CoreWeave Introduces Flexible Capacity Plans to Accelerate AI Innovation

Key Terms

flex reservations technical

Flex reservations are preliminary commitments investors make during a new securities offering when underwriters have the right to change the final price, size or allocation before the deal closes. Think of it like a flexible booking where the seller can adjust the number of seats and ticket price to make the sale work; for investors this matters because their order size or price may be scaled up or down, affecting allocation and potential returns.

spot instances technical

Spot instances are cloud computing virtual machines offered at substantially lower, variable prices because the cloud provider can reclaim them on short notice when capacity is needed; think of booking a deeply discounted hotel room that can be taken away if the hotel needs it. For investors, heavy use of spot instances can reduce operating costs and boost profit margins, but it also introduces reliability and predictability risks that can affect service uptime, revenue stability, and capital allocation decisions.

preemption signaling technical

Preemption signaling is when a company communicates that it will protect existing owners’ proportion of ownership by offering them first dibs on new shares or by taking steps to block rival bids, like putting up defensive measures. Investors care because such signals affect dilution, voting power and takeover chances—think of it like a homeowner announcing they’ll keep extra keys for current residents to prevent strangers from moving in.

inference technical

Inference is the process of drawing a conclusion from available evidence or data, like a detective piecing together clues to form a likely story. For investors it matters because these judgments turn raw reports, test results, or market signals into expectations about future performance, risk, or regulatory outcomes—so how someone infers from the same facts can change investment decisions and valuation.

mlperf technical

A standardized set of performance tests for machine learning systems that measures how fast and efficiently hardware and software can train and run AI models. Like a car mileage test for computers, MLPerf lets investors compare different vendors on speed, energy use and cost for AI workloads, helping predict which technologies will be more competitive, scalable and profitable as demand for AI grows.

New unified consumption framework aligns capacity with modern AI workload patterns

LIVINGSTON, N.J.--(BUSINESS WIRE)-- CoreWeave, Inc. (Nasdaq: CRWV), The Essential Cloud for AI™, today announced CoreWeave Flexible Capacity Plans, including Flex Reservations and Spot, designed to match the dynamic reality of modern AI workloads. With Flex Reservations and Spot, CoreWeave enables customers to move beyond the binary choice of reserved capacity versus on-demand capacity.

CoreWeave’s new unified consumption framework allows customers to align capacity needs to demand patterns.

While AI training cycles are often predictable, production-level inference is not. Traffic spikes and use fluctuates, forcing teams to either over-provision or risk delays. CoreWeave Flexible Capacity Plans give teams cost-efficient options for interruptible work and more control over how they run AI, enabling innovation without limitation.

“At production scale, infrastructure planning becomes as critical as deployment,” said Chen Goldberg, EVP of Product & Engineering at CoreWeave. “CoreWeave is setting the standard for the AI cloud by providing guaranteed capacity when it counts and flexible pricing when demand shifts. We’re bringing the original promise of the cloud - scale and efficiency - back to the AI pioneers pushing the limits of innovation.”

A Unified Capacity Framework for AI
CoreWeave Flexible Capacity Plans build on CoreWeave’s existing Reservations and On-Demand offerings, extending that foundation to reflect how modern AI workloads actually behave.

With the addition of Flex Reservations and Spot, CoreWeave introduces a unified capacity framework that redefines how production AI runs at scale:

• Reservations: Predictable, always-on capacity for steady workloads.
• Flex Reservations: A first-of-its kind model for guaranteed peak capacity with flexible economics for workloads that ramp or scale unevenly. Customers secure a capacity ceiling with a lower 24/7 holding fee, paying full usage rates only when instances are active.
• Spot: A new lower-cost option for interruption-tolerant work like batch analytics or backfills. Spot is delivered with explicit preemption signaling, allowing engineers to checkpoint and recover work cleanly.
• On Demand: Best-effort access for immediate, incremental capacity.

This framework allows customers to reserve what’s steady, protect what must be guaranteed, and shift interruptible work — aligning cost and certainty with real demand patterns. Flexibility of this kind isn’t theoretical, it’s already shaping how customers design and scale their AI workflows.

“At inference.net, our mission is to help teams get the most from AI models,” said Ibrahim Ahmed, CTO at inference.net. “ The most powerful AI models are not generic, off-the-shelf models, they’re specialized models trained on your data, for your problem. We built custom scheduling and orchestration software so that any team can train and deploy frontier-quality specialized LLMs on underutilized GPU capacity. Spot instances from CoreWeave are key to making this possible, giving our customers access to the compute they need to bring custom models to production at a fraction of the cost.”

Flex Reservations are available to Preview through CoreWeave account teams in eligible regions and SKUs, and Spot is generally available now.

CoreWeave’s AI cloud delivers industry-leading performance and efficiency through an end-to-end technology stack optimized for modern AI workloads. CoreWeave’s technology team consistently sets new standards for performance, demonstrated by an industry-leading MLPerf benchmark for AI workloads and its position as the only AI cloud to earn the top Platinum ranking in both SemiAnalysis ClusterMAX™ 1.0 and 2.0, which evaluate AI cloud performance, efficiency and reliability.

About CoreWeave
CoreWeave is The Essential Cloud for AI™. Built for pioneers by pioneers, CoreWeave delivers a platform of technology, tools, and teams that enables innovators to move at the pace of innovation, building and scaling AI with confidence. Established in 2017, CoreWeave completed its public listing on Nasdaq (CRWV) in March 2025. Learn more at www.coreweave.com.

View source version on businesswire.com: https://www.businesswire.com/news/home/20260310298766/en/

press@coreweave.com

Source: CoreWeave, Inc.