IREN Doubles AI Cloud to 23k GPUs, Raises AI Cloud ARR Target to >$500m
Rhea-AI Summary
IREN (NASDAQ: IREN) has announced a significant expansion of its AI Cloud capacity, doubling its GPU fleet to 23,000 GPUs through a $674 million procurement of additional 12.4k GPUs. The new order includes 7.1k NVIDIA B300s, 4.2k NVIDIA B200s, and 1.1k AMD MI350Xs.
The company has raised its AI Cloud annualized run-rate revenue (ARR) target to more than $500 million by Q1 2026. IREN's expanded infrastructure includes various GPU models across its British Columbia campuses, which can support over 60k Blackwell GPUs. The company notes increasing customer demand, with many seeking to contract capacity ahead of delivery.
Positive
- Doubled AI Cloud capacity to 23k GPUs with $674M new GPU procurement
- Raised AI Cloud ARR target to >$500M by Q1 2026
- Expanded hardware diversity by adding AMD GPUs alongside NVIDIA
- British Columbia campuses capable of supporting >60k Blackwell GPUs
- Strong customer demand with pre-contracting trends
Negative
- Significant capital expenditure of $674M for GPU procurement
- Potential impact on Bitcoin mining capacity requiring ASIC redeployment
- ARR target of $500M is not fully contracted and subject to execution risks
Insights
IREN doubles GPU capacity to 23k, investing $674M in Blackwell/AMD chips to target $500M+ AI Cloud ARR by Q1 2026.
IREN's announcement of doubling its AI Cloud capacity to 23,000 GPUs represents a significant scaling operation in the high-demand AI infrastructure space. The
The most telling aspect of this expansion is the shift toward pre-contracted capacity. Customers are now securing compute resources before the hardware is even installed—a strong indicator of the severe supply-demand imbalance in AI compute. This matches patterns we've seen with cloud hyperscalers where future capacity is being reserved years in advance.
From an infrastructure perspective, IREN's vertically integrated approach provides competitive advantages. Their Prince George campus leverages the company's renewable energy assets (supported by RECs) and their existing power contracts totaling 2,910MW across their portfolio. This power-first strategy has enabled them to scale AI infrastructure more rapidly than many competitors who face power acquisition bottlenecks.
The addition of AMD hardware alongside NVIDIA GPUs is particularly strategic—it diversifies supply chain dependencies while expanding their addressable market to customers with workloads optimized for AMD's architecture. This multi-vendor approach provides flexibility as availability between chip manufacturers fluctuates.
IREN's
The capital allocation shift toward AI infrastructure represents a significant diversification from IREN's original Bitcoin mining focus. This
The target of
This capacity expansion requires substantial capital. While not explicitly detailed, the release mentions "financing workstreams are underway" to support this growth. This likely indicates additional debt or equity raises will be necessary beyond operating cash flows. The
The business model transformation is notable—IREN is leveraging their existing power assets (2,910MW portfolio) and land holdings (>2,000 acres) originally secured for Bitcoin mining to support higher-margin AI compute services. This power-first approach differentiates them from pure-play cloud providers who often struggle with power acquisition.
Their multi-vertical strategy now spans three distinct but complementary business lines: Bitcoin mining (50 EH/s capacity), AI Cloud Services (GPU rental), and AI Data Centers (infrastructure development). This diversification potentially reduces cyclical exposure to Bitcoin price volatility by establishing revenue streams tied to the broader AI compute market, where demand currently exceeds supply significantly.
NEW YORK, Sept. 22, 2025 (GLOBE NEWSWIRE) -- IREN Limited (NASDAQ: IREN) (“IREN”) today announced it has doubled its AI Cloud capacity to 23k GPUs following the procurement of an additional 12.4k GPUs, and is now targeting more than
To meet rising demand, IREN has purchased 7.1k NVIDIA B300s, 4.2k NVIDIA B200s and 1.1k AMD MI350Xs for approximately
Deliveries will be staged over the coming months at IREN’s Prince George campus and are expected to support the >
- 1.9k NVIDIA H100s & H200s
- 19.1k NVIDIA B200s & B300s
- 1.2k NVIDIA GB300s
- 1.1k AMD MI350Xs
The addition of AMD hardware alongside further NVIDIA GPUs broadens IREN’s offering and addressable market.
AI Cloud demand continues to outpace global supply, with a growing appetite among customers to contract capacity ahead of commissioning.
IREN is well positioned to capture this demand, with its British Columbia campuses capable of supporting more than 60k Blackwell GPUs3, in addition to its Texas data centers. Financing workstreams are underway to support this growth, with any impact to IREN’s Bitcoin mining capacity expected to be mitigated by redeploying ASICs to other sites.
“As global demand for compute accelerates, customers are increasingly seeking partners who can deliver scale quickly,” said Daniel Roberts, Co-Founder & Co-CEO of IREN.
“Doubling our fleet to more than 23,000 GPUs in just a few months highlights the strength of our vertically integrated platform and our ability to meet urgent, long-term demand. The shift toward pre-contracting ahead of delivery provides further momentum for the growth of our AI Cloud business.”
About IREN
IREN is a leading developer, owner and operator of next-generation data centers powering the future of Bitcoin, AI and beyond utilizing
- Power & Land Portfolio: 2,910MW of grid-connected power secured across >2,000 acres in the U.S. and Canada, with an additional multi-gigawatt development pipeline.
- Next-Generation Data Centers: 810MW of operating data centers underpinning three verticals: Bitcoin Mining, AI Cloud Services and AI Data Centers.
- Bitcoin Mining: one of the world’s largest and lowest-cost Bitcoin producers with 50 EH/s of installed self-mining capacity.
- AI Cloud Services: delivering high performance cloud compute to AI customers with next-generation NVIDIA and AMD GPUs.
- AI Data Centers: end-to-end design, construction and operation of data center infrastructure tailored for AI workloads.
Contacts
| Media Megan Boles Aircover Communications +1 562 537 7131 megan.boles@aircoverpr.com Jon Snowball Sodali & Co +61 477 946 068 +61 423 136 761 | Investors Mike Power IREN mike.power@iren.com |
Assumptions and Notes
- AI Cloud ARR is calculated as GPU/hour pricing for contracted GPUs multiplied by 8,760 hours per year. ARR of >
$500m is presented as an illustrative run-rate measure of potential revenue based on a ~23k GPU deployment and internal company assumptions regarding utilization and GPU pricing. It is not fully contracted, there can be no assurance that it will be achieved, and actual revenue may differ materially. Assumes on time delivery and commissioning of GPUs. - Includes CPUs, storage, network fees and onsite services.
- >60k Blackwell GPUs reflects internal estimate of capacity based on 160MW power capacity across British Columbia sites, 1.1 power usage effectiveness and NVIDIA B200 reference architecture.
Forward-Looking Statements
This news release contains “forward-looking statements” within the meaning of Section 27A of the Securities Act of 1933, as amended, and Section 21E of the Securities Exchange Act of 1934, as amended, that involve substantial risks and uncertainties. Forward-looking statements generally relate to future events or IREN’s future financial or operating performance. Forward-looking statements include information concerning possible or assumed future results of operations, including descriptions of our business plan and strategies, revenue targets and trends we expect to affect our business. These statements often include words such as “anticipate,” “believe,” “may,” “can,” “should,” “could,” “might,” “plan,” “possible,” “project,” “strive,” “budget,” “forecast,” “expect,” “intend,” “target”, “will,” “estimate,” “predict,” “potential,” “continue,” “scheduled”. Forward-looking statements may also be made, verbally or in writing, by members of our Board or management team in connection with this news release.
These forward-looking statements are based on management’s current expectations and beliefs. These statements are neither promises nor guarantees, but involve and are subject to known and unknown risks, uncertainties and other important factors that may cause IREN’s actual results, performance or achievements to differ materially from any future results performance or achievements expressed or implied by the forward-looking statements, including IREN’s ability to successfully execute on its growth strategies and operating plans, achieve its targeted annualized AI Cloud revenue, continue to develop its existing data center sites, design and deploy direct-to-chip liquid cooling systems, and diversify and expand into the market for high performance computing solutions (including the market for cloud services and potential colocation services, along with other important factors discussed under the caption “Risk Factors” in IREN’s Annual Report on Form 10-K, filed with Securities and Exchange Commission (the “SEC”) on August 28, 2025 and our other filings with the SEC. These and other important factors could cause actual results to differ materially from those indicated by the forward-looking statements made in this press release. Any forward-looking statement included in this press release speaks only as of the date of such statement. Except as required by law, IREN disclaims any obligation to update or revise, or to publicly announce any update or revision to, any of the forward-looking statements, whether as a result of new information, future events or otherwise.