How CoreWeave became NVIDIA’s top GPU cloud partner in 2025: A deep dive into AI infrastructure’s rising star

What is CoreWeave and why is it at the center of AI infrastructure in 2025?

CoreWeave, Inc. (NASDAQ: CRWV) has undergone a dramatic transformation from its early days as a cryptocurrency mining operator to become one of the fastest-growing cloud infrastructure providers in the world. This pivot, which began in earnest in 2020, was fully realized by 2023 when the company divested entirely from crypto and repositioned itself to serve the burgeoning demand for GPU-accelerated computing. Today, CoreWeave is recognized as a “neocloud” champion, building AI-specific compute infrastructure for customers requiring scalable access to the most advanced GPU hardware.

With clients ranging from OpenAI and Mistral to xAI and Cohere, CoreWeave is now deeply embedded in the enterprise and foundational model development ecosystem. Its infrastructure—built around NVIDIA GPUs and optimized for AI training, inferencing, and high-performance workloads—has made it a favorite among both startups and Fortune 500 companies seeking cloud-native alternatives to AWS and Azure.

How Dell Technologies enabled CoreWeave’s Blackwell cluster deployment

In 2025, CoreWeave became one of the first cloud providers to deploy Dell Technologies’ (NYSE: DELL) next-generation PowerEdge servers optimized for Blackwell GPUs. Dell’s IR7000 series, including the XE9680L and newly launched XE9712, are fully rack-integrated, liquid-cooled AI servers capable of housing NVIDIA GB200 NVL GPUs. According to Dell’s Q1 FY26 earnings call, CoreWeave is now the world’s largest non-hyperscaler buyer of these AI systems.

This partnership has enabled CoreWeave to scale Blackwell-based clusters faster than any other cloud provider, deploying new infrastructure in under six weeks from order to operation. Dell’s vertically integrated supply chain and direct-to-data-center logistics have been key differentiators. For Dell, CoreWeave serves as a marquee customer validating its pivot toward multicloud AI infrastructure dominance.

By June 2025, CoreWeave had placed multiple rounds of large-volume orders with Dell, confirming that future clusters in both North America and Europe would be built entirely on Dell’s Blackwell-optimized platforms. Industry observers view this as a direct challenge to Hewlett Packard Enterprise and Super Micro Computer in the fast-evolving AI server market.

What makes CoreWeave’s funding model unique in AI cloud?

To finance its rapid expansion, CoreWeave has combined strategic equity raises with debt financing backed by long-term compute contracts. In 2024, it secured $2.3 billion in equity funding from BlackRock, Coatue, and Magnetar. In 2025, it followed with a $1.5 billion debt facility from JPMorgan Chase, Morgan Stanley, and Citigroup, with another $1.5 billion round expected to close later this year.

A crucial driver of investor confidence is CoreWeave’s contract with OpenAI, which reportedly includes $11.9 billion in locked-in compute consumption through 2027. This forward revenue visibility provides confidence to lenders and underpins the company’s ability to scale infrastructure with aggressive capex cycles.

Additionally, NVIDIA’s strategic equity stake in CoreWeave signals strong alignment between hardware supplier and infrastructure operator. By distributing GPUs preferentially to CoreWeave, NVIDIA ensures a competitive counterweight to hyperscalers while driving ecosystem diversity for its silicon.

How CoreWeave is reshaping the hyperscaler relationship with OpenAI and Google

On June 11, 2025, Reuters reported that CoreWeave had signed an agreement to supply GPU resources to OpenAI through Google Cloud. This marks a rare three-way collaboration between infrastructure competitors, suggesting that OpenAI is moving to a multicloud deployment model to increase flexibility and performance resilience.

The partnership allows OpenAI to run workloads across CoreWeave’s Blackwell clusters even when accessed via Google Cloud’s frontend. This breaks from the long-standing exclusivity with Microsoft Azure and highlights CoreWeave’s value proposition as a GPU-dense, neutral cloud partner capable of handling high-throughput inferencing at scale.

This development also signals growing interoperability in the AI infrastructure stack. Enterprises increasingly demand multi-cloud portability and vendor redundancy to avoid lock-in. CoreWeave’s integration with both sovereign data frameworks and open developer tooling has made it the go-to partner for such deployments.

What is the scale of CoreWeave’s current and future operations?

As of mid-2025, CoreWeave operates data centers across North America and Europe, with new sites under development in Germany, the United Kingdom, and South Korea. In Sweden, CoreWeave has partnered with EcoDataCenter to build GDPR-compliant Blackwell GPU clusters for European enterprise clients. This expansion reflects growing global demand for sovereign cloud infrastructure aligned with regional regulatory standards.

Internally, CoreWeave claims to manage over 400,000 NVIDIA GPUs, a figure expected to rise to more than 750,000 by Q4 2025. The majority of this growth will be driven by NVIDIA’s GB200 NVL and GH200 NVL2 platforms, housed in Dell XE9712 systems. This positions CoreWeave as the largest AI-native cloud provider by GPU volume outside of the Big Three hyperscalers.

How are enterprises and developers using CoreWeave’s infrastructure?

CoreWeave’s infrastructure is optimized for developers building and deploying machine learning models, digital twins, and simulation-heavy applications. Its developer platform, CoreWeave Kubernetes Service (CKS), allows containerized AI workloads to be deployed in minutes. GPU Mission Control, the company’s real-time orchestration layer, offers customers granular access to usage, performance metrics, and provisioning tools.

Enterprise customers are leveraging CoreWeave for model fine-tuning, batch inference, real-time generative chatbots, and simulation-based workflows in sectors such as automotive, finance, life sciences, and energy. The company’s acquisition of Weights & Biases in May 2025 further deepens its value proposition to AI/ML engineers by integrating model lifecycle tools directly into its cloud stack.

Average contract lengths are over 24 months, and over 60% of CoreWeave’s revenue is derived from long-term subscription agreements. This revenue model is designed to withstand volatility in spot GPU pricing and supports high utilization of capex-intensive clusters.

How is the market reacting to CoreWeave’s breakout performance?

CoreWeave’s IPO in March 2025 priced at $38 per share. By June, shares were trading above $165—a staggering 340% increase in less than three months. Institutional investors including Fidelity, Tiger Global, and Baillie Gifford have disclosed significant positions. Retail interest spiked following reports of the Google-OpenAI-CoreWeave infrastructure triangle and early Blackwell deployments.

Equity analysts remain split. Bullish forecasts highlight CoreWeave’s Blackwell-first positioning, rapid scale-up, and strong forward contracts. Bearish positions cite high debt leverage, dependence on NVIDIA, and the risk of a demand plateau in late 2026. As of this writing, analyst coverage reflects 45% Buy, 35% Hold, and 20% Sell recommendations.

Strategic outlook: CoreWeave’s model is now the reference architecture for AI-native clouds

CoreWeave’s rise from a crypto infrastructure footnote to a cornerstone of the global AI compute ecosystem is one of the most dramatic pivots in recent tech history. With Dell Technologies as its principal server partner and NVIDIA Blackwell at the heart of its clusters, CoreWeave is now the reference design for what the next generation of AI-native cloud providers might look like.

Its developer-first APIs, sovereign cloud compatibility, and hyperscaler interconnects make it highly relevant to modern enterprise buyers seeking AI performance at scale without full lock-in. The next stage of its evolution could include ASIC design, sovereign GPU supply chains, and regional IPOs, all of which could further differentiate CoreWeave in a rapidly crowding market.

For now, however, CoreWeave’s position appears unassailable among GPU-native cloud providers—and its infrastructure strategy is being closely studied by startups and legacy cloud providers alike.