CRWV

Unified AI Developer Platform & Reinforcement Learning

Cluster Description

Commitments to deliver an integrated, end-to-end developer stack spanning experimentation, RL/agent tooling, evaluation, and deployment, with strong open-source stewardship. Recent acquisitions and launches (W&B completion, OpenPipe, ServerlessRL, Marimo) expand and in some cases supersede earlier plans.

Commitments

• November 06, 2025: Continue investing in the technology layer that powers AI innovation to accelerate developer productivity and model performance.
• November 06, 2025: Support agent and model development via integrated products and partnerships across Weights & Biases, OpenPipe, and Marimo.
• October 30, 2025: Integrate Marimo’s technology into CoreWeave Cloud to unify the developer workflow across training, inference, data movement, and iteration.
• October 30, 2025: Augment Weights & Biases with Marimo to provide end-to-end tools for building, evaluating, and deploying AI applications.
• October 30, 2025: Maintain and accelerate development of the marimo open-source project.
• October 30, 2025: Keep the marimo notebook freely available under a permissive license with a transparent, community-guided roadmap.
• October 08, 2025: Publicly launch a fully managed Serverless Reinforcement Learning capability that scales to dozens of GPUs, delivering nearly 1.4x faster training and 40% lower costs versus local H100 setups.
• October 08, 2025: Integrate infrastructure, RL frameworks, and developer tools so enterprises can fine-tune LLMs and build reliable AI agents.
• September 03, 2025: Acquire OpenPipe and integrate its self-learning tools with CoreWeave’s AI cloud to expand the platform for scalable intelligent systems.
• September 03, 2025: Democratize reinforcement learning by enabling developers to post-train models on their unique workflows.
• June 18, 2025: Launch new AI cloud software products in preview to boost productivity for AI engineers and ML researchers.
• June 18, 2025: Integrate Mission Control into the W&B Models platform to simplify diagnostics by correlating infrastructure events with training runs.
• June 18, 2025: Provide W&B Inference for single-interface access to leading open-source models and introduce Weave Online Evaluations for real-time agent performance monitoring.
• May 05, 2025: Complete the acquisition of Weights & Biases to accelerate AI innovation and integrate hardware and software layers, superseding the March 04, 2025 expected-close announcement.
• May 05, 2025: Maintain W&B platform interoperability and accelerate investment to unlock new features for developers.

Federal Market Entry & Compliance-Grade AI Cloud

Cluster Description

Commitments to enter and scale in the U.S. federal market with secure, compliant AI cloud services, backed by performance leadership and targeted investments; multiple items announced together on the same date.

Commitments

• October 28, 2025: Enter the U.S. federal market via CoreWeave Federal to deliver secure, compliant, high-performance AI cloud services to agencies and the Defense Industrial Base.
• October 28, 2025: Pursue FedRAMP and other authorizations to meet evolving government cybersecurity and compliance requirements.
• October 28, 2025: Invest in teams, tools, and technologies tailored to federal mission needs.
• October 28, 2025: Expand partnerships with government agencies and systems integrators to deliver mission outcomes.
• October 28, 2025: Leverage advanced deployments and public sector experience (including W&B capabilities) to meet compliance standards and mission demands.
• October 28, 2025: Strengthen U.S. infrastructure and presence (including Washington, DC) and bolster government/security/legal/communications leadership.
• October 28, 2025: Continue advancing AI performance standards with deep integration of NVIDIA accelerated computing, networking, and orchestration technologies (supported by industry-leading MLPerf results).
• October 28, 2025: Introduce new solutions (e.g., AI Object Storage, ServerlessRL) and pursue strategic acquisitions (OpenPipe and intent to acquire Monolith AI) to expand the federal-ready portfolio.

Strategic Compute Partnerships with AI Leaders

Cluster Description

Anchor partnerships that commit compute, infrastructure, and services to leading AI organizations for training, inference, and AGI/agent development; newer agreements expand or supersede earlier ones.

Commitments

• October 15, 2025: Provide Poolside with a state-of-the-art cluster of NVIDIA GB300 NVL72 systems totaling more than 40,000 GPUs.
• October 15, 2025: Provide cloud solutions for Poolside’s Project Horizon and serve as anchor tenant and operational partner for the first 250 MW phase, with an option to expand by 500 MW.
• October 15, 2025: Partner with Poolside to deliver advanced AI cloud services supporting AGI development and enterprise agent deployment.
• September 25, 2025: Expand the compute agreement with OpenAI by up to $6.5 billion to power next-generation model training, updating and extending the March 10, 2025 agreement (total value now approximately $22.4 billion).
• May 22, 2025: Establish a multi-year partnership as Official AI Cloud Computing Partner to Aston Martin Aramco to accelerate engineering by migrating on-premises workloads to the cloud, with brand presence on the AMR25 car and wind tunnel.
• February 04, 2025: Deliver one of the first NVIDIA GB200 Grace Blackwell Superchip-enabled AI supercomputers to IBM to train its next-generation Granite models.

Global Infrastructure Expansion & Sustainability

Cluster Description

Commitments to expand data center capacity and sovereign AI infrastructure across the U.S. and UK, emphasizing renewable power, efficient cooling, and job creation; some items clarify or complement earlier announcements.

Commitments

• September 25, 2025: Commit $1.5 billion to power AI innovation and growth in the UK, complementing the September 16, 2025 UK capacity plan.
• September 16, 2025: Invest an additional £1.5 billion in UK AI data centre capacity and operations (bringing total UK commitment to £2.5 billion) with a priority on sustainability.
• September 16, 2025: Partner with NVIDIA and DataVita to deploy Grace Blackwell Ultra GPUs in Scotland using 100% renewable energy and closed-loop cooling.
• September 16, 2025: Plan an additional sovereign AI infrastructure deployment using NVIDIA GB300 and RTX PRO Blackwell Server Edition GPUs.
• September 16, 2025: Plan with local U.S. government to allocate up to $6 billion to equip a state-of-the-art data centre in Lancaster, Pennsylvania, refining the July 15, 2025 funding approach.
• July 15, 2025: Commit more than $6 billion to equip a new AI data center in Lancaster, PA, with an initial 100 MW build and potential expansion to 300 MW for advanced AI workloads.
• July 15, 2025: Create approximately 600 construction jobs and about 70 full-time roles at launch (scaling to around 175 over time) through the Lancaster project.
• July 15, 2025: Serve as tenant and co-developer with local partners for the Lancaster site, expand the AI data center network to 33 facilities (28 in the U.S.), and coordinate workforce/permitting with Pennsylvania leadership.
• January 13, 2025: Confirm two UK data centers (Crawley and London Docklands) are operational from a £1 billion investment and powered entirely by renewable energy.
• January 13, 2025: Plan to open 10 new data centers globally in 2025, targeting a total of 38 data centers by year-end.

Hardware Leadership & Performance Standards

Cluster Description

First-to-market hardware access, deep NVIDIA integrations, and record benchmarks that set performance standards for training and inference; later results and deployments reinforce and extend earlier milestones.

Commitments

• July 09, 2025: Be first to market with NVIDIA RTX PRO 6000 Blackwell GPU instances at scale, targeting up to 5.6x faster LLM inference and 3.5x faster text-to-video generation.
• July 03, 2025: Be the first AI cloud provider to deploy NVIDIA GB300 NVL72 systems and scale deployments globally, aiming for a 10x responsiveness boost and 50x output increase for reasoning model inference.
• July 03, 2025: Integrate GB300 NVL72 with CoreWeave’s cloud-native software stack (CKS and Slurm on Kubernetes) to improve operational efficiency.
• July 03, 2025: Collaborate with NVIDIA and IBM to achieve record MLPerf Training v5.0 results using nearly 2,500 GB200 Grace Blackwell Superchips to train Llama 3.1 405B in 27.3 minutes.
• April 15, 2025: Provide access to NVIDIA GB200 NVL72 rack-scale systems and a full stack of cloud services to advance AI model development and deployment.
• April 02, 2025: Deliver 800 tokens per second on Llama 3.1 405B and 33,000 TPS on Llama 2 70B (a 40% improvement versus prior H100 instances) to set new inference performance benchmarks.
• February 04, 2025: Launch NVIDIA GB200 NVL72 instances with up to 30x faster real-time LLM inference, up to 25x lower TCO and energy for real-time inference, and up to 4x faster LLM training.

AI Data Services: Object Storage and Managed Capabilities

Cluster Description

Data-layer commitments purpose-built for AI, eliminating hidden costs and improving throughput and integrity, supported by continued expansion of managed services.

Commitments

• October 16, 2025: Launch CoreWeave AI Object Storage, a fully managed object storage service purpose-built for AI workloads.
• October 16, 2025: Make a single dataset instantly accessible globally without egress or request/transaction fees.
• October 16, 2025: Maintain high throughput and data integrity across distributed GPU nodes and regions via LOTA, private interconnects, direct peering, and 400 GBps-capable ports.
• October 16, 2025: Introduce three automatic usage-based pricing tiers that reduce typical AI storage costs by more than 75% and eliminate egress, request, and tiering fees.
• October 16, 2025: Continue expanding the software ecosystem and managed services (e.g., ServerlessRL) to support AI development, testing, and deployment at scale.

Ecosystem Investment via CoreWeave Ventures

Cluster Description

Commitments to fund and accelerate the AI ecosystem with capital, compute-for-equity, technical collaboration, and GTM support; later announcements reaffirm and supersede initial launch details.

Commitments

• November 06, 2025: Fuel startups with capital and compute through CoreWeave Ventures to support AI ecosystem growth.
• October 16, 2025: Provide founders resources including direct capital investments, compute-for-equity transactions, and technical collaboration via CoreWeave Ventures.
• September 25, 2025: Launch CoreWeave Ventures to back founders building AI platforms and technologies, superseding the September 09, 2025 initial announcement.
• September 16, 2025: Support independent founders and startups through CoreWeave Ventures with capital and compute to accelerate the next frontier of computing.
• September 09, 2025: Offer a variety of investment models (including compute-for-equity) and accelerated access to CoreWeave’s AI cloud to help companies scale.
• September 09, 2025: Provide production-grade testing environments, product and GTM insights, and opportunities for deep technical partnerships to fast-track real-world AI use cases.

Organizational Scale-Up & Governance

Cluster Description

Commitments to scale go-to-market and governance through leadership appointments, capital markets actions, and strategic M&A to improve cost structure and financing flexibility.

Commitments

• October 16, 2025: Appoint Jon Jones as the first Chief Revenue Officer to lead the global revenue organization and scale products and sales.
• October 16, 2025: Extend CoreWeave’s leadership as the platform of choice for AI innovation and adoption by deepening vertical integration and continuing growth organically and via acquisitions.
• July 07, 2025: Acquire Core Scientific in an all-stock transaction to enhance data center infrastructure and operational efficiency, targeting $500 million annual run-rate cost savings by the end of 2027.
• July 07, 2025: Eliminate over $10 billion in cumulative future lease overhead over the next 12 years as part of the Core Scientific acquisition to streamline operations and reduce cost of capital.
• July 07, 2025: Pursue infrastructure financing strategies post-acquisition to fund committed capex and improve financing flexibility.
• March 27, 2025: Price the IPO at $40.00 per share for 37,500,000 shares (expected to close March 31, 2025), following the March 03, 2025 registration filing.
• March 24, 2025: Appoint a Chief Marketing Officer to enhance market presence and drive sustainable sales and product growth.
• January 09, 2025: Appoint a Chief Information Officer to lead IT and digital transformation to support hypergrowth and scaling needs.
• January 09, 2025: Appoint an independent director to strengthen governance and financial oversight as the company scales AI managed cloud solutions.

Industrial ML Expansion (Monolith AI)

Cluster Description

Commitments to expand CoreWeave’s platform into industrial and manufacturing use cases through the Monolith AI acquisition effort and related offerings; later references reinforce the strategy.

Commitments

• November 06, 2025: Broaden ML application into physics and engineering domains through the Monolith AI acquisition effort to extend CoreWeave’s technology footprint.
• October 06, 2025: Announce an agreement to acquire Monolith AI to offer a full-stack AI platform for industrial and manufacturing enterprises.
• October 06, 2025: Enable enterprises to shorten R&D cycles and accelerate product development by combining Monolith’s simulation/test-driven ML with CoreWeave’s AI cloud.
• October 06, 2025: Provide engineering ML tools (e.g., anomaly detection, test plan optimization, next-test recommendation) to reduce physical testing and improve quality.
• October 06, 2025: Scale Monolith’s AI-for-engineers mission to reach thousands more builders across industries.
• October 06, 2025: Serve as the Official AI Cloud Computing Partner to support Aston Martin Aramco’s first large-scale cloud computing facility in the automotive sector.

Potentially Out-of-Scope (Attribution Unclear)

Cluster Description

Commitments appear unrelated to CoreWeave’s AI cloud business and may pertain to a different company; included here for completeness and review.

Commitments

• December 18, 2024: Develop an integrated commercial process to recycle captured CO2 with low-carbon hydrogen to produce sustainable ethanol and ethylene under Project SECURE.
• December 18, 2024: Conduct Phase 1 Front-End Engineering Design, develop project plans, and complete documentation for NEPA review.
• December 18, 2024: Create high-quality jobs and support local communities throughout the project’s phases.