Churchill XI: Agility cites conditional $300M orders
A customer’s three-year Digit 5 order commitment covers 1,000 robots and remains subject to contractual milestones.
Sentiment and the balance of points
Rhea-AI Sentiment reads the wording of the document, how positive or negative its language is on a 1 to 5 scale. The balance of points shown with the takes weighs what the document actually discloses, so the two can disagree, for example when a trial that missed its main goal is described in upbeat language.
Churchill Capital Corp XI (CCXI) and proposed combination partner Agility Robotics discussed their business combination at Agility’s investor day. Agility cited deployment commitments across nine customer facilities and more than 65,000 hours of Digit 4 operational time. It said a customer committed to $300 million in Digit 5 orders for 1,000 robots under a three-year robot-as-a-service agreement, subject to contractual milestones. Agility expects early-adopter availability in the first half of 2027 and general availability by the end of 2027.
Agility estimated a near-term U.S. humanoid market of approximately $1 trillion by 2032 across manufacturing, distribution and logistics. It cited over 400,000 unfilled U.S. manufacturing positions in 2025, projected to reach nearly two million by 2033. The presentation described Digit 5’s battery as providing about 90 minutes of operation and charging in under nine minutes. Agility projected a bill of materials of approximately $30,000 per unit at annual production of 10,000 units. The Form S-4 registration statement had not yet been declared effective.
How this balance works
Rhea-AI gives every point it takes from this document a weight. Minor counts 1, Moderate 3 and Major 9, so one Major point outweighs several Minor ones. The bar adds up the weights on each side, and when neither side holds more than 65% of the total the balance reads Mixed.
It reads the document as published, with the same rules for every company, and it does not look at what the market expected or at how the stock traded, so a point can be objectively good on a day the stock falls.
Rhea-AI Sentiment measures something else, the tone of the wording.
Hollow bars mark forward-looking points. How the balance works
Positive
- Major point. Forward-looking: it has not happened yet and may not happen.Milestone-contingent $300 million orders cover 1,000 Digit 5 robots over three years. 49% of market cap
Negative
- None.
Filing Explained
Digit 5’s operation near people is a future, not completed, capability: Agility says a third-party lab will certify it for use outside work cells, while its off-board detection-and-stop system with FORT is still being developed for testing with early-access units before general availability.
Key Figures
Key Terms
physical AI technical
Robot as a Service financial
bill of materials financial
cooperative safety technical
fleet management platform technical
FAQ
AI-generated questions and answers. How Rhea-AI works. Not financial advice.
What Digit 5 orders did Agility report in the CCXI transaction?
What bill of materials cost did Agility project for Digit 5?
AI-generated analysis. How Rhea-AI works. Not financial advice.
Filed by Churchill Capital Corp XI pursuant to Rule 425
under the Securities Act of 1933, as amended,
and deemed filed pursuant to Rule 14a-12
under the Securities Exchange Act of 1934, as amended
Subject Company: Churchill Capital Corp XI (File No. 001-43020)
Set forth below is a transcript from a presentation by Agility Robotics, Inc. (“Agility”) in which the proposed business transaction between Churchill Capital Corp XI (“Churchill”) and Agility is discussed.
Agility Robotics (Analyst & Investor Day)
October 6, 2026
Corporate Speakers:
| ● | Anthony Rozmus; Agility Robotics; Head of Investor Relations |
| ● | Peggy Johnson; Agility Robotics; Chief Executive Officer |
| ● | Jonathan Hurst; Agility Robotics; Co-Founder, Chief Robot Officer |
| ● | Daniel Diez; Agility Robotics; Chief Business Officer |
| ● | Jennifer Hunter; Agility Robotics; Chief Operating Officer |
| ● | Michael Beer; Agility Robotics; Chief Financial Officer |
| ● | Pras Velagapudi; Agility Robotics; Chief Technology Officer |
Participants:
| ● | Jesse Sobelson; BTIG; Analyst |
| ● | Andres Sheppard; Cantor Fitzgerald; Analyst |
| ● | Andrew Kaplowitz; Citigroup; Analyst |
| ● | George Gianarikas; Canaccord Genuity; Analyst |
| ● | Andrew Boone; Citizens; Analyst |
| ● | David Kerr; Schaeffler; President, Humanoid Robotics |
| ● | Courtney Baines; Schaeffler; Advanced Technology Production Engineer |
| ● | Nelson Hsieh; Foxconn; Director, Central Strategic Investment Department |
| ● | Mark Delaney; Goldman Sachs; Analyst |
| ● | Ken Newman; KeyBanc Capital Markets; Analyst |
| ● | Chris Moehle; Robotics Hub; Analyst |
| ● | Craig Irwin; Roth Capital Partners; Analyst |
| ● | Winnie Dong; Deutsche Bank; Analyst |
| ● | Joseph Spak; UBS; Analyst |
PRESENTATION
Anthony Rozmus^ Great, thank you everyone for joining us today. My name is Anthony Rozmus, I’m the head of investor relations at Agility.
We’re excited to have you join at our Analyst and Investor Day. We appreciate you taking the time to be with us. Before we begin, I’d like to review our forward-looking statements.
The webcast of this event will be available for replay at Agilityrobotics.com/investors, where you’ll also be able to find a copy of today’s investor presentation. Please note that today’s webcast may contain forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995 and are based on management’s current expectations. They may include, without limitation, predictions, expectations, targets, or estimates, including regarding our anticipated financial performance, business plans, and objectives, future events and developments, and actual results that could differ materially from those mentioned.
These forward-looking statements also involve substantial risks and uncertainties, some of which may be outside of our control, and that could cause actual results to differ materially from those expressed or implied by such statements. These risks and uncertainties, among others, are discussed in Churchill Capital Corp’s filings with the SEC. We encourage you to review these filings for a discussion of these risks, including the publicly available registration statement on Form S-4, which has not yet been declared effective.
You should not place undue reliance on these forward-looking statements. These forward-looking statements are based upon information available to Agility and Churchill Capital Corp XI today and reflect the current views and expectations of Agility and Churchill Capital Corp XI, and we undertake no obligation to update or revise the item for any new information except as required by law. Actual results could differ materially from these contemplated by these forward-looking statements, including but not limited to the timing of development milestones, potential future customers and revenue, and competitive industry outlook, and timing and completion of our business combination.
Please refer to the presentation accompanying this webcast as well as the filings and potential filings by Agility, Churchill, or the combined company resulting from the proposed transaction with the SEC, including underheading risk factors. Turning to our agenda, here’s how we built out the day. In a few minutes, Peggy will lay out the market opportunity in front of us and where Agility fits in it.
Jonathan will then take you deep into our newest robot, Digit 5, the tech, the safety architecture, and the physical AI stack underneath it. We’ll hear directly from Heather Lee, Acting President and CEO of AUVSI on policy and its impact on this industry, and then we’ll open it up for some -- for your first round of questions for Peggy and Jonathan. After that, Daniel will walk through our commercial strategy, market momentum, and customer engagement, followed by a fireside chat with customers David Kerr and Courtney Baines from Schaeffler.
Jen will cover our operations and manufacturing strategy, and we’ll hear from Nelson Hsieh from Foxconn. Finally, Michael will talk -- will walk us through the financial profile outlook in more detail. Then we’ll open the floor up one more time for an extended Q&A with the full team.
2
Please feel free to step out at any point during the presentation or take a brief coffee break as needed. We’re excited to have you with us today and to share more about the Agility story. With that, I’d like to turn it over to Peggy Johnson, CEO of Agility.
Peggy Johnson^ Thank you. Thanks, Anthony. Good morning, everyone.
Thank you for being here with us, whether you’re joining us here or online. Glad to have you. A lot of you have been covering the robotics space for a while now, and you’re asking exactly the kind of very sharp, very skeptical questions that this industry needs.
So, I want to start by saying we built today around answering those questions. We’re meeting at a genuinely pivotal time for robotics. Two years ago, a humanoid robot meant a stage demo, dancing or doing backflips.
Today, it means a machine clocking real hours on a production line for real customers doing real work that used to be impossible to staff. That kind of very revolutionary transition from spectacle to shift work doesn’t happen by accident. It’s the culmination of years of purposeful innovation designed to deliver enterprise value.
Agility is at the center of this revolutionary tech, and today we’ll show you exactly why. For those I haven’t met yet, I’ve met many of you, but my name is Peggy Johnson, the CEO here. I’m an electrical engineer by training, and if you look at my career, it’s really been one long thread of helping emerging tech find its first real home inside of the enterprise.
I spent 25 years at Qualcomm, and I had a front row seat to that industry, the commercialization of the mobile phone from the antennas, the cell sites, and eventually the semiconductors that really scaled that industry. Then in 2014, I received a call from Satya Nadella. He was just stepping into his role as CEO of Microsoft, and he asked me to lead business development there for him across all of their hardware and software products, helping to bring a lot of that early-stage tech to market at Microsoft.
In 2020, I stepped in as CEO of Magic Leap, where I took a company built for the consumer to focus on solving enterprise problems because that’s where that tech actually had a home. I tell you all of that because that’s why I’m standing here today after decades of driving large enterprises to adopt new tech. I’ve learned one thing above everything else, and that is companies don’t buy tech.
They buy solutions to their problems. Let me tell you a little bit about Agility. We’re headquartered in Salem, Oregon, with teams in Pittsburgh, and we just opened an office in Fremont, California, and our mission is simple, to build robots that augment the human workforce.
And Digit is the result of that. It’s the first multi-purpose humanoid that’s actually commercially deployed and working today. Our focus on getting our robots into the operations of paying customers and outside the lab distinguishes us from other humanoid companies.
3
And when I first met Jonathan Hurst, our co-founder and the Agility team, two and a half years ago, what attracted me wasn’t a robot doing a backflip. That’s not what I saw. But this team had built already and done the unglamorous work of figuring out how to bring capable hardware to market that was actually useful.
And what they found along the way was true product market fit. They’d built a solution to an industry-wide problem, which was solving the growing manual labor gap. What the company needed was an experienced team to scale the solution, and that’s the job that I signed up for.
To help me do that, I want to introduce a few of the folks on our leadership team today helping make this vision a reality. You’ll be hearing from many of them today. Jonathan Hurst is our co-founder and chief robot officer, and I would also say that Damion is here as well.
Damion Shelton, there he is, our other co-founder and chairman of our board. With Jonathan, I would say everything about Digit’s physical design traces back to his vision for building a truly useful machine. We have Michael Beer, our CFO, who many of you know from Citibank or his time at Energy Vault, having led similar transactions to the one Agility is going through today.
He brings us an invaluable depth of experience. We have Daniel Diez, our chief business officer who leads our commercial engine. He too has a very successful track record of bringing very transformational tech to market, and he was with me at Magic Leap.
We have Jen Hunter, our COO. She brings a decade of experience from inside Amazon’s robotics organization and now runs the operations at our business. We have Pras Velagapudi, our CTO.
He has both a decade in academia and a decade in the robotics industry and he brings us autonomous capabilities and to our physical AI stack. Ana Lang, our chief legal and people officer, has a long career of guiding tech companies through very complex legal and organizational challenges, and so she was the perfect person to lead those functions at Agility.
And finally, not with us today, Marco Marroquin. He’s our chief hardware officer. He’s led hardware teams through their frontier of autonomy across GoPro, Lyft, and Tonal.
So together, this team combines 50 years of robotics experience and 80 years of scaling commercial enterprise tech. And lastly, I’d like to highlight the tremendously talented directors that we just announced. They’re joining our public company board.
Merline Saintil, Derek Aberle, and Pierre Gentin. These directors will serve alongside our co-founder Damion and myself, in addition to representatives from Churchill. Merline, Derek, and Pierre have led some of the world’s most consequential businesses and institutions, and I would say their collective experience together with Damion’s deep roots at Agility really will be instrumental for us as we work to make humanoid robotics an integral part of the global workforce.
4
And that is going to happen. Humanoids are going to transform labor. The pace of innovation in robotics is moving at an unprecedented rate.
Tech that previously only had a home in research labs is now at the center of unlocking real commercial opportunity, and we believe the reason humanoid robotics is having its moment right now in 2026 and not five years ago or five years from now is due to the evolution of AI. That real-world perception, the reasoning, the planning, it’s gotten good enough and fast enough to give every robotics company’s real intelligence almost overnight. This happened very quickly, and that’s a rising tide lifting every serious robotics company in this category.
But to be honest, a robot knowing what to do is not the same as a robot knowing -- being able to do it reliably and to do it thousands of times a day on a factory floor. That is called physical AI, and that’s a far more complex problem of turning intention into repeatable, safe, physical actions by the robot. That data, by the way, doesn’t exist anywhere yet.
There’s no vast stored internet of robot movements like there was for training LLM, so this kind of data has to be earned one use case at a time, and Agility has spent years gathering it through real-world deployments. This tech, by the way, isn’t arriving into a market that’s indifferent to it. It’s arriving into a labor market that is genuinely running out of people.
This is existential for many companies. In 2025, there were over 400,000 unfilled manufacturing positions in the U.S., and on the current trends, that number is projected to grow more than four and a half times to nearly two million by 2033, and it’s not some cyclical dip that corrects itself. This is a structural shift in the labor market.
We’ve got an aging workforce that’s retiring out of these roles faster than young people are choosing to enter them, and at the same time, the U.S. is working to reshore all sorts of manufacturing capacity back here, so the opportunity here is to fill these roles that businesses are struggling to staff today, so the jobs that Digit takes on first are the dirty, dangerous, repetitive ones, the ones like repetitively moving material, manual tasks with high injury rates and high turnover rates.
That’s why they’re hard to fill, and as Digit steps in, the people who are doing those jobs are able to move into higher value roles. They can begin supervising the robot fleets and maintaining the tech and the facilities that the automation is in. This is additive to the workforce because the alternative is a continuously understaffed one, so how big is this labor problem?
You would guess big, but it’s probably bigger than you think. The TAM estimates range broadly, but we believe that near-term total addressable market for humanoids just across manufacturing, distribution, and logistics in the U.S. will be approximately $1 trillion by 2032. This is a big, big market.
We sized our near-term addressable market around the three industries where the labor gap is most acute, the ones I mentioned, manufacturing, logistics, distribution. These are markets that share the same characteristics. They all have large labor pools, very structured and repeatable workflows, and also existing infrastructure built around the hallways, the walkways where their workforce currently operates, and I would say that last point is very, very important.
5
Digit doesn’t require a customer to move a bunch of things around to redesign their facility to deploy. Digit can walk in and fit into the aisles and the doorways and the workstations that already exist. Now, as Digit’s capabilities expand with each new generation, we’re getting more dexterous, we’re adding more skills.
That addressable footprint will expand with it. From today, bins and tote handling towards machine tending, kidding, quality inspection, and then we can move into new verticals over time, like retail, healthcare, and eventually, down the road, into the home. That’s a ways off, we believe.
But we’re sizing this business around the demands of a growing labor gap, and we’re building the roadmap to meet those demands. Now, you will likely observe there is no shortage of humanoid robots dancing on stages. There is, however, a shortage of humanoid robots on the payroll.
Our currently available robot, Digit 4, is the humanoid with the largest, the longest, I should say, continuous commercial track record in the industry. We’ve got over two years of live deployments, clocking more than 65,000 hours of real operational time, and we work at customer sites like GXO, Schaeffler, and Toyota Motor Manufacturing in Canada. That track record compounds in two ways.
First, commercially. It’s why we’ve been able to secure $300 million in multi-year orders for our Digit 5, subject to the satisfaction, obviously, of certain contractual milestones, product features, and specifications. We do anticipate Digit 5 will be available for our early adopters in the first half of 2027, and generally available by the end of 2027.
Second, technically, every one of those 65,000 hours of Digit 4 work generates operational data that no competitor that has just months of deployment has access to. They just don’t yet. And that data is becoming a flywheel for us.
It improves our physical AI, and that unlocks new skills, and that unlocks new workflows and new customers. That generates more deployments and more data. So, it all works for us here.
We believe this very real-world flywheel is very difficult to shortcut, and it is the core of Agility’s advantage. We’ve also made a very deliberate choice to govern in-house the parts of this business that can be the hardest to control. We design and assemble Digit ourselves at RoboFab in Salem.
It’s our purpose-owned, purpose-built facility. We have a capacity to scale to 10,000 units a year. Our approach to the tech, to customer deployments, and to the supporting infrastructure is why some of the most important strategic players in the ecosystem have backed us along the way are partners such as NVIDIA, Amazon, SoftBank, Scheffler, ABICO Group, and Foxconn, who is anchoring our pipe.
6
It is worth being precise about what we’re actually building, because Agility is not just building a single product. At the center is Digit, our general-purpose humanoid, and today’s Digit 4 is the version that’s been earning its reputation over the past couple of years. In the first part of 2027, when we make our next-gen robot, Digit 5, available, it will be available for early access customers.
We believe that commercial availability of Digit 5 will be a true product inflection point and not just a product iteration. I’ll go into why in just a bit, but around the robot sits something called Agility Arc, which is our cloud fleet management platform. It’s what lets a customer run multiple Digits as a coordinated fleet.
It connects to existing warehouse systems, to tracking metrics like uptime and throughput. It talks to other robots, and it gives us this operational visibility that we need across every deployment. We also provide the services layer for our customers who need to run this in live production, so we’ve got deployment, maintenance, ongoing support, whether a customer adopts Digit through our robots as a service or purchase it outright.
We see a real revenue opportunity beyond just Digit sales here through licensing of our proprietary hardware tech into the broader robotics industry as well. We’ve spent a decade actually building that tech, and it’s valuable beyond our own robot. And we plan to capture that value through a new high margin revenue stream over time.
And this is why we believe right now is the right time to go public, because it’s not just because the market is demanding humanoid solutions, but because our tech and our ops are ready to support that demand, and we’re bringing a large and growing order book with us. So, let’s talk about Digit 5. Every humanoid in this industry, including our own Digit 4, has been boxed inside of a fenced work cell.
Today’s safety standards were actually written for machines that don’t dynamically balance like humanoids do and move about on their own. So, this is a new category of risk that existing frameworks weren’t really built to underwrite. Digit 5 is our first humanoid designed to work near people.
So, at general availability, Digit 5 will combine AI-based human detection, safety cues, and an independent safety controller that can stop or power the robot down before any human contact at all. Digit 5 will also deliver on the hardware upgrades customers have been asking us for, more payload, an ultra-fast charging battery, and swappable hands to equip Digit with the flexibility to take on more kinds of work. And here’s why that matters.
Once a robot can be trusted to move through a facility full of people, every new skill it learns adds value across the whole operation, not just one fenced-off task. And that’s what turns Digit 5 from a robot that does one thing well into a platform that can absorb a customer’s entire workflow. So, it’s why Digit 5 will be an inflection point for fulfilling labor gaps.
That’s the opportunity, and it’s why we believe Digit 5 will capture it. None of this works, though, without the technology underneath actually delivering real value. And there is no one better to take you through that than the person who spent his entire career on exactly this problem.
7
Please join me in welcoming Agility’s co-founder and chief robot officer, Jonathan Hurst.
Jonathan Hurst^ [Very good]. All right. Thank you, Peggy.
Good morning, everyone. So. Peggy just walked you through why this market exists and why we think Agility is really built to win it. My job for the next few minutes is to go a little bit deeper into the robot itself and the AI underneath it, the safety architecture that we believe is the gating factor for the entire industry.
My career-long mission has been to build machines that can move and physically interact with the world the way that humans and animals do. And this is very different from the rigid position control of historic robots and automation. For a long time, this was a basic science problem, and it lived in a university lab.
So, through my graduate work at Carnegie Mellon and later as a professor at Oregon State, we made real progress on the core science of dynamic balance and physical interaction. But at some point, the blockers stopped being scientific, and they started being about finding applications and creating a product and building an engineering organization, all things that universities aren’t built to do. And that’s when we founded Agility.
So, we brought along four of our robots. You’ve all seen them over there. We brought Cassie and three generations of Digit.
And each one of these kind of marks a distinct stage of our company maturity and how we got here. And this first stage was really an R&D era built around Cassie, our first robot. So, this is where we solved some of the core engineering problems, like our first cycloidal actuator design.
That technology gave us the force control to have this compliant interaction with the world and exceptional durability, even with repeated ground impacts during walking and running. And really nothing else on the market could match that. Whether it’s something we could brought by.
We had to develop it. Now, we sold Cassie to research programs at Berkeley, Michigan, Caltech, Georgia Tech, a bunch of others. It was really the only bipedal machine on the market that was capable of this kind of biomechanically relevant movement.
Also, this is when we began our investigations into what everybody now calls physical AI, really before it was popular. And you can see here that Cassie is moving like a thing alive, even in these early experiments. This was all reinforcement learning, sim-to-real transfer.
So, the next stage of the company is, call it the proof-of-concept era. And this is Digit 3. So, this is the platform that we used to test literally hundreds of potential use cases and find where the technology that we understood really creates value for a customer, where’s the right match here.
8
And that’s how we landed on bin handling, on tote manipulation as our beachhead market, just the very first use case of many. And this problem is structured enough to automate it. It’s not a totally general environment.
But every site is different enough, every workflow is different enough that single purpose automation hadn’t solved it and hasn’t been able to solve it. So, they had people trying to do this job, and as Peggy pointed out, hard to hire people for these sorts of roles. That work led directly to our third stage with Digit 4.
And this is commercial deployment, the first commercial deployment of [a humanoid robot]. And this was with GXO, the first time a humanoid robot has worked in a customer’s workflow and actually been paid for that work. This is the same platform that’s still out there at companies like GXO and Schaeffler and Toyota.
And it’s racking up all those operational hours that you heard about this morning. Now, an important point I want to make about why are we building a humanoid robot. We did not set out to build a machine that looks like a person.
And in fact, we actually set out to avoid building machines that look like people because I don’t want to fall in this biomimetics trap of copying all the wrong things, right? And the early machines, of course, don’t look anything like a person. We build robots that can do many things in human spaces.
We’re building multipurpose human-centric robots. And in designing the hardware, we start from the first principles of the physics of movement and the engineering. We’re building things, remember, out of motors and metal.
We’re not using muscle and bone. So, the form may look very different from a person, and we always let the design follow the requirements. Now, it does end up looking like a humanoid because the requirements have pointed it that way.
And it’s been a decade of figuring out exactly the function that gets us there. So, to work in aisles, like narrow spaces, go through doorways, you can’t be big and wide and squat to be narrow and lift heavy things up high. So, you can’t be statically stable because you wouldn’t fit in the space.
So that means you have to be actively balancing if you want to have a very small footprint and still be able to lift things up high, the way that we do on two legs. And you can think -- we could probably do that on wheels, and we can. That is possible.
But if you’ve ever been pushed while standing on a Segway versus standing on your own two feet, which one recovers better. And that intuition holds up under the physics as well. So, legs really are the right solution for dynamic balance.
Having an upright torso lets it lean and balance in the direction that it’s moving. And that gives us room to mount the arms and the sensors, put in the batteries and the compute and everything else in a way that’s actually helpful for the balance of the robot. And in a place that lets the robot see in human spaces and reach up high and still be able to turn in place.
9
Having two arms rather than one arm lets it lift bigger objects. So, a relatively small robot can lift very large things and move them through a huge workspace. It also lets you reposition things from one hand to the next.
So, it allows you to have much simpler end effectors while still being able to do all the dexterity that you want to do. So, the core advantage for this form is that ability to move those very large objects. And once you have the two arms, the two legs, an upright torso, people expect a face because it’s already starting to look a little bit like a person.
And they need somewhere to read the cues of what the robot’s about to do. And you have a real opportunity here with a robot like this to communicate on human terms. The same cues that you might read on a coworker in a way that doesn’t have to be trained.
You just understand it. So, we’ve built just enough expression into the robot to meet that expectation without chasing realism into the uncanny valley and trying to make something look too much like a person. So that brings us to our newest robot, Digit 5.
And this is the fourth stage that we’re in right now as a company, which is our scaling moment. This is when we grow. The single biggest blocker standing between where we are today and having thousands of robots deployed across hundreds of facilities or even homes is not intelligence and it is not dexterity.
It’s not the ability to walk upstairs or fold a fitted sheet. It’s about safety. Digit 5 is designed to operate in close proximity to people.
But this has been a major engineering effort because Digit is a new class of machine. As a humanoid robot, it’s dynamically stable. So, unlike a stationary robot arm, it can fall.
It can land on somebody’s foot. Simple risks like that. And it’s a category of risk that the existing regulatory and insurance framework really haven’t anticipated.
So, until that’s addressed, every humanoid, regardless of how capable it is, has to operate inside of a fenced work cell requiring that installed infrastructure and disconnecting it from the rest of the facility and the people in it. So, creating this cooperatively safe humanoid robot, it’s not the sort of thing where you can just put a safety hat on an existing robot. It is a completely holistic, top-to-bottom design exercise.
It touches nearly every system of the machine. It influences the shape of the limbs and the body. It requires dedicated computing and sensor arrays and provides sort of hard requirements for the entire electrical system to ensure that they do exactly what you expect of them 99.999-something percent of the time, provably. We’ve spent years inventing the strategies and building the engineering best practices and working with partners like NVIDIA to bring a solution for this forward. Digit 4 was the first humanoid to pass a field evaluation from an OSHA-recognized independent testing lab on an active customer production line. Digit 5 builds on that work.
10
And when it’s generally available, it’ll be the first humanoid robot that is capable of safely operating in close proximity to people. Digit 5 will combine AI-based human detection that continuously distinguishes a person from, say, a statue of a person or a photograph of a person, and it tracks how somebody is approaching an environment. It can provide safety cues so that people around the robot can intuitively understand what it’s about to do.
And it’s built with that certifiably safe motion control so the robot can slow down, stop, set its payload down, and even sit down on the ground and power off its motors in a statically stable position before the person can physically make contact with it. So with any sort of AI-based technology that this is, you don’t let the guardrails down right away. It’s not all or nothing.
It’s not that the robot is safe or it isn’t. It is a gradual improvement of safety features with deployed proof points. If you think about the autonomous vehicle industry, AV companies, they didn’t go straight from a research machine to a driverless one.
They used safety drivers for years, collecting all the data that they needed to prove out the capabilities to earn that full autonomy safely over increasingly complex environments one validated step at a time. And that’s how we think about cooperative safety. But this approach that we have now is our first step towards that side-by-side with humans.
This is the mechanism by which we collect the data to responsibly expand what Digit is trusted to do next. This is basically the safety driver right here, the first step out of the work cell on the path towards really trusted safety in human environments and around people. And this is the unlock that really matters most to customers.
Once a robot can be trusted to move through a facility full of people, every new skill it learns adds value across the entire organization. And that’s what changes Digit from being a robot that does just one or two things well in a normal space to a platform that really works across those full facility workflows. We’re also not building every layer of the safety stack alone.
Agility is the first company to build NVIDIA’s new Halos system for robotics into a safety system, integrating NVIDIA’s IGX Thor compute and Halos core software into Digit’s human detection system. We’re using NVIDIA’s own (Technical Difficulty) for third-party safety certification. That gets us a validated industrial-grade safety foundation underneath our own system.
Now, along with cooperative safety, our early deployments gave us more insight into customer requirements to scale this technology. These are requirements that apply not just to Digit, but they would apply to any humanoid if it’s going to be successfully deployed in industrial settings. The robot has to have a battery that allows it to work across a factory’s three shifts a day and do that reliably. On Digit 4, that was kind of gated by a two-to-one run-to-charge ratio. In other words, for every two hours of work, the robot needed about an hour on the charger. And that is very normal for batteries.
11
Across two to three shifts a day, that is a lot of downtime and a lot of coordination and overhead for a facility trying to run continuously. So, we looked hard at solutions for this, including things like swappable batteries as the fix. And we passed on that, the same way cell phones and electric vehicles did, because swapping batteries means extra mechanical complexity, a human or another machine in the loop, spare packs for inventory and to maintain as a hazardous material.
Digit 5 solves this instead with a new fast-charge battery system. So, we have about a 90-minute run time on the battery pack that charges in under nine minutes. That’s really important.
That gives us a ten-to-one work-to-charge ratio, which works out to more than 20 hours of productive work in a 24-hour day. And this is a good example of something that is perhaps an unglamorous upgrade, but is really critical for moving these deployments to scale. Digit has to lift heavy loads over and over again, up to the amount that’s recommended in these OSHA-regulated facilities.
And our cycloidal transmissions that we started a decade ago on Cassie match the strength needed, but they also achieve this 95% to 96% efficiency in the drive, which is really unusual. And that gives us force control that nothing else on the market matches for something like this, for this kind of use case. And 10 years in now, they’re essentially indestructible.
We have observed no meaningful wear, even in this punishing application. Another important thing is that Digit has to be cost-effective to the point where it is economically compelling for customers, compared to the fully burdened labor costs. That’s really what we’re comparing against.
We’re on a path that ensures the necessary performance, of course, while being on the path to low-cost manufacturing with common processes like machining and casting aluminum, thermoforming and injection-molding plastics, and shaping and bonding sheet metal. There are no exotic manufacturing technology pieces needed for any part of Digit.
Finally, I would say that it needs to offer the flexibility of general-purpose automation. Our function-first humanoid form helps with this because it can do so many of the workflows that were designed for people to do.
But we also have the advantage of changing hands for different tasks, putting tools directly on the robot. Our tool change system makes it easy to use our in-house manipulators that can lift these 50-pound bins, which is something that nothing off the shelf can do, while still swapping to different tooling for other tasks like machine tending or kitting or sequencing. Of course, the hardware is the foundation, but AI really unlocks the generality of this hardware.
So new AI approaches to robot behavior and control means it can learn new skills with less engineering development time and be deployed across the facility to perform this work where it’s needed most. So talking about AI, when most people talk about AI, they are usually thinking about semantic AI up here. That is the perception, the reasoning, the same models that are behind Google’s Gemini or Anthropic’s Claude, and it lets Digit see what’s in an environment and understand it.
12
And this layer of semantic AI has gotten very capable, very fast because it’s trained on the entire internet. It is a rising tide. This is something I think of like the internet or electricity is a commodity at this point.
It’s becoming that way. It lifts every serious robotics company, including us. But there’s no internet-scale data set for how to control the motors of a bipedal robot.
That data doesn’t exist anywhere, and it has to be created, and that’s the physical AI. And that layer turns the intention into safe and repeatable physical action. It is the harder problem.
It is far more specific to our hardware, and it’s the piece of the puzzle that Agility creates and owns in-house. So, we build that layer in a couple of steps. I guess I would say the foundation, of course, is being able-bodied and having the right hardware so the capability is even there in the first place.
Then we can be teaching skills, learning from demonstration means teleoperation of the robot or motion capture or computer animation, many sources of data for that, that gives the robot a hint, a strong starting point for a skill, how to grasp this or how to walk over there and how to do it at 95% reliability, something like that. But then reinforcement learning in simulation lets the robot practice and refine that behavior on its own because Digit’s body doesn’t have the same physics as a human body exactly. It has to find its own version of the movement.
This is no different from a person having to practice to learn how to do something well. Over time, this extends into reinforcement learning in the real world across fleets of robots, all coordinated and fed back through our fleet platform Arc. That is the flywheel that Peggy talked about.
Every deployed hour generates data that allows us to improve the physical AI and get better and better at all of our tasks. The quality of the data is really important. That’s the thing that we think really matters.
The robot must learn from relevant experience which can only be gained on the job. So, that flywheel compounds, it’s very hard to shortcut that. And because we are focusing on the quality of data for these relevant use cases, it also means we don’t need to make the enormous compute bets that some of our competitors are making with no guarantee that it gets them into a working robot any faster. We think that that makes us both more data efficient and more capital efficient.
So, we have been discussing individual robots, but to deploy thousands, they have to coordinate together. Arc is our cloud fleet management platform, the layer that you don’t see, but it does a lot of the work. This coordinates robots across a facility, integrates with other automation already running there, and gives us real-time health and diagnostics on every unit. It’s also core to the data collection. So, every one of those signals feeds the same physical AI flywheel that I just described.
13
Over time, we see Arc as a meaningful product in its own right, and not just plumbing underneath one robot. So, this brings me back to another point about safety. I’ve walked you through safety in terms of hardware upgrades and then the physical AI, and then Arc on purpose, in that order, because safety is not a separate topic from any of it.
You really cannot just add safety onto humanoid at the end, or layer it on top of a finished machine. It has to be built into the hardware, built into the AI, built into the fleet software, all the way through. So, one more thing before I hand this off.
Digit 5 is a platform that is designed to be upgraded. We can retire subsystems for better ones. We can be improving actuators, batteries, manipulators, and sensors, and swapping those in as pieces and parts throughout the years.
That continuous upgrade path is how Digit stays useful for years in the field, and how we get to whatever comes next. If you zoom out far enough, the destination isn’t a warehouse. It’s robots that are genuinely part of everyday life.
They’re helping in homes, they’re working in stores, they’re moving a package the last few feet from an autonomous delivery vehicle to your front door. That big picture is Agility’s vision. And we believe that Digit 5 is just the next real step towards that big vision.
So, this is really an exciting time to be in the field of robotics. Our continued technology breakthroughs and our commercial success depends on policy, keeping pace with what we’re building, and on the people writing our laws being part of this conversation from the start. So, I want to hand things over to a different kind of discussion.
Join me in welcoming Heather Lee, Acting President and CEO of the Association for Uncrewed Vehicle Systems International, or AUVSI, who will join Peggy for a conversation on the policy landscape shaping this industry. Welcome.
Peggy Johnson^ Thank you. Thanks, Jonathan. So we are going to have a fireside chat with Heather Lee.
Heather, it’s a pleasure to welcome you, and thank you very, very much for joining us today.
Heather Lee^ Thank you for having us.
Peggy Johnson^ Yes. So AUVSI, for those of you who don’t know, is the world’s largest trade association representing what’s called the uncrewed systems, robotics and other autonomous technologies industry. And they’ve really become the voice in Washington on the policies that will shape the future of robotics and autonomy. More recently, the organization has broadened that leadership across advanced robotics and physical AI, bringing together industry and policymakers around issues ranging from the national competitiveness and supply chain security to federal procurement, workforce development, and the need for a coordinated national robotics strategy.
14
And Heather is the Acting President and CEO. At AUVSI, she’s held a number of senior strategy and operational leadership roles. And today, she oversees the engagement, the industry development, the strategic programs, the meetings, finance, and administration.
So, lots to do there. And she brings -- I think you would say, Heather, you bring a particularly interesting perspective to this conversation as prior to AUVSI, she spent nearly two decades at CTIA. Some of you may know that as the trade association for the U.S. wireless industry. Heather’s been at the center of much of AUVSI’s ongoing work, and I am delighted to have her with us today to talk about that work and where robotics policy is headed and what it means for our industry.
So, we are excited to have you here today, Heather. With AUVSI being at the forefront of the robotics policy discussion and from Agility’s perspective, the conversation has really evolved over the last year very, very quickly. Can you give us a sense of what has changed and walk us through some of the key robotics initiatives that Congress and the administration actually are working on right now?
Heather Lee^ Absolutely. But before we dive into the specifics of that, I’d love to take a moment and congratulate you, Peggy, and the whole Agility team on today’s event. You’re heading toward a major milestone for the American robotics industry.
I think we can all agree that being the first is never easy, but the transparency that Agility is embracing as the only U.S. publicly listed pure play humanoid robotics company is going to send a strong message to industry that American innovation and leadership are thriving in this space. So, a huge congratulations to you and the team.
Peggy Johnson^ Thank you. Thanks, Heather.
Heather Lee^ Of course. And also, I want to take a quick moment to recognize Mark Aitken from your team who recently became the chair of our robotics subcommittee, and we’re looking forward to working closely with him on that.
But to your question, yes, a lot has happened in a year. I think one of the biggest signs of that increased activity is that Michael Robbins, our former president and CEO, testified three times in the House on robotics, and all of this happened before August of this year. One of the testimonies were on -- was on the GUARD Act, which Agility endorsed, and as always, we appreciate your partnership on that.
The big theme that we’re seeing is that Congress and the administration are focused on protecting the U.S. robotics industry from adversaries, especially the CCP, the Chinese Communist Party. And part of the reason why I believe the U.S. government is so locked in on this is that AUVSI spent years showing what happened with drones, where China captured the U.S. market. They’ve seen that playbook once, and they don’t want it to happen again in robotics. They see robotics as essential to both national security and economic competitiveness.
15
So, let me start with Congress and what’s been going on there. With Congress, the biggest vehicle there is the NDAA, the annual defense bill, and both the House and Senate versions have robotics language in them. The Senate version would bar federal agencies from buying or operating adversary-made ground robots, like humanoids and quadrupeds, and that language comes from two bipartisan bills, the American Security Robotics Act and the Humanoid ROBOT Act.
The House version tees up a federal security review of the robotics sector that could lead to procurement restrictions, and it has a provision tied to the GUARD Act, which I’ll come back to a little later. On the administration side, there’s also been just as much happening, and the biggest move came from the FCC. In August, it blocked new foreign-made advanced robots from being imported and sold here, and that applies to every country.
Companies can apply for a conditional approval, and the FCC is using that to push manufacturing onshore, but officials have called this an interim step, so I would expect to see more on that front. Commerce is also running a Section 232 National Security Investigation into robotics and industrial machinery. While the deadlines for the decision have passed on that without an announcement, I wouldn’t necessarily count that out, and I wouldn’t read that as the issue going away.
The definition is broad enough that humanoids could be swept in, so we’re still watching this closely. But with all of this, we’re keeping a close eye on balancing between protecting ourselves against adversaries and acknowledging that we can’t succeed on our own. We need our allies to help us scale.
Many of our members rely on allied supply chains, and our policy needs to reflect that, and that’s what we’re working toward.
Peggy Johnson^ Yes, no, you make great points there. Speaking of China, one of the forces clearly shaping this discussion is the geopolitical and economic competition between the U.S. and China, and we’re increasingly seeing robotics discussed not simply as an emerging technology, but in the context of supply chain security, domestic manufacturing, economic competitiveness, and, again, national security. How significant has the U.S.-China dynamic become in shaping robotics policy, and do you expect it to remain a major catalyst going forward?
Heather Lee^ Yes, that’s a great question. It is definitely the defining issue in robotics policy right now, and we expect it to stay that way. The CCP sees advanced robotics as a way to unlock both military capability and economic value, and it’s investing to get ahead of us.
The U.S. still leads in innovation, but we can’t take that for granted. We’ve got to lead and stay ahead on quality, and we have to scale. Scale is where we’ve seen this movie before.
China is running the same playbook in robotics that it ran with drones. It subsidizes both sides of the market. State money goes to robotics companies, and government procurement creates the demand.
16
So that, in combination, lets Chinese robots be exported at prices that no free-market competitor can match. With drones, these tactics got China roughly 80% of the U.S. commercial market. So, a similar dependence in robotics would be an even bigger security problem.
I mean, these systems are mobile, they’re sensor-rich, and they’ll be working inside American factories, hospitals, homes, and military facilities. So that’s why it matters so much that Congress and the administration act now and that we get it right while we still have this window.
Peggy Johnson^ Yes. Thank you. Notably, AUVSI was instrumental in encouraging policymakers to use federal procurement itself to help build and scale the domestic drone industry, and especially with Department of War’s Drone Dominance Program, was the name of that program.
Just last week, I was selected to be part of the Department of War’s Project Meridian. You might have seen that announcement. That basically stood up to support the development of a study to address the future of warfare, including adjustments or transformations in the Department of War’s approach to warfare to maximize U.S. warfighting advantage over time.
Do you see any similar procurement opportunities emerging for advanced robotics? And specifically, do you think we’ll see the federal government and perhaps the national security community in particular begin using procurement more deliberately to accelerate the adoption and scale of American-made robotics?
Heather Lee^ Yes. First, congratulations on Project Meridian.
Peggy Johnson^ Thank you.
Heather Lee^ I think that’s a huge testament and a real vote of confidence in you and the work that Agility is doing.
So that’s great. But yes, we think we will see similar procurement opportunities and what’s happened on the drone side shows us how. You’re absolutely right in that AUVSI has been working not toward just defensive policy, but also offensive policy.
We -- the restrictions on Chinese drones only went so far. And as you stated, what really changed the market was when DOW decided to buy American at scale through drone dominance and robotics needs the same thing. Big part of what we’ve been pushing for is playing that offense and not just defense.
The procurement bans and the NDAA and the FCC’s import restrictions are all defense. And they’re an important first step because they send a market signal, but that doesn’t build the industry by itself. What builds it is the government deliberately investing, both in finished systems like Digit, but and also in the supply chain underneath them, including components and critical minerals.
17
So, a few examples would be federal contracting that intentionally buys American-made robotics. Another would be export-import bank financing where we’ve endorsed legislation that specifically includes robotics and critical minerals. And critical mineral support, like the Magnets Value Chain Support Act, which we’ve also endorsed.
And doing this isn’t just good for the industry. It’s good for national security. The Department of War gives real operational value from logistics support to taking on dangerous tasks so soldiers don’t have to.
So, it’s very important to get the offensive side right as well.
Peggy Johnson^ I agree. Finally, we do have a midterm election approaching and there’s naturally a lot of discussion about what the political landscape in D.C. will look like next year, including the possibility of a divided government. From where you sit, do you see, how durable do you think is the bipartisan consensus around robotics, which we do have today?
And regardless of which party controls Congress, is there enough alignment around the economic and national security importance of robotics to keep this agenda moving forward?
Heather Lee^ Yes. So, AUVSI is nonpartisan, so we won’t predict the election. But we’re confident that this agenda holds up no matter who controls Congress. So, we believe that it is very durable.
And let’s just look at who’s sponsoring the bills now. We’ve got Cotton and Schumer on the American Security Robotics Act. We have Cassidy and Coons on the Humanoid ROBOT Act.
Moolenaar and McClellan on the GUARD Act. And on the National Commission on Robotics Act, Obernolte and McClellan in the House. And in the Senate, Hickenlooper and McCormick, who I know has been a great partner to Agility, due in large part to your presence in Pennsylvania.
This kind of bipartisan alignment doesn’t happen by accident. Both parties definitely see robotics as central to national security and economic competitiveness. And concern about China crosses the aisle.
Here are a few things that we’re going to be watching through this lame duck period and into the next Congress. The NDA, as mentioned before, which has a good chance of carrying procurement restrictions on adversary-made robots. The GUARD Act, which cleared its Energy and Commerce Subcommittee after Michael’s testimony.
And now we’re working with the Hill on how it fits with the FCC’s new rules, including making sure that allies are protected. The Communications and Technology Transparency Act from Chair Guthrie and Ranking Member Pallone, which would limit future FCC covered list actions to foreign adversaries and add congressional oversight. Drones and robotics have been at the center of the covered list, so we’re bringing our members’ experience to the conversation.
18
And the National Commission on Robotics. If that doesn’t get done this year, it’s going to be a top priority for AUVSI for the next Congress. The U.S. needs a national robotics strategy and a bipartisan commission is the right way to build one.
Peggy Johnson^ Well, Heather, thank you. This is clearly an important moment for the robotics industry, and I think your comments underscore both the opportunity in front of us and the importance of industry and government working together as the tech moves toward much more broader adoption. So, thank you for joining us today. Very much appreciated.
Heather Lee^ Thank you so much. It was a pleasure and honored to be with you. Congratulations again.
Peggy Johnson^ Thank you. Okay. Now we’d like to offer a brief Q&A.
Anthony, I’ll hand it over to you to walk us through a Q&A, and I think Jonathan’s going to join me on stage.
QUESTIONS AND ANSWERS
Anthony Rozmus^ Great. Thanks, Peggy. For this Q&A session, the first one here, let’s just focus on questions for Jonathan and Peggy.
We have a few handheld microphones that will be going around. If you could just state your name before you ask your question and the firm you’re with, that’d be great. One of our team members will find you, and then we’ll get started here.
So, I think Jesse, the first one right here.
Jesse Sobelson^ Hey, guys. Good to see you. Jesse Sobelson with U.S. Bancorp, BTIG. So, with the safety unlock here, what share of the pipeline workflows were off limits or behind barriers to open -- that opens with Digit 5? And I guess what I’m asking is, from Digit 4 to Digit 5, you’ve enabled collaborative work with people. What’s been unlocked in workflows for that machine?
How much of the work are we seeing is available robotics today from that?
Peggy Johnson^ Yes, I’ll start. You probably have some to add. One thing that when we are inside the fenced-off area, we’re doing one thing.
But typically, what a human worker does is a number of things throughout the day. So, they have a workflow that may take them from one part of the facility to another, to another, and they’re doing different things. So, what it unlocks is the entire start to finish of that human’s workflow.
Again, that human that they can’t find right now. And that is what allows us to really expand. And it has a multiplicative force on what Digit can do from being inside to coming outside the work cell.
19
Jonathan Hurst^ Yes, the other thing to keep in mind is when you have to build a physical barrier, you’re installing infrastructure. And if you are going to install infrastructure to do a thing, maybe you just use some piece of purpose-built automation. The real value of a humanoid robot is it can just walk into this kind of already existing as-built environment and start doing the task.
The customers that we started to work with are more about seeing the future and working with us towards that future. That’s why they started with us with Digit 4. It wasn’t so much that the building walls, installing this thing, and then having a humanoid robot in there doing it is the optimal solution at the time.
It’s growing into what Digit 5 is. Thank you.
Anthony Rozmus^ Good. Right here.
Andres Sheppard^ Good afternoon, everyone. Andres Sheppard from Cantor Fitzgerald. Good to see you again.
And congrats on all the great success on very soon to be the first pure play, publicly traded humanoids company in American history. So very exciting. Peggy, quick question for you is, what are you most excited as we reach this milestone?
And Jonathan, maybe for you, what are some KPIs that we analysts or investors should be tracking to monitor your progress? What are the things that we should be paying closest attention to? Thank you.
Peggy Johnson^ Well, this milestone is coincident with us coming out of the work cell. So that is really exciting. It is the start of the expansion beyond manufacturing logistics and distribution into retail, healthcare, adding skills.
We can add them very quickly now with the physical AI that Agility has incorporated. So, the world of opportunities really opens up. And my team hates it when I use this analogy, but it reminds me of the early days of the mobile phone industry when we didn’t know where mobile phones were going go.
And they just went, we couldn’t even think back then of where they would be at today. And that’s what it feels like when we can have a robot walking around a facility, doing a number of different things, very much replicating what a human would have done is opening up a lot of opportunity for us.
Jonathan Hurst^ Yes, I think a good thing to track is operational hours of robots. Because just getting to that bar of having these machines that you are allowed to deploy because they meet the safety requirements and they have the uptime and the performance that a customer is even willing to allow them on their floor is a pretty high bar to get over. So, when we start racking up tens and then hundreds of thousands of operational hours, that’s a pretty significant moat and a significant demonstration of organizational strength for how we’re able to support these things in the field engage with our customers and really provide value.
20
Anthony Rozmus^ I’ll go to Andy right next to Andres.
Andrew Kaplowitz^ Thank you. Andy Kaplowitz for Citigroup. Jon or Peggy, any gating factors that we still need to think about for that ramp-up that you guys talked about on Digit 5 first half and then more widespread in the second half.
And then I think you signed an MOU on Digit 5 safety infrastructure with FORT Robotics less than a week ago. So, it seems like you’re still focused on building out the safety sort of parameters, if you may. So maybe just update us on sort of what that MOU is and what it does for you.
Peggy Johnson^ Yes, why don’t you start with that? Because it’s all part of a broader safety strategy.
Jonathan Hurst^ Yes, just briefly, remember I mentioned that safety is an ongoing process, right? And so just getting out of the work cell is kind of the first step on the path towards really trusting these things to do a lot of things. FORT makes a number of components that we use in our safety strategy.
So, it’s just like -- it’s a piece of the supply chain like you would imagine any number of other components for the machine. Does that answer the question?
Andrew Kaplowitz^ Yes.
Jonathan Hurst^ Yes.
Peggy Johnson^ So any gating factors, I mean, obviously getting the certification on safety is something that is left to third party. Though we have a big voice in the -- in setting the security or the safety standards in the industry. We’re sitting on the regulatory bodies, as you heard Mark Aitken is, our head of safety, Kevin Reese is as well.
So, we are a voice forming those standards, but it is going to be a nationally recognized test lab, a third-party organization that will give us that certification. We did seek that for a Digit 4 inside the work cell and we will do the same for Digit 5 outside the work cell.
Jonathan Hurst^ I don’t think there’s any, there are no like technological bottlenecks or things that are where we’re like, I’m not sure, we’ll have to figure this out. There’s nothing like that. We have our arms around the problem because we’ve had our robots deployed with our customers.
So, we have very well defined exactly what the use cases are that we will be entering into and have written that into the entire engineering requirements for the whole formal product process to create Digit 5. The real challenge right now is execution as an organization, as a company, to successfully deploy these and service them and build out the skills and make them better. That’s it.
21
Anthony Rozmus^ Go ahead, George.
George Gianarikas^ Hi, everyone. George Gianarikas from Canaccord Genuity. I think it’s safe to assume that you had to put some physical limitations on the robot to achieve the safety status that you did.
So I’m just curious, what can Digit 5 do that’s better than a human?
Jonathan Hurst^ Yes. Well, Digit 5 is a lot stronger.
Peggy Johnson^ Yes.
Jonathan Hurst^ It can lift up. So when you say lift a 50-pound bin, it’s not lifting it just one time. It’s doing that as part of its task all day long. And it’s not a nice, solid, convenient 50 pounds. It’s maybe a 50-pound ball bearing that’s rolling around in the bin.
So, when we were testing out the manipulators and trying to figure out how we’re going to grasp this thing early on, we got a steel ball bearing, rolled it around in the bin and picked it up. And engineers were dropping it all the time. It’s actually really hard even for a person to kind of handle that and lift that up because that’s the specification.
So really meeting and getting to all of those edges of the human kind of OSHA-based recommendations, that’s what Digit 5 can do right now.
Peggy Johnson^ And I would add the superhuman capability. You could do that for three shifts.
Jonathan Hurst^ Yes.
Peggy Johnson^ You wouldn’t put a human on a three shift lifting 50 pounds over and over and over again.
But Digit, you can. So that’s the start of this superhuman capability we’re going to see with the robot as we continue to upgrade various parts.
Jonathan Hurst^ But you’re not wrong. Like I’ve said a couple times today, in Digit 5, there’s a lot more robot in that robot. There’s a lot more sensors and all kinds of things in there to create that safety strategy as one of the features of many that we’ve had to integrate.
And that does make the robot bigger. It does make the robot heavier than if we didn’t have to have those things, but we do. So getting to that point, and now over the next decades, you’re just going to see these machines get lighter and cheaper and everything else.
But it is meeting the customer requirements. Like it does meet the needs of the customer. So, we’re already there with Digit 5.
22
Anthony Rozmus^ I think we just have time for one last one and then we could come back up with the full management team. Andrew, let’s go there.
Andrew Boone^ Andrew Boone, Citizens. Thanks for taking the question. Jonathan, I really wanted to go back to the data flywheel that you talked about.
Just help us understand how data starts to compound in terms of tasks that you guys can do and how this becomes something that is more of like a company-wide mode.
Jonathan Hurst^ So this comes down to throughput and uptime, right? So every time, if say we’re moving those bins you saw in that example with GXO, if we slip and drop one of those bins, all of those parts are now scrap because they might have a piece of grid on them or something that can’t be reused. So, it’s a fairly expensive thing when you do have a flaw or mistake like that.
We really need to get these machines to the 99.9% reliability, something that is at least as good as a human would be doing in those environments and sometimes significantly better. And we need to get, in some cases, the throughput up. It needs to be at least in human capability for those use cases.
In this use case, the robot doesn’t have to be very fast, but it has to keep up with the machine that it’s feeding. As we are then deploying these machines, we can test and experiment with all of the little corner cases. You can kind of give out this, the data collection is when it does fail, why exactly did it fail?
Now you can test that specific scenario again and then 100 more times with slightly stochastically different grasps or grips, and that’s that practice thing. You engineer a way for the robot to be exploring in all of the corner cases where it may have failed one day, to get really good at that. And you explore and how do you improve the path in order to be more efficient and how quickly you get from one place to another.
And I believe it’s going to end up looking more and more natural as it does that because it’s going to look a little bit more like how people take their movements. And that’s what the data flagpole gets you. It gets you the throughput and the reliability.
The early deployments, you’re going to get to demonstrating the task, certainly good enough to show that you can do it, good enough to show it on video. But going from there to actually reliable deployment, it’s an order of magnitude difference. That’s what that flywheel is for.
Anthony Rozmus^ Great. Again, we’ll have one more session at the end here. Peggy and Jonathan, thank you.
Next up, I’d like to introduce Daniel Diez, Agility’s Chief Business Officer.
23
Daniel Diez^ Hello, everyone. Thank you for joining us today. So my job is to help you understand how our strategy, all of what you’ve heard on the technology side is actually being deployed and executed on the ground.
I’m going to talk about who’s actually deploying Digit today, what they’re using it for, how that translates into orders and the pipeline behind all the numbers you’ll later see from Michael. We’ve deliberately chosen a core use case in three verticals as our beachhead markets, manufacturing, logistics, and distribution. We’ve chosen these industries because they share some really critical traits.
Now, as you heard from Peggy earlier, they have very large labor pools. They’re under constant hiring pressure and the cost of labor is going up. And thanks to sort of inflation, tightening labor market, and we’ve got other factors, the silver tsunami, young people not going into these areas, all of this is creating an increased need for labor.
Automation and humanoid specifically can offer relief for that expanding labor gap and really can help increase productivity as established economies like the U.S. look to increase onshore manufacturing. Now, these industries are also built around facilities with very structured environments and that lets us repeat common tasks from deployments across different customers and verticals without having to reinvent what the robot is doing every single time. So in another way, it’s a very efficient means of deploying this technology.
They also provide an environment where cooperatively safe robots can operate outside of the traditional safety fences or walled off work cells. That all humanoids currently have to operate in. Critically, these facilities are already built for people to work in.
And so, we’re not asking the customer, recreate or redesign your entire operation around us. Like you might see with other forms of automation where they have to accommodate the technology to get the automation. Digit doesn’t require that.
And so, a lot of the incredible opportunity and the near term for us to deploy these humanoids in these facilities is really about taking on these tasks that have been super difficult to recruit for. So, I’m going to spend a few minutes on where we’re getting commercial traction because this is the part of the story that I think separates Agility from a lot of what you’re seeing out there in the industry right now. Today, we have deployment commitments across nine different customer facilities.
We’ve got more than 65,000 hours of real operational time with Digit 4 already on the clock. And as we sit here today, Digit is running in environments where safety can slip and downtime just isn’t an option. We’ve had recent wins with global leaders in new manufacturing and consumer brands as well.
And we’re expanding our footprint well beyond from where we started. So, when you ask customers why they choose to expand with us instead of just running another pilot, it comes down to five critical factors. Safety, accuracy, uptime, throughput, and ROI.
24
And you’re going to hear this from us again and again and again. We think these are the metrics that you should be paying attention to when you look at the humanoid industry as a whole. So let me put some of the numbers behind a couple of these things.
At GXO’s facility near Atlanta, Digit 4 was handling Spanx orders and it handled more than 100,000 totes filled at a 98% accuracy. So you heard Jonathan, what is required for us to actually be successful. Agility is already performing at those levels with the numbers we’ve published.
At Schaeffler’s Cheraw, South Carolina facility, Digit’s been moving totes of automotive bearing components onto a line for washing. And it’s handled well over a million pounds of material in the process. Every single one of these deployments, every customer I talk to comes back to the same pitch.
And it’s not a technology pitch. This goes back to what Peggy said. It’s an economic pitch.
Can we solve the problem that you have in operations? Our agreement with our customers is that the job has to get done according to the success metrics that we set together when we entered that commercial agreement. Digit is there to work, not demo.
So how do customers adopt Digit? They do it in really just one of two ways. We’ve got a robot as a service offering and then just a purchase offering.
So under Robot as a Service or RaaS as we call it, they pay a price set at a discount to their fully burdened human labor rate. So, what that means is that the ROI isn’t hypothetical or something that they think about a few years out. That customer sees value from day one of operation because the economics of the robot beat the labor costs they’re already paying for the work they often can’t find people to fill in the first place.
So under our ownership model, which is slightly different here, the customer will purchase the Digit 5 outright. And on that path, we deliver breakeven in just about, it’s just over a year, 1.1 years is the economics of the timeline. In either model, Digit becomes more economically compelling the more skills it has learned.
The robot provides relief from a whole host of issues that all of these manufacturers and warehouses are facing. High turnover, the tax of new employee training, loss of efficiency, which all happens when these shifts go unstaffed and that happens all the time.
So, the unit economics should improve even more when customers deploy Digit over three shifts of operation, something Digit 5 has been designed to do. Either path has the potential to deliver real measurable savings against a customer’s existing labor spend. And that’s deliberate.
What we’ve built our pricing on are very conservative assumptions. So, this case is designed to hold up even before you factor in the cost curve that Jen will walk you through in a bit, are any of the pricing optimization that we expect as customers start to really experience that value themselves.
25
Okay, so that economic case is only half the picture for us. The other half is why this business model compounds in our favor as well as the customers. So, RaaS and ownership aren’t just two ways to price a sale. They’re actually very two, they’re two different revenue profiles for the company.
So under our current, again, deliberately conservative assumptions, a single Digit deployment generates on the order of somewhere around $0.5 million of cumulative revenue to Agility over its five-year useful life under a RaaS contract. And that’s recognized steadily in a highly recurring pattern over the life of that contract.
Under ownership, a single deployment generates roughly $400,000 over the same five years, but the profile is slightly different. A larger portion of that revenue lands right up front when they purchase that robot. And then with recurring software and service revenue, it builds on top of that.
So, today, we anticipate that the majority of our 300 million in committed Digit 5 orders will be structured as robot as a service. They’ll be RaaS contracts, contingent upon, of course, meeting all the constructional milestones that we’ve set up in the contract. And it’s not an accident. RaaS is the lowest friction way for a customer early in their automation journey to say yes, because it reduces upfront capital requirements and it de-risks the decision.
But RaaS to us -- RaaS is at least as attractive to us as it is to the customer. It builds a recurring revenue stream, a growing installed base of robots in the field. And because the customer’s operations come to depend on that robot, being there every shift, every day, not quitting, not getting into fights, it’s a very strong retention play for us.
So over time, as humanoids become a normal line item in a customer’s capital planning process, rather than an investment on new technology, we expect a natural migration toward ownership, which brings in larger upfront revenue recognition while we retain the recurring Arc and service revenue on top of that.
So, here’s how that becomes a flywheel rather than just a sales model. Every RaaS dollar will also deliver valuable data. You’ve heard this a number of times now, because every deployed hour under contract is an hour of real operational data flowing back in through Arc.
So what does that do? It improves our physical AI. It unlocks the next skill or workflow that we can sell into a customer site or into a next customer that has a similar operation just like it. And at the same time, a growing recurring revenue base funds the manufacturing scale that drives our own bill of materials down. And Jen will share more details on that in a bit.
So, lower unit economic costs expands our margin, which we believe could exceed 70% as the business matures. And margin expansion funds faster deployment, further cost reduction, and more aggressive commercial motion, which brings in the next round of customers and the next round of recurring revenue. Recurring revenue funds the data and the cost curve, and the data and cost curve fund the next round of recurring revenue. That loop is a flywheel. And it’s why we think this model gets more efficient, not less as it scales, the opposite of most industrial hardware businesses.
26
So, we move customers through all this through a program we designed called Customer Acceleration Program. And it is a three-part program that involves a proof of technology, an onsite proof of concept, and then a three-month pilot deployment under our RaaS model. We’re selective about who we admit into this program. Many of the enterprise customers we engage with have use cases that can support at least 50 robots. So, we won’t engage until we know you’ve got the wherewithal and the need to actually scale what we’re doing in the near term.
So, these 50 robots are all beyond the RaaS model. And it works out to a minimum opportunity of around 25 million over the five-year useful life. So, 50 units times $100,000 a year times five years. That scale matters because it means we’re spending our own time and engineering effort on opportunities that can actually move the needle. And customers can then scale that same engagement into broader RaaS deployments as Digit 5 becomes more widely available by the end of 2027 and early 2028.
When a customer actually signs one of these commitments, it tells you something beyond the dollar figure that’s behind it. It means that Agility is actually earned buy-in, and not just from their tech and their R&D teams who are tire-kicking innovation, but from the business units who actually have to operate these robots. Because by the time, by that point, there’s already proven value on the floor, not just a promising pilot with the results on paper, but something that’s actually delivering for them.
And it’s built for one specific kind of customer, one that can point to a real labor shortage and has the appetite to scale human deployment in the near term. This is exactly the type of proof that gave us and one of our most experienced customers with Digit 4 the confidence to commit to $300 million in Digit 5 orders. And that’s structured under a three-year robot as a service agreement covering 1,000 Digit 5 robots.
This customer has been running Digits in live operation environments for years, handling real material handling work, shift after shift, day after day. When a customer with that much experience with a Digit decides to go deeper, that’s a stronger signal that there’s a real market for Digit than about anything else I could tell you.
Their decision to expand is informed by what they already have seen Digit do on their own production floor, in a real factory. And they see Digit 5’s new safety capabilities, the ability to work in closer proximity with people, as the unlock for taking humanoid robots further across their operation.
So under this agreement, this customer plans to expand Digit across multiple facilities, moving beyond material handling into a broader set of workflows that Digit 5 has been designed to do. Okay, what you should take away from this announcement isn’t just the number, although $300 million is a good number, it’s that the most experienced customers, the ones with the most real hours on Digit anywhere in the world, and arguably one of the most experienced deploying all humanoids, chose to expand with Agility. And that’s exactly the type of proof this business should be judged on and built on.
27
So, recently we announced a new partnership with FORT Robotics to further enhance Digit 5 safety infrastructure. And I’d like to go into a little bit more detail on that, just because it matters as much commercially as it does technically.
We’re jointly developing a solution that’ll extend our cooperative safety approach beyond the robot itself and connect Digit 5’s planned onboard safety system architecture to an off-board safety system that provides safe human detection and a stop command in the facility around it. So, in practice, that’ll give customers a second independent layer of protection and continuity of safety coverage as deployments grow from one to a fleet of robots. With every added layer, Digit 5’s safety system becomes more resilient, so the robot can be trusted in a more diverse set of environments.
And so, why does this belong on a commercial update and not just a technical one? Because in every new sales conversation we have, safety is a long pole in the tent. It’s what customers, environment, health and safety and insurance teams all want answered before they’ll agree to sign anything.
And so, a formalized, independently built safety architecture that is backed by a company trusted across warehousing, manufacturing and defense is going to shorten that conversation. It’s a direct input into how fast we’ll be able to move deployments from pilot to scale. It’s important that I tell you, this is a years-long effort, and it’s going to start with these redundant safety systems.
We will deploy them together, test them during these early access customers, test them with the early versions in the robot before we get to general availability. So, by the time we’re at GA, we have a solid safety system, but we’ll take the same approach that we’ve taken to everything, pragmatic, safety first, and solidly tested before we roll it out.
So, put together, this is what our commercial pipeline looks like after Digit5’s planned launch in 2027. You’ve got more than 300 million committed multi-year customer orders for Digit 5 that are subject to all the contractual milestones that we’ve agreed to, structured under our RaaS model with qualified potential commercial pipeline that exists right now that is in multiples of that figure. And as Digit 5’s safety architecture matures with partners like NVIDIA and FORT, we expect that pipeline to convert even faster. All right.
Now, I’m going to spend a bit of time with some customers from one of our larger partners at Schaeffler. So, I’d like to bring a bit more of this operational perspective, thank you so much, into the conversation. And so, our next two customers come from a company that knows humanoids from both sides of the equation. It builds the parts that go into them and it puts them to work in their factories.
Schaeffler is a German auto parts or motion technology company with an 80-year long history. They’ve got 110,000 employees around the world, 250 locations, 55 countries they operate in. It is one of the largest family-owned businesses in the world and its bearings and actuators and drive systems are in vehicles and industrial machinery around the world.
28
Schaeffler is now applying that manufacturing expertise to humanoid robots and estimates that its portfolio of components can cover half the bill of materials of a typical humanoid. Schaeffler is also an investor in Agility and a customer of ours.
And in 2024, they invested in Agility and agreed to begin deploying Digits for use across their entire global network of plants. And Digits now would work in Cheraw, South Carolina, doing shift work every day for Schaeffler.
David Kerr, and I just want to make sure that we have them on the line. Great. David Kerr is President of Humanoid Robotics at Schaeffler where he leads the development and industrialization of motion technology specifically for the humanoid sector, drawing on more than two decades of automotive engineering and manufacturing experience.
Courtney Baines is an Advanced Production Technology Engineer at Schaeffler and is the subject matter expert on Agility. She works directly with us and directly with Digit in their Cheraw facility. And between them, they can tell you what it’s going to take to actually scale the humanoid supply chain and what it takes to run a humanoid on the factory floor.
So, please join me in welcoming David and Courtney.
Thanks so much. Okay. Thanks for joining me, guys. I appreciate it.
Let’s see. So, David, I wanted to start with you. Schaeffler was early here, early in investing, early in trying to figure out how to build out the business, early to deploy. What I want to understand is, where did the impetus come to start so early? And where do you see humanoids fitting across the organization as you’ve now been in it for a number of years?
David Kerr^ Yes, first of all, thanks for having Courtney and I here today. It’s a pleasure to be with you guys in this important day. I mean, the impetus comes from a couple of things.
First of all, we see physical AI and humanoids in particular as a competitive necessity, especially in our plants. So we see this from, I think it was mentioned earlier from you, Daniel, that we just can’t find the workers that we need. And certainly when they need to do dull or dirty jobs, it’s really difficult to even retain them even if you can get them in the door.
So, we really think humanoids and physical AI is going to fill this void that is only going to become bigger as the world’s population starts to age out. And as you mentioned, we have 100 manufacturing plants around the world. So for us, this is not just a regional topic in the U.S., it’s a global topic.
If you then compound that with the benefits of physical AI that we expect to get, there’s a productivity gain. So, if you look at these productivity gains and then you add in the fact that we need to fill a worker gap, this is really a good impetus into why we’re so involved with physical AI. And then if you look on the supply side, we’re a motion technology company. We’ve been pioneering motion for 80 years and all of the products that we have for either industrial or automotive, most of these translate very, very nicely over into physical AI and humanoids.
29
Daniel Diez^ Thank you, that’s great.
And so, Courtney, you’ve been leading on the deployment side for Schaeffler here in the U.S. You have Digit on the factory floor in Cheraw, South Carolina. Can you tell me a bit about what has it been like for this early deployment? And specifically, what was it about Digit and Agility that made you start with us?
Courtney Baines^ We really wanted to implement humanoids because we see them as the truly the next step in innovation to achieve flexible and modular productions. Schaeffler has implemented thousands of robotic solutions all the way from industrial robots to co-bots to mobile robots. So, we saw this as the obvious next step and we wanted to get in early.
When choosing the first use case to be in Cheraw, there were a lot of factors we were looking at. First and foremost, I was looking at scalability. These are standard Schaeffler products that they’re moving. It’s something that we can build upon skills and do more and more in the future.
Feasibility, I needed a use case that was currently within the ability of the Digit back a year and a half when we launched it, that was Digit 4 Alpha. So it was a lot of bulk handling and movements such as that.
And then last but not least was sustainability. Dave touched on it a bit, but we’re really looking to automate any jobs that are repetitive, ergonomically challenging or physically demanding to take that strain off of our human operators.
Daniel Diez^ So help me understand, Courtney, how Schaeffler is thinking about this. As we’ve been there now, as you mentioned, for well over a year, year and a half, and we’re now evolving into the Digit 5 platform. For Schaeffler, what has to be true on the technology side, on the safety side, in terms of the operational metrics that you track, what needs to be true in order for you to start to scale as we are starting to do at Schaeffler?
Courtney Baines^ First and foremost for scalability is going to be safety. Safety comes above all else, especially in our plants. We need to ensure we have a safe environment for any of our human operators or just humans in general in the facility.
Right now, given the lack of ISO safety standards, it is necessary to put guarding around any humanoid deployment we have. And that can be difficult in choosing initial use cases just because we can’t block any critical pathways, we can’t prevent access to a machine in case we need to perform maintenance or something goes down.
So, I think Schaeffler, like a lot of companies looking into humanoids, you have a list of use cases that you really want to do and really want to implement with, but it’s going to be hard to do until we get the safety standards.
30
Safety standards can be established, humanoids can be built to them, and humanoids can be certified to those standards that will enable us to put them in the same vicinity as a human. So all the other things, factors you listed are very important. We care about accuracy, repeatability, uptime, maintainability, but nothing comes above safety for us.
And I think that’s going to be the first real hurdle we have to go through, and then we work on all those other hurdles.
Daniel Diez^ I think it’s one of the major cultural alignments between Agility and Schaeffler is that that is the first question we ask, it’s the first question that you ask. David, we’ve now been together for a while. I think you’ve laid out the case for why humanoids in operation along with Courtney and why this is going to be important.
How the market is growing and how the need will be growing for Schaeffler on the operational side. That has opened up a massive opportunity for Schaeffler in motion technology. How do you see Schaeffler’s role as a leader in supplying this market and not just being a customer of finished robots, but supplying components and even designing componentry that’s going to help evolve the humanoid market?
David Kerr^ That’s a great question. Our first goal is really globally to be a leading supplier of physical AI and in particular humanoid motion technology. So, again, a lot of what we do in automotive translates really nice. We’re not starting from zero. We do a billion euros a year worth of actuators for automotive and industrial. So, translating that already into physical AI and humanoid space, that’s already begun.
And like I said, we don’t start from zero. So, that makes it a very compelling opportunity for us. What really helps too is that our executive team then is at the foresight to really require the team at Schaeffler to begin deployment.
And what happens with deployment is from my side or my supply side situation is that we can learn from these deployments. So, we see how this physical AI and these humanoids are actually used in the plants. We can measure, we can simulate then what is going on with the components, and we can design components that are not under-engineered or over-engineered, but they’re engineered for the right application with the right specification.
And so, it really lines up great with the portfolio we have and translating that over and then also taking what we’re learning as a deployment partner.
Daniel Diez^ Great. Well, we have limited time today, but I really wanted this audience to understand Schaeffler as a company and the role it’s playing in humanoid, the humanoid market in general, but how we’re together really advancing the deployment of humanoid technology and then the advancement of the design and the manufacturability of these robots. So, I thank you both for helping to help the audience understand a bit more about this.
And I please everyone join me in thanking Dave and Courtney for the time today.
31
All right, with that, I will now introduce you to our Chief Operating Officer, Jennifer Hunter.
Jennifer Hunter^ Thanks, Daniel. And thank you to David and Courtney for that conversation.
It’s one thing for us to describe demand, but it’s very different to hear it right from the customer themselves. My job is to answer the question that naturally follows all of that. Can we actually build enough product fast enough to keep up with the market demand?
I spent a decade at Amazon’s operations organizations before joining Agility. And if there’s one lesson that I learned in being there is that the hardest part of installing and adopting technology and robotics isn’t the robot itself. It’s really everything around it. So, let’s talk through what we’ve built.
RoboFab is really our foundation. It is the world’s first purpose-built assembly facility of humanoids located in Salem, Oregon. We stood it up in 2023, designed from day one to really assemble and produce humanoid robots rather than retrofitting an existing production facility.
RoboFab uses a modular architecture that lets us scale output as our demand grows. With the ability to support production volumes up to 10,000 units per year. This also does not require any additional large capital outlay.
One detail that I’d point to, many of our engineers sit side by side on the manufacturing floor next to the people building those robots every day, not in a separate building. That creates a tight feedback loop and it really allows us to engage between design, manufacturing, and what we’re learning directly from our customers’ facilities. And it’s the meaningful reason that we’ve been able to iterate as fast as we have.
We don’t manufacture or design all of our components ourselves and we don’t outsource everything either. We’ve been very, very deliberate about where we add inherent value and differentiation. We design and produce the systems that are hardest to replicate and matter most to the performance in-house in RoboFab.
We start there with our actuators. Those represent the largest portion of Digit’s bill of materials and are the single biggest driver of the cost curve. For components that are more standardized, we lean on manufacturing partners who can bring scale that we’d otherwise spend years building ourselves. This approach gives us control over the product that actually differentiates us while keeping the business capital efficient as we grow.
On the supply side, we’ve developed a supply chain roadmap which optimizes for quality, lead times, and cost. We source domestically wherever feasible. That’s a deliberate choice. It gives us tighter quality control and a supply chain that is far less exposed to the geopolitical uncertainty that this industry is watching very closely right now.
A critical question that we get is whether our product quality and reliability will hold as volume ramps. And the honest answer there is that we treat this as an engineering discipline, not an afterthought. This begins upstream with our suppliers. We qualify, we audit, and we provide real-time feedback to every inbound component to ensure it meets both our specification requirements and our quality needs.
32
In addition, every Digit unit that we produce goes through rigorous testing before it ever leaves RoboFab. And because, as I mentioned previously, our engineering and manufacturing teams sit side by side, any issues get caught immediately and corrected at the design level. Not just remediated on the line.
Our discipline around quality and reliability also extends to each of our customers’ facilities. In order to further minimize downtime for our customers, we service all of our customer units with the support of a third-party field service provider. That is our first line of defense. It allows us to, again, minimize that customer downtime and to really scale very quickly across all of our customer deployments.
The design of Digit is also facilitating that. We design in a way that allows for quick repair in the field with field-replaceable units snapping on and off the torso, like arms and legs.
And as we mentioned previously, because Arc gives us live health data and diagnostics on every robot that’s deployed, a lot of that turnaround happens even before the technician ever reaches the customer facility. For issues that might require further remediation, we have a service center that’s located in our RoboFab facility in Salem, Oregon. We’re also standing up a second facility in our Pittsburgh location.
Our current customer turnaround time for any service-related items that go off-site to the service center is under five days. We’re actively working that down to under three. So, what I know you’ve all been waiting for, let’s talk about the bill of materials, the cost structure, because that’s really where the operations story turns into a financial one.
Over the past 10 years, we’ve designed and built our product made for work. It’s that experience that has allowed us to drive Digit 4’s bill of materials down to roughly $125,000 per unit. We’re running that same playbook on Digit 5 today.
It’s the plan that delivers a clear path towards a bill of material which costs approximately $30,000 per unit at Digit volumes of 10,000 a year. I want to be direct about something. This cost curve is not simply a spreadsheet exercise and it isn’t riding on one single lever either.
There are really three different mechanisms that drive this cost curve. Engineering redesigns, supplier initiatives with strategic partners, and finally, pure volume-based tier discounts. It’s worth discussing each of these areas separately so you can see why we’re confident about each.
The first is the piece that we control directly. Since we have developed our own design, which is the core to our dynamic stability and our payload capacity, we also own the ability to redesign the actuator components from start to finish driving costs down. Today, our actuators are the main driver of this category and they’re amongst the biggest drivers of Digit’s bill of materials, comprising roughly 25% of the overall cost of Digit today.
33
Our plan to reduce costs here comes from better design, iteration, better manufacturing processes, and really the learning that we’ve built up over years of producing these products in-house at RoboFab, real-time on the floor, not simply through a forecast.
The second area for enabling cost and reductions there is a different lever entirely. These are supply chain initiatives. These initiatives are applicable across all of our components, including the more standardized subsystems, things like cameras, sensors, PCBs. And it’s these components that we’ve begun developing strategic partnerships alongside key partners to help us drive costs out through scale that we don’t have to do alone.
We don’t have to build ourselves. Bringing in manufacturers with existing global supplier relationships and precision manufacturing really drives better pricing, shorter lead times, faster than years of enabling that on our own.
The third area of cost reductions will come through growth as volume scales, and we can benefit from that volume-based tier discounting. This delivers roughly 20% of our projected costs down as we move into volumes of 1,000 and then 10,000 units annually.
Putting these three levers together, what we control ourselves, what we get from scaled manufacturing partners, and volume-based discounts are the basis for a clear path of a bill of material moving from $150,000 at commercial launch of Digit 5 to a target of 30,000 per unit at commodity levels. It’s not one bet on engineering alone. It’s engineering on the piece that differentiates us and partnership on the piece that doesn’t need to.
The attractive unit economics we’ve discussed are only really meaningful if you have the infrastructure required to scale production and support customers. We believe that this is one of Agility’s most important competitive advantages. Over the past decade, we’ve invested in not only building a robot, but the broader platform that is really required to deploy humanoids at commercial scale.
That includes RoboFab, Arc, the software program we discussed earlier that really manages the deployment and fleet of the robots, a predominantly domestic supply chain, and ownership of the highest value components and subsystems within Digit. That’s the operational engine behind everything you’ve heard today.
I’d like to now bring on our next guest who represents a company that has backed Agility for years. Foxconn is the world’s largest contract electronics manufacturer. Founded in Taiwan in 1974, Foxconn builds hardware at enormous scale for many of the best known names in technology.
Foxconn is also a longtime investor in Agility. And this year, they led our internal pipe in the business combination with Churchill Capital XI. Robotics is now central to where Foxconn is going. The company has named it one of the three emerging industries in its long-term strategy alongside electric vehicles and digital health.
34
When the company that knows more about -- more than almost anyone about manufacturing at scale chooses to invest in a company like Agility, it says something about where humanoids are heading and the path to get there.
Nelson Hsieh is a Director in Foxconn’s Central Strategic Investment Department. He brings more than 15 years of experience in both investment and fund management. Nelson can speak to why Foxconn invested, why it chose to lead the internal pipe, and how one of the world’s greatest manufacturers thinks about the future of humanoid robotics.
Please join me in welcoming Nelson Hsieh.
Welcome, Nelson.
Nelson Hsieh^ Yes. Thanks, Jen.
Jennifer Hunter^ Thank you for joining us today. So, what I’d love to start off with is how Foxconn is thinking about its role in the humanoid robotics space.
01:46:17 Nelson Hsieh Nelson Hsieh^ Okay. That’s a very good question. For Foxconn, before I answer this question, maybe I can give you some figures.
For Foxconn, we produce nearly 40% of world’s consumer electronics. Actually, we operate in more than 240 campuses in 24 countries. And every season, we employ 900,000 employees. So, why I need to say this, because all this gives us the practical view of humanoid can help Foxconn, how this bring the value to Foxconn.
Also, you may heard of Foxconn have 3+3+3 strategy. Robotics is one of our core industry we focus on, along with electric vehicle and digital health. The problem for us is never to purchase the equipment. That’s easy for us.
The key question now is that people who are willing to do repetitive job, that’s the problem for us, especially now in China. So, in this field, actually, we deploy it because we are our toughest customers, that’s for our top customers, like Apple and some other media, something like that. And also we build robotics because bring the robotics to dependable products is a manufacturing problem. And we are very good at that.
And also, we not only build, we partner because in our concept, we believe no player in this field can define the whole market. That’s why we also partner with Agility. There’s something we can cooperate with.
So, other than that, I would like to talk more about Foxconn development on robots. And actually this year, we unveiled our AI power industrial robots at GTC in March. And we showed that again in Computex in June. Also, we bring to Paris in VivaTech.
It’s a wheel to arm robotics. It can use to pick up parts, press it. Also, it can do the collaborative assembling.
35
Actually, for two wheels, it can work very well, especially for within a few step of workstation. And for legs, like Agility, it works very well in unconstructed environment. And from our view, we believe both will coexist.
And finally, the market will be segmented by the task. And other than the robot we bring to market, we also have FastBrain. And currently, it’s FastBrain 2.0. It’s our enforced large language model. It connects AI model and also simulation and physical operations so that the machine can really understand the true environment.
Our position is quite simple. We would like to move the people to the high value work, not to repress them. And also, deploying the robots and also building robots are different disciplines. We happen to have both. So, in our position, we are happy to help the company who already have commercial progress like Agility.
In our view, we are not trying to own every robot. We are not trying to define the technical paths. We would like to be part of the ecosystem and help the ecosystem grow.
Jennifer Hunter^ Thank you, Nelson. And in that vein, why Agility? Why have you bet on Agility through the demonstration of your interest and investment in the internal pipe in this transaction?
Nelson Hsieh^ Okay. Actually, as I just mentioned, robotics is one of our core strategy. So, from our view, the question is never should we be in humanoid.
The question is who is in this field already ready to scale and how we can help them to move faster. So, in terms of investment variation, actually, our team have looked at so many humanoid robotics globally, including U.S., Europe, also China. We always ask us three key questions.
Is this company already have paid customers? That’s very key for us.
And secondly, can this company scale in big volume? The second one.
And third, most importantly, in our factory, can this robot can safely work with people?
These are three questions we examine all these companies. And just a few of them really pass all three criteria and actually pass all three. Oh, as Daniel just mentioned, actually have engaged, not engaged, have a deployment with Schaeffler, GXO, and Toyota.
And he just showed us some numbers, like they have nine customer facilities. Also, they have like 30 [pylon] customers, potential customers. And also, when we do the evaluation, I remember the number. Agility also have 300 million multi-year contract.
No customer will sign a multi-year contract if it’s just a demo. So, Agility is ready to enter the mass production already. And also, they have a factory in Oregon. They design to like up to 10,000 units.
36
So, Agility in this field is ahead based on our criteria. And also, we find that in the [impact run], the existing investors also do the follow-on investment. So, we clearly understand the management team and existing investor stand the same boat with us. So, that’s why we would like to lead this round and be the anchor investor. Yes.
Jennifer Hunter^ Thank you. And we agree that real-world deployments definitely trump demos and hype videos.
Nelson Hsieh^ Yes.
Jennifer Hunter^ So, thank you for that. One final question for you, Nelson. Given the tightening export controls, how, and that’s really coming across both on the AI side and compute robotics components, what’s the realistic risk to manufacturing’s geographical mix?
Nelson Hsieh^ Okay.
Jennifer Hunter^ And how are you positioned as Foxconn to really address that rapid growth that is expected in the U.S.?
Nelson Hsieh^ Okay. Actually, for Foxconn, we don’t focus policy. We also don’t speculate regulations.
But indeed, this export control would bring some risk to our customers. Something like they have to -- like they are sourcing supply chain, something like that. And for Foxconn, it’s not just moving a location from China to U.S.
It’s not that easy. We have to qualify like supply chain components or also manufacturing capability in U.S. And in our Q2 call in August, actually, we already said to the public that we will build our R&D and also manufacturing capability in many states.
And for robotics, it’s the same situation. So, it’s not a greenfield point for us. We already built our footprint and also the manufacturing capability here.
I’m not saying that that will be easy for us in the future. But the key thing is we would like to stay close to our customers. And we already built a footprint here.
So, yes, that’s the key point. Yes.
Jennifer Hunter^ Wonderful. Thank you, Nelson. Thank you very much for your continued support, Foxconn support. And we look forward to continuing to work together in the future.
With that, I’d like to hand it over to our Chief Financial Officer, Michael Beer. He will connect and communicate all of the things that we’ve spoken about earlier, the commercial momentum, the manufacturing discipline, and really bring that all together into the financial picture.
37
Michael?
Michael Beer^ Great job, Nelson.
Okay. Thank you, Jen, Nelson. And thank you to all the analysts and investors for joining us here today.
I spent over a decade in your seat as an analyst at Bear Stearns, Wolfe Research, Citibank, amongst others, covering public companies across the logistics, mobility, and infrastructure landscape. I’ve also scaled dynamic innovative technology companies and served as a public company CFO. Drawing on that experience, my objective today is to tie together what you’ve heard, help you understand the economics of a Digit deployment, as well as our value proposition longer term, and explain how we intend to measure success.
I’ll focus on four key takeaways I want to leave you with.
First, proven demand. We have validated customer traction with commercial engagements today based upon a proprietary design built for scale, $300 million in committed customer orders subject to the achievement of certain milestones and real deployments.
Next, healthy growth outlook. We articulated how a robust pipeline and growing demand picture will translate into future business wins through our turnkey offering, combining Digit with embodied AI and Arc.
Next, how our anticipated margins are set to expand. We have attractive deployment economics with multiple levers to expand margins as we add skills and we drive down costs.
And lastly, capital efficient growth, built on investments we’ve already made with access to an advantaged cost of capital to support future deployments.
Let’s start with the demand and the scale of the opportunity. Peggy outlined our near-term opportunity in manufacturing, distribution, and logistics, industries where the labor gap is acute. There are over 400,000 unfilled manufacturing jobs today with nearly 2 million potentially unfilled by 2033.
That is a structural shift compounded by changes in an aging labor force, reshoring, rising transportation costs, and trade and immigration policies. Further, the humanoid form factor is perfectly suited for those functions and in facilities initially designed for human labor over decades. In the U.S. alone, there’s over 20 billion square feet of industrial space, much of which is over 40 years old and yet to benefit from the upgrades and the capital investment of a brand new facility.
In more modern facilities, we expect Digit to work alongside other forms of automation, not an either-or. The broader market opportunity is estimated to approach $1 trillion by 2032. Our opportunity expands as Digit learns new skills and addresses more of the work customers need done.
38
And because the value of the work is linked to labor costs, we believe our offering provides a natural hedge against labor rate inflation. We bring real experience to that opportunity, more than 65,000 hours of operational experience with approximately 98% accuracy at Schaeffler and GXO. That experience helps us to understand what customers require and what it takes to deliver consistently.
It also informed our recently announced next generation product, Digit 5, by truly incorporating the key requirements our customers demand for real-world deployments designed to address a binding constraint on humanoid deployments, which is working cooperatively alongside people. To bring new customers into that process, we established the Customer Acceleration Program in late 2025.
It provides a structured path to validate performance, integration and economics before expanding. Customers pay approximately $500,000 to participate in meaningful commercial commitment. We’ve signed four new CAP customers and are engaged with more than 30 prospective customers through active commercial discussions.
The $300 million in committed Digit 5 orders demonstrates how that translates into demand. That commitment relates to 1,000 robots under three-year RaaS contracts. It represents potential multi-year value realized as contractual milestones are met and robots are deployed.
With the recent announcement of Digit 5 and the subsequent rollout next year, our goal is to end 2027 with a potential customer pipeline of up to 100 customers and up to 25 committed customer facility deployments. Meanwhile, we believe our committed orders will commensurately increase to reflect our goals.
From there, we have five growth levers.
One, adding additional robots within a given facility. Two, expanding across a customer’s facility network. Three, introducing new skills and workflows that move up the ladder, expanding our share of the wallet. And four, adding new customers. And lastly, expanding into new verticals.
Over time, our skills roadmap moves from material handling into component handling, precision operations, and more complex autonomous tasks. Each new skill can expand the opportunity within customers we already serve.
And those levers reinforce each other. More skills create more applications within a facility and validated applications create opportunities to expand across facilities. As you can imagine, satisfying an unmet need or a labor gap for a role with a fully burdened labor rate of, say, $40 an hour is far more valuable to a customer than one at $30 an hour. Assuming they can find a person to fill that role in the first place.
Now, let’s turn to what we earn on those deployments. As Daniel discussed, customers are currently leaning towards our Robots-as-a-Service offering, particularly for initial deployments because it reduces upfront friction and more closely resembles the ongoing labor cost.
39
So, I’ll use that model to illustrate these economics. We are looking at an assumed five-year useful life per Digit. And these are product level economics, including deployment and ongoing support, but exclude corporate R&D and SG&A, which I’ll talk about in a moment.
Under RaaS, we receive annual subscription payments and an approximate $25,000 deployment fee. Over five years, those payments generate approximately $500,000 in illustrative cumulative revenue per robot, or $100,000 per year on average. For the customer, the annual subscription represents a significant savings to their fully burdened labor costs, particularly when you factor in our ability to operate for three shifts per day.
And the customer also receives access to our Arc fleet management tool and maintenance services. Assuming a full-time, fully burdened labor rate of, say, $30 an hour, this would equate to roughly $180,000 per year in total costs the customer would otherwise have incurred. And that’s without the impact of sick days, lunch breaks, or other variables and variability that could eat into productivity.
Our costs include the upfront bill of materials, approximately $15,000 to deploy the robot, and approximately $15,000 annually to deliver software and maintenance. At a launch bill of materials of roughly $150,000, that’s approximately $240,000 in cumulative costs against that $500,000 in cumulative revenue for an approximate 50% product margin over the useful life.
We recover our initial bill of material investment in approximately one and a half years in that scenario. Specifically, that payback measure is in the time until RaaS subscription receipts exceed the robot’s bill of materials. Deployment and ongoing service costs are captured in the lifetime margin calculation. As we execute down the cost curve Jen described, both margins and capital recovery will improve.
At approximately 1,000 units of annual production, illustrative RaaS margins each will reach approximately 70% with payback below one year. And at approximately 10,000 units, margins will reach approximately 75% with payback below six months. That progression assumes no pricing upside.
It reflects engineering, manufacturing, and supply chain improvements alongside higher volumes. Lower costs increase what we earn on each deployment and reduce the time required to recover our initial robot investment. On that basis, 2,500 deployed robots represents approximately $250 million in annual revenue potential.
And at 10,000 units, that would represent approximately $1 billion. These are illustrative installed-based scenarios with no deployment date attached. Actual annual revenue depends on deployment timing and contract terms.
For customers purchasing Digit as a capital asset, ownership brings more revenue forward, approximately 65% in year one, assuming a five-year useful life, and more than covers our bill of materials at the time of sale. We retain recurring software and maintenance revenue with illustrative five-year product margins of approximately 40% at launch, increasing towards 70% at scale. Pricing provides an additional opportunity under either model.
40
We offer savings against fully burdened labor costs while helping fill jobs customers struggle to staff. That allows us to tailor pricing to the value delivered. As Digit gains skills and addresses higher wage work, we believe our pricing opportunity increases.
Greater utilization strengthens the customer’s economics further. Neither opportunity is required to achieve the cost-driven margin progression shown here.
And that brings us to capital efficiency. We’ve already invested in the infrastructure required to scale, RoboFab supporting up to 10,000 units annually, ownership of Digit’s high-value systems, a predominantly domestic supply base and Arc for deployment and fleet management. We also have deep proprietary pool of physical AI data earned through years of real-world deployments. Our commercial arrangements provide access to operating data across both RaaS and customer-owned deployments.
Each deployment, therefore, contributes to the data, service knowledge and workflow experience that can improve reliability, accelerate the development of new skills and make subsequent deployments more efficient. All the capital we’re raising as part of this transaction is intended for deployment-related activity, building robots, putting them to work, supporting them and advancing the engineering and software that expands their capabilities.
We’re well-capitalized today. Going public gives us substantial resources to accelerate that activity. And it does not change our discipline around corporate costs.
Our illustrative model assumes approximately $60 million in SG&A in 2026, growing approximately 20% per year through 2028 and approximately $115 million in R&D, growing approximately 15% annually over that same timeframe. Longer term, those assumptions are approximately 10% and 15% of revenue respectively. That’s leverage. The platform and the organization we built should support a substantially larger deployment base, allowing revenue to grow faster than the operating expenses over time.
Our infrastructure capital expenditure assumption is approximately $8 million per year through 2028 and approximately 0.5% of revenue longer term, really very capital efficient.
Funding robots deployed under RaaS is a separate requirement. These are illustrative model assumptions, but provide a good framework for understanding how we manage -- how we intend to manage spending as we scale.
Now, briefly on the transaction. We’re going in public in partnership with Churchill Capital XI, which trades under the ticker CCXI here on NASDAQ. Churchill is a proven sponsor that has helped bring critical technology companies including Oklo and Infleqtion to the public markets.
We expect over $620 million in gross proceeds from our SPAC transaction, including more than $420 million from Churchill’s cash and trust and $200 million in the common stock pipe, which was secured before the announcement, anchored by Foxconn and priced at our $2.5 billion transaction valuation.
41
Agility and Churchill have filed a publicly available registration statement on form S-4 with the SEC, which has not yet been deemed effective. We do plan to close this transaction this quarter, customer -- subject to customary closing conditions and trade on NASDAQ under the ticker AGLT.
Finally, how should you measure our progress? The model starts with units. Bookings become backlog and backlog converts into deployments as product, site, and workflow readiness allow. Customer readiness can be as important as manufacturing capacity in determining that timing.
For RaaS, revenue depends on active robots and the portion of the period in which they operate. The ownership model brings upfront hardware revenue followed by recurring software and services. And that mix affects revenue timing, margin and cash requirements.
Our disclosures will focus on product milestones, bookings, and backlog conversion into deployments and expansion across customer facilities. Active installed units, adoption mix, recurring revenue per robot and costs will serve to help connect that commercial progress to financial performance.
We look forward to seeing many of you at CES and continuing the conversation through our business updates and results call.
With that, I’ll turn it back to Peggy for closing remarks.
Peggy Johnson^ All right. Thank you. Thanks for hanging in there with us today.
And thanks to the rest of the Agility team who spoke, Michael, Jonathan, Daniel and Jen, along with our partners who joined us. Thank you very much.
As I said at the start of the day, that you would leave with a clear answer to these questions. Why Agility and why now? Let me bring that together in one place.
You heard from Jonathan why this is happening now. Physical AI has finally caught up to the real world and Agility is the company that spent a decade earning the harder half of that problem. You heard why Digit 5 safety architecture won’t just be a feature.
And it’s the unlock that lets us work in close proximity to people and why that’s fundamental to scaling humanoids. You heard it made real by the people building alongside us, customers doubling down and after years of hands-on experience with Digit. And you heard from Jen and Michael why the economics behind all of it aren’t a forecast.
They’re already underway. None of that happens without the policy environment, keeping pace with all of it, which is exactly why we wanted to have that conversation in this room today and not at the side of it. It really is front and center.
42
So, here’s the simple version. This is a business already generating revenue, commitments from customers who’ve already seen it work and we are asking to raise growth capital against demand we already have. We think that’s a rare thing to find this early in a category like ours.
And we’re glad you spent the afternoon with us learning more about Agility. We’ve got time now for one more round of questions with anything we didn’t cover or we didn’t go into well enough. Anthony will come up and help us with that.
Thank you.
Anthony Rozmus^ Thank you, Peggy.
While we bring up management team on the stage for our final Q&A, I’d like to mention that Pras, Agility’s Chief Technology Officer, will also join. As before, we have a few handheld mics being passed around. We ask you to state your name and your firm that you’re with before we -- once you -- before you ask your question.
And with that, we can get it started. We can go right here. Mark, right here. And then Ken will hit next.
Mark Delaney^ Mark Delaney with Goldman Sachs. Thank you very much for putting the presentation on today and for having me. I was hoping you could speak a little bit more on the path to the 10,000 units that you spoke of having the capability to produce in your Oregon fab.
How much of getting to that 10,000 level is about new technology development, maybe on the software side with more skills, upgrades to the V5 platform with different hardware, or is it more just kind of blocking and tackling and ramping the supply chain? Thanks.
Peggy Johnson^ Great. I’ll have Jonathan and maybe Pras talk about the technical things we’re doing. And then, Jen, you can fill in on volume and supply chain.
Jonathan Hurst^ Yes. I mean, it just starts with, okay, what does it take to build a robot? And then planning out, what are all of those physical pieces? And then kind of laying out the factory of what it takes to actually achieve that. That’s kind of the short and simple answer.
Pras Velagapudi^ With the V5 product, one of the major pieces that we incorporated is a lot more modularity to be able to move more quickly between different use cases. So, the capability to swap out end effectors to handle new areas of case and item manipulation is really part of the unlock and scale in addition to the safety. So, the Digit 5 design is really us gearing up to match that increased demand with increased capability through software unlocks and incremental hardware unlocks rather than generational ones.
Peggy Johnson^ Well, Jen, you have your mic on, right?
43
Jennifer Hunter^ I would just add a couple of things. From an operational readiness perspective, there’s nothing standing in our way between today and 10,000 units being assembled within RoboFab. We’ve also begun, as I mentioned earlier, building that supply chain.
We know that the suppliers that we’re engaging with can grow alongside us really to meet that demand. So, really, what it comes down to is the pace in which our customer can ingest those units. And with any new disruptive technology and the program that Daniel and team have put together, we’re really accelerating that pipeline and that deployment schedule.
Anthony Rozmus^ Great. We’ll take the next one from Ken.
Ken Newman^ Hi. Thanks. This is Ken Newman from KeyBanc Capital Markets. I really appreciate all the color on the opportunity to drive costs down on a unit basis going forward.
And I do think that there’s this broader thought that as the adoption of new humanoids kind of comes to market, that there’s an opportunity for the selling prices on a per unit basis also to come down, right? And so, to help kind of drive that adoption even further, the conversation today kind of seems like it’s assuming a static type of pricing environment. But I’m just curious if you could talk a little bit about how you think about the momentum in ASPs and how that should trend as your competition also scales.
Peggy Johnson^ Yes. I think I’ll have Michael address that.
Michael Beer^ Sure. So, the fact of the matter is you shouldn’t think of us like you do a piece of hardware like a TV, a flat screen TV. We won’t follow that same cost curve.
We are competing against the equivalent fully burdened labor hour. And as a result, and as we add additional skills, there’s a real defensibility around that, particularly when you’re fighting against things like inflation and so forth. So, we believe that as we introduce new skills, at a minimum, it’ll allow us to defend our ASPs.
And arguably, we’re unlocking more value for the customer and can actually increase both the share of wallet and climb the ladder of those individuals where we can -- where we can truly unlock value.
Anthony Rozmus^ In the back there, and then we’ll get Craig next.
Chris Moehle^ Hi. Chris Moehle, Robotics Hub. Have a semi-related follow-up question because I think two of the more important things you hit on were the safety system and the time it takes to build that and then the flywheel effect that you mentioned for your customer data.
But do you have a way to concisely relate that to barriers to competitive entry and your ability to maintain your actual unit price?
Peggy Johnson^ Jon?
44
Jonathan Hurst^ I’m sorry, yes, can you repeat that last part? Relate that to the barriers to entry and --
Chris Moehle^ Barriers to a hypothetical competitor getting $1 billion tomorrow and your ability to maintain your unit price moving forward.
Jonathan Hurst^ Sure. So, I mean, a lot of what goes into this is really about the engineering that went into it. It’s about understanding the product requirements because we’ve actually been deployed. And it’s going from that idea of a group of engineers imagining what a general purpose machine should be like and then building it, and then encountering reality as you actually try to deploy these things in a general purpose environment and need to write down very specific things about the first five, 10 use cases and workflows you’re going to do and then incorporate that into a machine.
And so, like the two or three years that it has taken us to invent and create the safety strategy and integrate that into our design, it’s a pretty fantastic moat. Any other machine out there that doesn’t have some of the basic features of the sensors that you see and the way the E-stop is and the safe motion certifications throughout and a lot of that, it’s going to be a huge effort to redesign a machine to catch up. Yes.
Pras Velagapudi^ Yes. Speaking a little bit around the flywheel effect, when we talk about the flywheel, especially as it pertains to AI, one of the big things to relate it to is the cost of deployment. So, when we go into a new customer workflow, there is a certain amount of NRE to bring the robot to that new space. As we build up this flywheel, what happens is that cost of deployment, the incremental cost of deployment is going down.
It’s more similar to scenarios we’ve seen in the past and that makes it faster and easier to move into that new space. So, that really lets us sort of hold the line, but also increase the speed and decrease the incremental work of deployment. Similar to almost the mental equivalent of reducing the change to the environment, we’re reducing the amount of change to process that people might need to contemplate in bringing a robotic solution in because there’s more adaptability in the platform itself to meet their processes coming from that diversity of data that we’ve seen in the past.
Jonathan Hurst^ Yes. As the robot gains more skills, it also gains more ability to gain more skills, if that makes sense.
Anthony Rozmus^ Craig?
Craig Irwin^ Thank you. Craig Irwin from Roth Capital Partners.
So, I realize the 65,000 hours that you quote are for Digit 4, right? And the $300 million in committed orders is Digit 5. I was curious if you could maybe talk a little bit about how well defined the tasks or workflows are on your customer side in those committed orders.
45
And then maybe if you could share a little color on any training or final engineering involved. And many of us have to keep financial models. Is this something that’s going to impact revenue progression?
Can it be linear based on your RoboFab production? Or is it likely to be a little bit more lumpy based on customer commissioning to their facilities? If you could just unpack this picture for us.
Peggy Johnson^ Daniel, did you want to jump in?
Daniel Diez^ Sure. So, when it comes to the committed orders, the agreement is structured in a way that the robot is required to meet, be able to perform certain use cases. And each use case unlocks new tranches of robots. It’s worth noting that every single use case that is contemplated or specified in that contract are part of the Digit 5 design program.
So, those are not skills that are unachievable or sort of a stretch, if you will. So, that was very deliberate.
I think the other part of it is that the KPIs for those use cases, we’ve talked a lot about that throughput, accuracy, uptime, those are also things that have to be mutually agreed upon between the customer and Agility. And the reason for that is, I mean, uptime and accuracy are sort of table stakes. It’s the throughput piece that we really have to look at as you put Digit in new environments, you have to make sure that you’re accommodating for what the customer is asking for.
And so, I half-jokingly will say, if a customer asks Digit to walk through a maze before it gets to the task and then walk back to complete the task, throughput will be a little bit different than if it’s just walking in a straight line. And so, those things have to be considered. But all of it is stipulated in the agreement, and none of it is something that is outside of our ability to achieve.
Craig Irwin^ The revenue recognition?
Michael Beer^ Yes. So, from a revenue recognition perspective, this is over, it’s a three-year contract spread over four years. So, there will be a ramp component to that.
On a RaaS contract, it’s based on the services provided. So, there’s a time component based on, multiplied through by the number of robots that are deployed in any given period. But as you can tell, having three years of service over a four-year period, the ramp should be relatively quick and you should be able to model that out.
Craig Irwin^ Thank you.
Anthony Rozmus^ Perfect. I think we have time for two more questions here.
We’ll go to Winnie, and then I think, Joe, you had your hand up as well.
Winnie Dong^ Hi. Thank you. Winnie Dong from Deutsche Bank.
46
Can you talk about competition, who you view as strongest competitors within the landscape? Understand that there’s enough in the space for everyone to sort of grab, but longer term, what do you think is your competitive mode? And how sticky do you think your products could be in terms of clients potentially deploying two different suppliers, but then how difficult or easy is it for them to transfer in and out?
Jonathan Hurst^ So, I’ll start by saying you have to reframe a little bit about how you think about competitors. It isn’t humanoid robots because almost none of the humanoid robots that exist compete with us. They have two legs and two arms maybe, but they aren’t picking up bins and moving them around and doing industrial work. They don’t have any safety, et cetera.
So, the competition actually, that shrinks you down to not a lot. And then the competition is really the status quo. It’s having people do the work. It’s other forms of purpose-built automation that teams are always working to develop.
But we’re on such a ramp of starting with a beachhead, but then one product, one SKU, that is able to start to do many, many different use cases and tasks. So, that’s one perspective I want you to be aware of.
Another I would say is like think about who out there is really, really good at robotics. I think like Boston Dynamics knows what they’re talking about with robotics. That’s going to be good competition.
It’s really important for us to have competition because if we didn’t, it would be -- we would worry about market validation. The fact that others are starting to see this and see the value of it is really pretty important.
Winnie Dong^ Thank you.
Jonathan Hurst^ Anybody else want to add? Or did we cover it?
Peggy Johnson^ Yes, I think you covered it. That commercialization lead is really --
Jonathan Hurst^ Yes.
Peggy Johnson^ -- takes us away from the pack because we’ve been out there in these facilities, having to figure out how to connect to the warehouse management systems, the IT infrastructure, meet all the compliance, all of their safety, all their regulatory bars that we’ve had to meet. It is not easy.
You come in and you have to become enterprise hearted. We have to become like what a conveyor belt is or a put wall or any of the other types of automation, let alone from just being a humanoid. It’s all those other things around it that have really given us quite a lead over the competitive humanoid set.
47
Anthony Rozmus^ Joe?
Joseph Spak^ Joe Spak from UBS. I wanted to maybe just spend a second on the RaaS model, which you described as enabling pretty rapid customer adoption. But there’s a -- it strikes me there’s a little bit of almost financial paradox, at least in the beginning, because like the faster you ramp, the bigger drain on your balance sheet.
Now, I know that flips over time, especially as you sort of get to the scaled numbers. But even at the scale numbers you provide, that’s $300 million out the door. And then I think you said you make that back in maybe half a year or so.
So, I guess, the question really is, is if we could think a little bit further out, how do you see this business model evolving from a financing perspective? Is the plan to always sort of self-fund this on the balance sheet? Do you envision some sort of either asset backed or debt capital markets type of initiatives? Just how you think about that as you scale.
Michael Beer^ Yes. And this is one of the reasons why we’re going public today, right? Access to capital markets, not necessarily using expensive equity dollars, but tapping into sub-10% debt capital market type resources. And obviously, holding the robots on balance sheet to service that RaaS offering, that’s something that we’ll definitely want to take into consideration.
The capital that we’re raising today is allowing us to really accelerate the initial build to satisfy this $300 million order. And obviously, there’s going to be some successive orders on the back of this. We wanted to articulate here today where we intend on being basically at the end of 2027 and those subsequent orders and how that’s going to sort of flow through the machine.
But one of the reasons we want to tap capital markets today is to really have a more advantaged cost of capital. And that’s what it’s going to come down to years from now. That’s what we’ll be talking about is how do we deploy resources with a really attractive cost of capital.
Anthony Rozmus^ Great. Thanks, team. That wraps up today’s Analyst and Investor Day.
We appreciate everybody coming out. And for those of you who join us online, we’ll be in touch soon. As many of you, we expect this transition to close later this year, subject to customary closing costs throughout the fall would be at investor conferences and again in the new year to present at CES.
So, thank you, all, for joining us today. Have a great rest of your day.
48
About Agility Robotics
Headquartered in Salem, Oregon, with offices in Pittsburgh, Pennsylvania and Fremont, California, Agility’s mission is to build robot partners that augment the human workforce. Agility’s groundbreaking general-purpose humanoid robot, Digit, is the first multi-purpose, human-centric robot that is Made for Work® and commercially deployed today. With more than 65,000 hours of real-world operation combined with industry-leading safety standards, we’re pioneering a new era of automation that enhances human potential. To learn more, visit agilityrobotics.com.
Agility, the Agility logo, Digit, Agility Arc, RoboFab, and Made for Work are trademarks of Agility Robotics, Inc. All rights reserved. Third-party trademarks are the property of their respective owners.
About Churchill Capital Corp XI (Nasdaq: CCXI)
Churchill is a blank check company formed for the purpose of effecting a merger, amalgamation, share exchange, asset acquisition, share purchase, reorganization or similar business combination with one or more businesses. It may pursue an initial business combination target in any business or industry.
Additional Information About the Proposed Transaction and Where to Find It
The proposed transaction will be submitted to shareholders of Churchill for their consideration. Churchill and Agility Robotics have jointly filed an initial registration statement on Form S-4 with the Securities and Exchange Commission (“SEC”) on September 9, 2026 (File No. 333-298781) (as amended from time to time, the “Registration Statement”), which includes preliminary and definitive proxy statements/prospectus to be distributed to Churchill’s shareholders in connection with Churchill’s solicitation of proxies for the vote by Churchill’s shareholders in connection with the proposed transaction and other matters described in the Registration Statement, as well as the prospectus relating to the offer of the securities to be issued to Agility stockholders in connection with the completion of the proposed transaction. After the Registration Statement has been declared effective, a definitive proxy statement/prospectus and other relevant documents will be mailed to Churchill shareholders as of the record date established for voting on the proposed transaction. Before making any voting or investment decision, Churchill and Agility stockholders and other interested persons are advised to read the preliminary proxy statement/prospectus and any amendments thereto and, once available, the definitive proxy statement/prospectus, as well as other documents filed with the SEC by Churchill in connection with the proposed transaction, as these documents will contain important information about Churchill, Agility and the proposed transaction. Shareholders may obtain a copy of the preliminary or definitive proxy statement/prospectus, as well as other documents filed by Churchill with the SEC, without charge, at the SEC’s website located at www.sec.gov or by directing a written request to Churchill Capital Corp XI, 640 Fifth Avenue, 14th Floor, New York, NY 10019.
49
Forward-Looking Statements
This communication includes “forward-looking statements” within the meaning of the federal securities laws. Forward-looking statements may be identified by the use of words such as “estimate,” “plan,” “project,” “forecast,” “intend,” “will,” “expect,” “anticipate,” “believe,” “seek,” “target,” “continue,” “could,” “may,” “might,” “possible,” “potential,” “predict,” “should,” “would” or similar expressions that predict or indicate future events or trends or that are not statements of historical matters, but the absence of these words does not mean that a statement is not forward-looking. We have based these forward-looking statements on current expectations and projections about future events. These statements include statements relating to, without limitation: hosting of the Analyst Day and its anticipated timing, location, program content and format, including the expected program highlights, presentations, live question-and-answer sessions, webcast, presentation slides and replay; Agility’s expectations regarding Digit 5, including its engineering for cooperatively safe work at scale, its product roadmap, software and autonomy roadmap and the role of Physical AI, and the use of real-world customer deployments and customer feedback to inform its development; Agility’s technology roadmap, commercial strategy, commercial momentum, market opportunity and plans to scale humanoid robotics across industrial applications, including plans for scaling deployments and manufacturing; Agility’s operating model, long-term financial profile and priorities as it scales; Agility’s plans to become a public company through the proposed business combination with Churchill; and the expected availability of the Analyst Day webcast, presentation slides and replay. These forward-looking statements are provided for illustrative purposes only and are not intended to serve as, and must not be relied on as, a guarantee, an assurance, a prediction or a definitive statement of fact or probability. Actual events and circumstances are difficult or impossible to predict and will differ from assumptions, many of which are beyond the control of Agility and Churchill.
These forward-looking statements are subject to known and unknown risks, uncertainties and assumptions that may cause Churchill’s actual results, levels of activity, performance or achievements to be materially different from any future results, levels of activity, performance or achievements expressed or implied by such statements. Such risks and uncertainties include: that Agility is pursuing an emerging technology, faces significant technical challenges and may not achieve commercialization or market acceptance; Agility’s historical net losses and limited operating history; Agility’s expectations regarding future financial performance, capital requirements and unit economics; Agility’s use and reporting of business and operational metrics; Agility’s competitive landscape; Agility’s dependence on members of its senior management and its ability to attract and retain qualified personnel; the potential need for additional future financing; the capital requirements of Agility’s business plans; Agility’s ability to manage growth and expand its operations; potential future acquisitions or investments in companies, products, services or technologies; Agility’s reliance on strategic partners and other third parties; Agility’s reliance on global supply chains and the risk that disruptions, tariffs, or trade restrictions could delay production, increase costs, and limit Agility’s ability to fulfill customer orders; Agility’s ability to maintain, protect and defend its intellectual property rights; risks associated with privacy, data protection or cybersecurity incidents and related regulations; risks associated with product liability, workplace safety regulations and potential injuries arising from the deployment of humanoid robots alongside human workers; the use, rate of adoption and regulation of artificial intelligence and machine learning; the evolving regulatory landscape for AI technologies across multiple jurisdictions and the risk that failure to comply with new or changing AI laws could result in enforcement actions, fines or restrictions on Agility’s ability to develop or deploy its products; uncertainty or changes with respect to laws and regulations; uncertainty or changes with respect to taxes, trade conditions and the macroeconomic environment; the combined company’s ability to maintain internal control over financial reporting and operate a public company; the risk that the proposed transaction may not be completed in a timely manner or at all, which may adversely affect the price of Churchill’s securities; the failure by the parties to satisfy the conditions to consummation of the proposed transaction, including the approval of Churchill’s shareholders; the possibility that required regulatory approvals for the proposed transaction are delayed or are not obtained, which could adversely affect the combined company or the expected benefits of the proposed transaction; the risk that shareholders of Churchill could elect to have their shares redeemed, leaving the combined company with insufficient cash to execute its business plans; the level of redemptions of Churchill’s public shareholders; the ability of Agility to grow and manage growth, maintain relationships with customers and retain its management and key employees; costs related to the proposed transaction; the occurrence of any event, change or other circumstance that could give rise to the termination of the business combination agreement; the outcome of any legal proceedings or government investigations that may be commenced against Agility or Churchill ; failure to realize the anticipated benefits of the proposed transaction; Agility’s estimates of expenses and profitability; the evolution of the markets in which Agility competes; the ability of Churchill or the combined company to issue equity or equity-linked securities in connection with the proposed transaction or in the future; and other factors described in Churchill’s filings with the SEC. Additional information concerning these and other factors that may impact such forward-looking statements can be found in filings and potential filings by Agility, Churchill or the combined company resulting from the proposed transaction with the SEC, including under the heading “Risk Factors.” If any of these risks materialize or assumptions prove incorrect, actual results could differ materially from the results implied by these forward-looking statements. In addition, these statements reflect the expectations, plans and forecasts of Agility’s and Churchill’s management as of the date of this communication; subsequent events and developments may cause their assessments to change. While Agility and Churchill may elect to update these forward-looking statements at some point in the future, they specifically disclaim any obligation to do so. Accordingly, undue reliance should not be placed upon these statements.
50
In addition, statements that “we believe” and similar statements reflect Churchill’s beliefs and opinions on the relevant subject. These statements are based upon information available to us as of the date of this communication, and while we believe such information forms a reasonable basis for such statements, such information may be limited or incomplete, and Churchill’s statements should not be read to indicate that we have conducted an exhaustive inquiry into, or review of, all potentially available relevant information. These statements are inherently uncertain and investors are cautioned not to unduly rely upon these statements.
An investment in Churchill is not an investment in any of Churchill’s founders’ or sponsors’ past investments, companies or affiliated funds. The historical results of those investments are not indicative of future performance of Churchill , which may differ materially from the performance of Churchill’s founders’ or sponsors’ past investments.
Participants in the Solicitation
Churchill , Agility and certain of their respective directors, executive officers and other members of management and employees may, under SEC rules, be deemed to be participants in the solicitation of proxies from Churchill’s shareholders in connection with the proposed transaction. Information regarding the persons who may, under SEC rules, be deemed participants in the solicitation of Churchill’s shareholders in connection with the proposed transaction will be set forth in the proxy statement/prospectus when it is filed by Churchill with the SEC. You can find more information about Churchill’s directors and executive officers in Churchill’s final prospectus related to its initial public offering filed with the SEC on December 16, 2025 and in the Annual Reports on Form 10-K filed by Churchill with the SEC. Additional information regarding the participants in the proxy solicitation and a description of their direct and indirect interests will be included in the proxy statement/prospectus when it becomes available. Shareholders, potential investors and other interested persons should read the proxy statement/prospectus carefully when it becomes available before making any voting or investment decisions. You may obtain free copies of these documents from the sources described above.
No Offer or Solicitation
This communication does not constitute an offer to sell or the solicitation of an offer to buy any securities, or a solicitation of any vote or approval, nor shall there be any sale of securities in any jurisdiction in which such offer, solicitation or sale would be unlawful prior to registration or qualification under the securities laws of any such jurisdiction. This communication is not, and under no circumstances is to be construed as, a prospectus, an advertisement or a public offering of the securities described herein in the United States or any other jurisdiction. No offer of securities shall be made except by means of a prospectus meeting the requirements of Section 10 of the Securities Act of 1933, as amended, or exemptions therefrom. INVESTMENT IN ANY SECURITIES DESCRIBED HEREIN HAS NOT BEEN APPROVED BY THE SEC OR ANY OTHER REGULATORY AUTHORITY NOR HAS ANY AUTHORITY PASSED UPON OR ENDORSED THE MERITS OF THE OFFERING OR THE ACCURACY OR ADEQUACY OF THE INFORMATION CONTAINED HEREIN. ANY REPRESENTATION TO THE CONTRARY IS A CRIMINAL OFFENSE.
51