CoreWeave

CoreWeave, Inc. is an American artificial intelligence cloud computing company headquartered in Livingston, New Jersey, that rents large clusters of NVIDIA GPUs to AI developers and enterprises and builds the software to manage them. Founded in 2017 as a cryptocurrency mining operation, CoreWeave began moving into GPU cloud services in 2019 and launched its cloud platform in 2020.^[7][35] Many investors treat it as the bellwether of the "neocloud" category, a cohort of GPU-dense providers built to absorb AI training and inference workloads that hyperscalers cannot fully service.^[40] The company went public on the Nasdaq stock exchange on March 28, 2025, under the ticker symbol CRWV, raising approximately $1.5 billion in the largest U.S. technology IPO since 2021.^[1][29] In the second quarter of 2026, revenue rose 112% year-over-year to $2.58 billion, and the company reported a revenue backlog of about $104 billion at June 30, 2026, not counting more than $25 billion of new customer commitments signed early in the third quarter.^[36][38] CoreWeave says nine of the ten leading AI model providers use its platform.^[31] The buildout has been financed largely with debt, about $35 billion by mid-2026, and in October 2026 the company's shares and bonds fell sharply as investors questioned that borrowing.^[36][39]

History

Founding and early days (2017-2018)

CoreWeave was founded in 2017 by Michael Intrator, Brian Venturo, and Brannin McBee; the company also lists Peter Salanki, its chief technology officer, as a co-founder.^[7][66] All three original founders came from commodities and energy trading. Intrator and Venturo had worked together at Natsource Asset Management, and in 2013 they helped start Hudson Ridge Asset Management, a natural gas hedge fund that Intrator ran as chief executive and where Venturo was a partner and portfolio manager.^[3][7][65] The founders had built a machine learning model to help them pick investments in the data-heavy energy industry, and McBee, previously a proprietary trader at the commodity hedge fund Windy Bay Power, ran the data firm they used.^[3][65]

The company was formed in September 2017 as The Atlantic Crypto Corporation LLC, became Atlantic Crypto Corporation in 2018, and operated as an Ethereum mining company using NVIDIA GPUs.^[7][34] Its first GPU sat on a pool table in the founders' Wall Street office; the founders later turned Venturo's grandfather's garage in New Jersey into a data center.^[7] At the time, Bitcoin and other cryptocurrencies were rising sharply in price.^[7] Venturo later said the company had been "the largest Ethereum miner in the world for like two and a half years" and at one point ran 50,000 NVIDIA consumer GPUs.^[3]

Pivot to GPU cloud computing (2019)

Following the cryptocurrency market crash of 2018, Ethereum mining became significantly less profitable. During the 2018-2019 "crypto winter" the company bought distressed mining hardware, and Intrator wrote in 2021 that its GPU fleet had grown from a single card "into hundreds, then tens of thousands" across seven facilities.^[7] In 2019 the founders began using that inventory to sell GPU computing to other businesses, and in December 2019 the company changed its name to CoreWeave, Inc.^[7][34] It launched the CoreWeave Cloud Platform in 2020 and shut down its remaining Ethereum mining unit in September 2022.^[7][35]

The company initially targeted clients in visual effects (VFX) rendering, scientific computing, and other GPU-intensive workloads. CoreWeave differentiated itself from traditional hyperscale cloud providers like Amazon Web Services, Microsoft Azure, and Google Cloud by offering GPU-specialized infrastructure at lower costs and with faster provisioning times.^[7]

Rapid growth and the AI boom (2020-2023)

After OpenAI released ChatGPT in late 2022, demand for GPU compute surged, and Microsoft agreed to rent NVIDIA GPUs through CoreWeave to meet OpenAI's needs in a deal worth potentially billions of dollars.^[7] In April 2023, the company raised $221 million in a Series B funding round led by Magnetar Capital, with participation from NVIDIA, former GitHub CEO Nat Friedman, and former Apple executive Daniel Gross. TechCrunch reported that the round valued the company at $2 billion before the new money.^[53] Magnetar added $200 million in May 2023, bringing the round to $421 million.^[54]

In January 2023, CoreWeave acquired Conductor Technologies, a cloud-based rendering startup that simplified access to cloud resources at scale. This acquisition enhanced CoreWeave's offerings to VFX studios and brought its headcount to 95 employees. Mac Moore, former CEO of Conductor Technologies, began leading CoreWeave's Media and Entertainment division.^[59]

In June 2023, in the MLPerf Training v3.0 round, a cluster of 3,584 NVIDIA H100 GPUs co-developed by Inflection AI and operated by CoreWeave completed the benchmark based on the 175-billion-parameter GPT-3 large language model in under 11 minutes. CoreWeave said the result was more than 29 times faster than the next best competitor's submission.^[60][61] In July 2023, CoreWeave announced a $1.6 billion data center in Plano, Texas, and in August 2023 it raised a $2.3 billion debt facility led by Magnetar Capital and Blackstone, for which it posted the NVIDIA GPUs it bought as collateral.^[62][55][7] In December 2023, a $642 million minority stake sale led by Fidelity valued the company at $7 billion.^[56]

Pre-IPO expansion (2024)

In May 2024, CoreWeave raised $1.1 billion in Series C funding led by Coatue Management, with participation from Magnetar Capital, Altimeter Capital, Fidelity Management & Research Co., and Lykos Global Management.^[2] The round valued the company at $19 billion, according to press reports.^[57]

Also in May 2024, the company secured a $7.5 billion debt financing facility led by Blackstone and Magnetar Capital, which Blackstone called "one of the largest private credit financings in history."^[4] A further $650 million revolving credit facility was announced in October 2024.^[57] In CoreWeave's filings the Blackstone facility appears as DDTL 2.0. CNBC reported that it carried an interest rate of about 11%, against more than 14% on the 2023 facility, and co-founder Brian Venturo said CoreWeave structured each customer deal to cover the debt used to buy the GPUs serving it.^[3][7][35]

In November 2024, CoreWeave closed a $650 million secondary sale of existing shares led by Jane Street, Magnetar Capital, Fidelity, and Macquarie Capital, with Cisco and Pure Storage also investing; Bloomberg had reported that CoreWeave was targeting a $23 billion valuation for the deal.^[57][58] During 2024, CoreWeave hired key executives to prepare for its public listing. Nitin Agrawal, previously vice president of finance at Google's cloud unit, became Chief Financial Officer in March 2024, and Sachin Jain, who had been a senior vice president at Oracle Cloud and before that a vice president at Google Cloud, became Chief Operating Officer in August 2024.^[65]

When did CoreWeave go public?

CoreWeave went public on March 28, 2025, listing on the Nasdaq under the ticker CRWV. The IPO priced at $40.00 per share, with 37.5 million shares offered (36.59 million sold by CoreWeave and 910,000 by selling stockholders).^[1] The offering raised approximately $1.5 billion, making it the largest U.S. technology IPO since UiPath's debut in 2021; TechCrunch described it as the largest AI-related listing to date.^[3][29] At $40 a share the company was valued at about $19 billion on its shares outstanding, and more on a fully diluted basis.^[63] The stock opened at $39 and closed flat at $40 on its first day of trading.^[29]

CoreWeave had first marketed the shares at $47 to $55 and planned to sell 49 million of them, but it cut both the price and the size of the offering in the final days before listing as investor demand fell short during a weak market for technology stocks; the Nasdaq fell almost 3% on the day of the debut.^[29][63] NVIDIA anchored the deal with a $250 million order at $40 a share, which CNBC described as helping to save the IPO.^[7][29] "There's a lot of headwinds in the macro," CEO Michael Intrator said on CNBC on the day of the listing. "And we definitely had to scale or rightsize the transaction for where the buying interest was."^[29]

The shares more than doubled during May 2025 and reached an intraday all-time high of $187.00 on June 20, 2025, before falling back; they closed 2025 at $71.61.^[64] Glenn Hutchins joined the board in February 2025 and has been its lead independent director since March 2025.^[65]

Acquisition activity (2025)

After the IPO, CoreWeave made several acquisitions, mostly of software companies, and tried to buy one of its data center landlords.

On May 5, 2025, CoreWeave completed its acquisition of Weights & Biases, the developer tools and observability platform for AI model training, evaluation, and monitoring.^[16] The companies did not disclose the price; The Information reported that the deal could be valued at about $1.7 billion.^[68] When it announced the deal in March 2025, CoreWeave said the Weights & Biases platform was used by over 1,400 organizations and that the combination would give customers an "end-to-end experience" running from bare-metal GPU clusters through model lifecycle tooling.^[69] In June 2025, CoreWeave introduced three new software products in preview: Mission Control Integration, W&B Inference, and W&B Weave Online Evaluations.^[70]

In July 2025, CoreWeave announced an all-stock agreement to acquire Core Scientific, a publicly traded Bitcoin miner that had been hosting a significant portion of CoreWeave's compute capacity. The implied equity value of approximately $9.0 billion would have given CoreWeave roughly 1.3 gigawatts of gross power across Core Scientific's national data center footprint, with an additional one gigawatt or more of expansion potential.^[17] Core Scientific stockholders were offered 0.1235 newly issued CoreWeave Class A shares per Core Scientific share, valuing the target at about $20.40 per share, a 66% premium to its closing price before deal talks were reported.^[17] On October 30, 2025, Core Scientific terminated the merger agreement after it did not receive the stockholder votes needed to approve it.^[18] Investors led by Two Seas Capital had argued that the all-stock offer undervalued Core Scientific's growth in AI hosting, and the proxy adviser ISS recommended a vote against it.^[71] Intrator said CoreWeave respected the vote and looked forward to continuing its commercial partnership with Core Scientific, that its strategy remained unchanged, and that it would keep pursuing "opportunistic and strategic mergers and acquisitions."^[71]

CoreWeave also bought several smaller AI software companies in 2025. It agreed on September 3 to acquire OpenPipe, whose tools train AI agents with reinforcement learning.^[72] It announced a deal for Monolith AI, a British company that applies machine learning to physics and engineering problems, on October 6 and completed it in November, paying with $185 million of non-interest-bearing convertible notes.^[73][35] On October 30 it agreed to acquire Marimo, the company behind the open-source marimo notebook, issuing 844,735 Class A shares.^[74][35]

2026 contracts

In the first half of 2026 CoreWeave signed a run of multibillion-dollar customer agreements, and its revenue backlog rose from $66.8 billion at the end of 2025 to $99.4 billion at March 31, 2026 and about $104 billion at June 30, 2026.^[8][23][36]

On March 4, 2026, CoreWeave announced a multi-year strategic partnership with Perplexity to power Perplexity's inference workloads on CoreWeave Cloud, with Perplexity running on dedicated NVIDIA GB200 NVL72 clusters. Perplexity began running inference workloads using CoreWeave Kubernetes Service and Weights & Biases model management tools.^[22] In a reciprocal arrangement, CoreWeave deployed Perplexity Enterprise Max for use across its own organization.^[22] CoreWeave shares rose about 5.7% in pre-market trading on the day it was announced, according to 24/7 Wall St.^[78]

On April 9, 2026, CoreWeave and Meta announced an expanded long-term agreement worth approximately $21 billion that runs through December 2032.^[19] The order form, signed on March 31, 2026, covers new capacity through December 20, 2032 and the exercise of an option for additional capacity under Meta's earlier order through April 10, 2032.^[43] It adds to the up to $14.2 billion contract signed in September 2025, bringing the two companies' combined commitments to roughly $35 billion.^[20][42] The dedicated capacity will be deployed across multiple locations and will include some of the initial deployments of the NVIDIA Vera Rubin platform, which Meta will use to scale its AI inference workloads.^[19]

On April 10, 2026, the day after the Meta announcement, CoreWeave disclosed a multi-year agreement with Anthropic to support the development and deployment of the Claude family of AI models. Compute under the agreement was scheduled to come online starting later in 2026, beginning with a phased infrastructure roll-out that could expand over time.^[31] Announcing the deal, Intrator framed CoreWeave's expanding role above raw compute: "AI is no longer just about infrastructure, it's about the platforms that turn models into real-world impact."^[31] CoreWeave declined to disclose the value; Intrator told Bloomberg it was a multibillion-dollar contract.^[20][77] With Anthropic signed, CoreWeave said, nine of the ten leading AI model providers used its platform; the company told CNBC that the one it did not yet work with was Elon Musk's xAI.^[20][31] The stock rose 11% on the announcement, closing at $102.00.^[20][64]

On April 15, 2026, CoreWeave announced that Jane Street, the quantitative trading firm, had committed approximately $6 billion to use its AI cloud platform and had separately bought $1 billion of CoreWeave Class A stock at $109.00 per share.^[21] The agreement includes NVIDIA Vera Rubin technology, which Jane Street said it would use for training large, complex models on massive volumes of noisy data.^[21]

Second half of 2026

CoreWeave reported second-quarter 2026 results on August 11, 2026. Revenue of $2.58 billion beat analysts' estimates, and the company raised its 2026 revenue forecast to $12.4 billion to $13.2 billion; the shares rose 19% the next day.^[38][64] It said it had added more than $25 billion of net new customer commitments early in the third quarter, which were not yet in its June 30 backlog, and that its contracted power had grown to about 4.2 gigawatts by August 11.^[36][48]

On August 20, 2026, CoreWeave announced a multi-year agreement with Hudson River Trading to build a research platform on NVIDIA Vera Rubin NVL72 and HGX B200 clusters connected by Spectrum-X Ethernet; Bloomberg described it as a multibillion-dollar deal.^[50][51] In September 2026 the company sold $4.2 billion of 2.875% convertible senior notes due 2033 and said that short-term contracts it had signed in the third quarter were priced at about $40 million of annualized revenue per megawatt.^[47][48] On September 30 it made Vera Rubin NVL72 systems available, with Cognition as the first customer running production workloads on them, and on October 6 it said it would be the sole tenant of 240 MW of AdaniConneX data center capacity in Navi Mumbai, its first expansion into India.^[49][52]

On October 14, 2026, CoreWeave's shares fell 15.5% and the yields on its unsecured bonds rose to about 15%, as investors grew concerned about how heavily AI infrastructure companies were borrowing.^[39] Intrator told employees in a memo that "there will be periods of time when the market misunderstands and acts irrationally."^[39]

Who founded CoreWeave?

In June 2025, three months after the IPO, Bloomberg estimated CEO Michael Intrator's net worth at about $10 billion, according to TechCrunch.^[13] All three original co-founders had each sold more than $150 million of shares before the IPO, and together they held about 80% of the voting power after it through ten-vote Class B shares.^[13][34]

What is CoreWeave's business model?

CoreWeave operates as a specialized cloud computing provider focused on GPU-accelerated workloads. Unlike general-purpose hyperscale cloud providers that offer a broad range of computing services, CoreWeave concentrates on delivering high-performance GPU infrastructure optimized for AI training, inference, rendering, and high-performance computing (HPC).

Revenue model

CoreWeave primarily generates revenue through long-term contracts with enterprise customers who reserve GPU capacity. Customers buy access to the platform over the contract term on a take-or-pay basis, committing to pay for a minimum amount of services whether or not they use it, and at the end of 2025 its committed contracts had a weighted-average duration of about five years.^[34][35] Revenue backlog was about $104 billion at June 30, 2026, up from $66.8 billion at the end of 2025 and $99.4 billion at March 31, 2026.^[8][23][36] Remaining performance obligations, a narrower accounting measure, were $103.7 billion at June 30, 2026; CoreWeave expected to recognize 41% of that within 24 months and 39% in the following two years.^[37] In May 2026, Intrator said ten customers had committed to spend at least $1 billion each with the company.^[30]

The company's customer concentration has been a notable feature of its business and a recurring topic in equity research. Microsoft accounted for 35% of CoreWeave's revenue in 2023, 62% in 2024 and about 67% in 2025.^[34][32][35] Contracts with OpenAI, Meta and others have since reduced the dependence on one customer: CoreWeave's largest customer, which its quarterly report does not name, accounted for 71% of revenue in the second quarter of 2025 but 36% in the second quarter of 2026, when a second customer accounted for 26% and a third for 10%.^[37]

Pricing and competitive positioning

In 2023 CoreWeave's website advertised compute that was "up to 35x faster and 80% less expensive than legacy cloud providers."^[88] It has also claimed up to 20% higher GPU cluster performance than alternative solutions (see Performance benchmarks below).^[83] In August 2026, Intrator said that "pricing and margins for our Blackwell and Vera Rubin SKUs are setting new highs, while pricing for prior generation SKUs is at or above where it was years ago."^[38] In September 2026 the company said that short-dated contracts of about three to six months signed in the third quarter were priced at approximately $40 million of annualized revenue per megawatt of power.^[48] In mid-October 2026, its public price list showed on-demand rates of $49.24 an hour for an eight-GPU HGX H100 instance, $50.44 for HGX H200, $68.80 for HGX B200, and $42.00 for a four-GPU GB200 NVL72 instance, with GB300 NVL72 and HGX B300 systems priced on request; it advertised discounts of up to 60% for committed usage.^[87]

Software stack

The acquisition of Weights & Biases gave CoreWeave a software business that includes a model training platform (W&B Models), an inference serving product (W&B Inference, introduced in preview in June 2025), and an LLM evaluation and observability suite (Weave).^[70] The 2025 purchases of OpenPipe, Marimo and Monolith added reinforcement learning tools for AI agents, a notebook environment for AI developers, and machine learning software for industrial engineering.^[72][73][74] Together these products extend CoreWeave from renting compute into the software that developers use to train, evaluate and deploy AI models.^[69][70]

On October 21, 2026, at NVIDIA's GTC Berlin conference, CoreWeave announced the CoreWeave Zero Trust AI Framework, which it called "a customer-controlled security model" designed to keep CoreWeave out of the customer's trust chain.^[108] The framework groups six controls: customer-held encryption keys through Remote Key Encryption, customer authorization of provider access through Support Access Management, hardware-rooted attestation, confidential computing, a commitment not to use customer prompts, training data, model weights or outputs to train CoreWeave's own models, and MACsec encryption across CoreWeave's backbone.^[108][109] The confidential-computing layer was announced as a plan: CoreWeave said it "intends" to extend NVIDIA's third-generation Confidential Computing and would begin with select bare-metal environments.^[108]

Technology and infrastructure

Kubernetes-native architecture

CoreWeave's cloud platform is built on a Kubernetes-native architecture designed specifically for large-scale, GPU-intensive tasks. The CoreWeave Kubernetes Service (CKS) is a managed Kubernetes service that runs clusters on bare-metal servers with no hypervisor layer, and CoreWeave says it supports clusters scaling to more than 100,000 GPUs.^[84]

CKS clusters use bare-metal nodes with NVIDIA BlueField Data Processing Units (DPUs) for offloading node and resource management processes. The platform natively integrates with workload orchestration tools like Slurm, KubeFlow, and KServe.^[84]

SUNK (Slurm on Kubernetes)

CoreWeave developed SUNK (Slurm on Kubernetes), an implementation of Slurm deployed on Kubernetes using Helm. SUNK is designed to be a unified training system for demanding AI workloads, delivering production-grade reliability and operational visibility for large, long-running training jobs.

A standard SUNK deployment uses two Helm charts, one for the operator and cluster-wide dependencies and one that defines a Slurm cluster, and its compute images ship per CUDA version, built on NCCL test images. Each Slurm compute node runs as a Kubernetes Pod running slurmd, placed one-to-one on a Kubernetes node, and groups of nodes are declared as NodeSets that set the image, resources and replica count used to define and scale the cluster.^[85]

CoreWeave Mission Control monitors cluster health end-to-end, detects silent hardware issues and GPU stragglers, and mitigates failures before they stall synchronous training.^[26][86] Mission Control integration with Weights & Biases, announced in June 2025, surfaces hardware-level events directly inside the W&B experiment tracking interface.^[70]

Networking

CoreWeave's GPU clusters use NVIDIA InfiniBand with SHARP (Scalable Hierarchical Aggregation and Reduction Protocol), providing high-bandwidth, low-latency interconnects between GPUs.^[84] Its Blackwell instances scale to as many as 110,000 GPUs on NVIDIA Quantum-2 InfiniBand networking.^[15] Some newer deployments use Ethernet instead: the research platform announced with Hudson River Trading in August 2026 runs on NVIDIA Spectrum-X Ethernet.^[50]

GPU hardware

In mid-October 2026, CoreWeave offered NVIDIA GPU systems across several generations:^[49][87]

CoreWeave says it became the first cloud provider to make NVIDIA GB200 NVL72 instances generally available, in February 2025.^[79] In April 2025 it brought thousands of Grace Blackwell GPUs online for Cohere, IBM, and Mistral AI, the first customers to gain access to its GB200 NVL72 systems.^[15][80] Cohere used the systems for its North enterprise agent platform and reported up to three times more training performance for 100-billion-parameter models than with the previous Hopper generation, IBM used them to train its next-generation Granite models, which support its watsonx platform, and Mistral AI received its first thousand Blackwell GPUs to build new open-source models.^[80]

In July 2025, CoreWeave said it had become the first AI cloud provider to deploy the NVIDIA GB300 NVL72 platform for customers; the platform offers up to a tenfold improvement in user responsiveness and a fivefold improvement in throughput per watt compared to the previous-generation Hopper architecture.^[9] In January 2026, CoreWeave said it expected to be among the first cloud providers to deploy the NVIDIA Rubin platform in the second half of the year.^[26] It announced the availability of Vera Rubin NVL72 on September 30, 2026, with the AI coding company Cognition as the first customer running production workloads on it; Meta's and Jane Street's April 2026 agreements both include early Vera Rubin capacity.^[49][19][21]

Performance benchmarks

In June 2025, CoreWeave, in collaboration with NVIDIA and IBM, made what it called the largest-ever MLPerf Training v5.0 submission on NVIDIA Blackwell, using 2,496 Blackwell GPUs in GB200 NVL72 systems. The system trained the Llama 3.1 405B model in 27.3 minutes, a record result on the most complex model in the benchmark suite.^[81] In the MLPerf Training v6.0 round, CoreWeave said in June 2026 that it had trained DeepSeek-V3 in 2.02 minutes on 8,192 GB300 GPUs and reached the Llama 3.1 405B quality target in 9.77 minutes on 4,096 GB300 GPUs.^[82]

CoreWeave has reported up to 20% higher GPU cluster performance compared to alternative cloud solutions. The figure comes from the company's own analysis, which estimated model FLOPS utilization above 50% on its Hopper clusters against 35% to 45% in published training benchmarks.^[83] In September 2026, Cognition reported up to 4.8 times more total token throughput for its SWE-2 inference workloads on Vera Rubin NVL72 than on GB200 NVL72, both measured on CoreWeave's cloud.^[49]

Data centers

CoreWeave operated 32 data centers running more than 250,000 GPUs, with more than 360 MW of active power, at the end of 2024, and 43 data centers with more than 850 MW at the end of 2025.^[34][35] By May 2026 it had passed one gigawatt of active power across almost 50 data center locations, and it reported about 1.5 GW at June 30, 2026, with a target of more than 1.85 GW by the end of 2026.^[27][36][38] Contracted power, capacity that CoreWeave has secured but not yet brought online, grew from about 3.1 GW at the end of 2025 to about 3.7 GW at June 30, 2026 and about 4.2 GW by August 11, 2026.^[8][36][48] The company has said it aims to bring most of that contracted power online by the end of 2027 and that it is well on its way to more than 8 GW by 2030.^[23][27]

As of May 2026, all of CoreWeave's operating capacity was leased from data center partners, none of which held more than 17% of it. Intrator said the company would accelerate development of its own self-build sites, "which will provide us with greater operational control and long-term financial upside," and expected the first to come online later in 2026.^[27]

United States

Europe

When it announced the continental sites in June 2024, CoreWeave put its total European commitment at $3.5 billion, combining the $1.3 billion UK investment with the $2.2 billion continental expansion.^[10] The European data centers are designed to meet regional regulatory and operational requirements, providing data-sovereign cloud services with low-latency performance.^[12] In September 2025 CoreWeave announced a further £1.5 billion for the UK, which it said brought its total UK investment to £2.5 billion.^[99]

India

Funding and financials

Funding rounds and capital structure

The table draws on company announcements, filings and press reports.^{[2][4][11][21][25][34][35][36][45][46][47][53][54][55][56][57][58][103][104]}

CoreWeave said it had secured more than $20 billion of debt and equity in 2026 by early May and more than $30 billion by August 10.^[30][46] Its total indebtedness rose from $21.6 billion at the end of 2025 to about $25 billion at March 31, 2026 and about $35 billion at June 30, 2026.^[35][23][36] The capital stack is a mix of delayed draw term loans secured by GPUs and customer contracts (DDTL 1.0 through 5.5), senior unsecured notes, convertible notes, OEM and vendor financing, and a revolving credit facility.^[35][46] Borrowing costs generally fell through early 2026: the 2023 DDTL 1.0 facility carried an actual interest rate above 14% and DDTL 2.0 about 11%, according to CNBC, while DDTL 3.0 was priced at SOFR plus 4%.^[7][11] The non-recourse DDTL 4.0 facility of March 2026, which Moody's rated A3, was the first investment-grade rated financing secured by HPC infrastructure and an associated customer contract; it was priced at SOFR plus 2.25% for its floating-rate tranche and about 5.9% for its fixed-rate tranche.^[45] The DDTL 5.5 facility of August 2026, rated Ba2 by Moody's and BB+ by Fitch, was priced at SOFR plus 5.5%.^[46]

Key investors

Major investors have included:

• Magnetar Capital (first invested in November 2021 through $50 million of convertible notes, then led the 2023 Series B and co-led the 2023 and 2024 debt facilities; CNBC described it in March 2025 as CoreWeave's biggest equity investor)^{[4][7][53][55][103][104]}
• NVIDIA (strategic investor since Series B in 2023; bought $2 billion of stock at $87.20 a share in January 2026, which Reuters calculated nearly doubled its stake from 6.3% and made it the second-largest shareholder)^[25][44][53]
• Blackstone (led the $7.5 billion DDTL 2.0 facility; anchor investor in the $8.5 billion DDTL 4.0 facility of March 2026)^[4][45]
• Coatue Management (led the $1.1 billion Series C)^[2]
• Jane Street (co-led the November 2024 secondary sale; bought $1 billion of stock at $109.00 a share in April 2026)^[21][58]
• Fidelity Management & Research (led the December 2023 secondary sale)^[56]
• Altimeter Capital
• Macquarie Capital
• Nat Friedman (angel investor, former GitHub CEO)
• Daniel Gross (angel investor)

How fast is CoreWeave's revenue growing?

CoreWeave's IPO prospectus put its revenue at $16 million for 2022, $229 million for 2023 and $1.9 billion for 2024.^[34] It reported full-year 2025 revenue of $5.13 billion, more than 2.6 times its 2024 total, against a net loss of $1.17 billion and adjusted EBITDA of $3.09 billion.^[8][14] In November 2025 it had cut its 2025 revenue forecast to $5.05 billion to $5.15 billion because a third-party data center developer had fallen behind schedule, and the shares fell 16% the next day.^[64][100] The company reiterated on its Q1 2026 earnings call that annualized revenue should exceed $30 billion by the end of 2027.^[30] In August 2026 it raised its 2026 revenue forecast to $12.4 billion to $13.2 billion, from $12 billion to $13 billion, and forecast third-quarter revenue of $3.4 billion to $3.6 billion.^[38]

Is CoreWeave profitable?

Despite rapid revenue growth, CoreWeave has not yet achieved profitability on a net income basis due to heavy capital expenditures on GPU hardware, data center buildouts, and debt servicing costs: it reported net losses of $594 million in 2023, $863 million in 2024, $1.17 billion in 2025 and $1.37 billion in the first half of 2026.^[35][37] The company's profitability picture is, however, sharply positive on an adjusted EBITDA basis.

The company's Q4 2025 results showed revenue of $1.57 billion (up 110% year-over-year), adjusted EBITDA of $898 million (a 57% margin), and a net loss of $452 million.^[8] In Q1 2026 CoreWeave reported revenue of $2.08 billion, up 111.6% year-over-year, beating Wall Street's consensus of $1.97 billion.^[23][30] Adjusted EBITDA rose 90.9% to $1.16 billion, a 56% margin.^[23][30] The net loss came to $1.40 per diluted share, as interest expense roughly doubled from a year earlier to $536 million.^[28] The scale of the buildout was visible on the balance sheet: property and equipment reached $36.4 billion as of March 31, 2026, on first-quarter cash purchases of property and equipment of $7.7 billion, while operating activities generated $2.98 billion of cash in the quarter.^[23] Summarizing the quarter, CEO Michael Intrator said it was "the strongest bookings quarter in CoreWeave's history, with revenue backlog reaching nearly $100 billion."^[23] Its second-quarter revenue guidance of $2.45 billion to $2.6 billion fell short of the $2.69 billion analysts expected, and the shares dropped as much as 10% in extended trading.^[30]

In Q2 2026, revenue rose 112% to $2.58 billion, slightly above the $2.56 billion analysts expected, while the net loss widened to $626 million, or $1.14 per share, as net interest expense rose to $640 million from $267 million a year earlier; adjusted EBITDA was $1.51 billion, a 59% margin.^[36][38] Cash purchases of property and equipment reached $14.1 billion in the first half of 2026, property and equipment stood at $46.7 billion at June 30, 2026, and CoreWeave raised its 2026 capital expenditure forecast to $35 billion to $39 billion, from $31 billion to $35 billion in May.^[36][38]

Major customers and contracts

Microsoft

Microsoft has been CoreWeave's largest customer since 2023, when it began renting NVIDIA GPUs through CoreWeave to meet demand from OpenAI for training and inference of frontier models.^[7] Microsoft accounted for 62% of CoreWeave's total revenue in 2024 and about 67% in 2025.^[32][35] The Information reported in late 2024 that Microsoft planned to spend about $10 billion on CoreWeave's services by the end of the decade.^[76] Reliance on a single customer has fallen as newer contracts ramp up: CoreWeave's largest customer, which its quarterly report does not name, accounted for 71% of revenue in the second quarter of 2025 and 36% in the second quarter of 2026.^[37]

OpenAI

CoreWeave has built a substantial relationship with OpenAI through a series of escalating contracts:

The initial OpenAI agreement, announced March 10, 2025, was worth up to $11.9 billion over five years and included OpenAI taking a $350 million equity stake in CoreWeave.^[5][7][34] Successive expansions in May 2025 (up to $4 billion) and September 2025 (up to $6.5 billion, through May 31, 2031) brought the cumulative direct commitment to approximately $22.4 billion.^[5][33][35] CoreWeave financed the equipment for its OpenAI contract in part with the $2.6 billion DDTL 3.0 facility of July 2025.^[11] The OpenAI partnership complements OpenAI's existing commercial relationships with Microsoft and Oracle, as well as its joint venture with SoftBank on the Stargate project. Reuters reported in June 2025, citing unnamed sources, that CoreWeave would also supply computing capacity to Google Cloud, which would resell it to OpenAI; CoreWeave, Google and OpenAI declined to comment.^[75] That arrangement would give CoreWeave a third channel to OpenAI demand, on top of its direct contracts and its Microsoft-mediated capacity.

Meta

In September 2025, CoreWeave secured a $14.2 billion contract with Meta to provide AI cloud infrastructure.^[6] Under the original order form, signed on September 25, 2025, Meta committed to pay CoreWeave up to approximately $14.2 billion through December 14, 2031, with the option to materially expand its commitment through 2032.^[6][42] On April 9, 2026, CoreWeave and Meta announced an expanded agreement that adds approximately $21 billion in committed value through December 2032, bringing the cumulative total between the two companies to roughly $35 billion.^[19][20] The new order form, signed on March 31, 2026, includes new capacity through December 20, 2032 and the exercise of the option under the earlier order through April 10, 2032.^[43] Capacity under the expanded deal will use some of the initial NVIDIA Vera Rubin platform deployments and is distributed across multiple CoreWeave locations to support Meta's AI inference workloads.^[19] CoreWeave said the same day that it was raising $3 billion of new debt to fund the Meta deal, CNBC reported.^[20]

Anthropic

On April 10, 2026, CoreWeave announced a multi-year agreement with Anthropic to support the development and deployment of Anthropic's Claude family of AI models.^[20][31] Anthropic will run production-scale workloads on CoreWeave's cloud platform, with compute coming online starting later in 2026 and expanding through a phased rollout.^[20][31] CoreWeave declined to disclose the contract's value, and Intrator described it to Bloomberg as a multibillion-dollar contract.^[20][77] With Anthropic onboard, CoreWeave said, nine of the ten leading AI model providers were using its platform; the company told CNBC that the remaining one was xAI.^[20][31]

Perplexity

On March 4, 2026, CoreWeave entered into a multi-year strategic partnership with Perplexity to support Perplexity's inference workloads.^[22] Perplexity runs inference on dedicated NVIDIA GB200 NVL72-powered clusters via CoreWeave Kubernetes Service, and uses W&B Models to manage models from experimentation through production.^[22] The infrastructure supports Perplexity's Sonar and Search API ecosystem. In a reciprocal commercial arrangement, CoreWeave deployed Perplexity Enterprise Max for its own employees to use across internal search and research workflows.^[22]

Jane Street

On April 15, 2026, Jane Street, a global quantitative trading firm, committed approximately $6 billion to use CoreWeave's AI cloud platform and made a separate $1 billion equity investment at $109.00 per share.^[21] The agreement includes NVIDIA Vera Rubin technology, which Jane Street uses for training large, complex models on massive volumes of noisy data.^[21] Jane Street had already co-led CoreWeave's November 2024 secondary share sale.^[58]

Other notable customers

In addition to the marquee contracts, CoreWeave provides infrastructure to:

• Cohere: Personalized AI agent platform North runs on GB200 capacity, with training performance up to three times higher than on previous-generation Hopper systems.^[80]
• IBM: Trains the next-generation Granite models on GB200 systems for the IBM watsonx platform.^[80]
• Mistral AI: Received its first thousand Blackwell GPUs to build new open-source models.^[80]
• Hudson River Trading: Signed a multi-year agreement in August 2026 for a research platform built on Vera Rubin NVL72 and HGX B200 clusters, after first becoming a customer earlier in 2026.^[23][36][50]
• Cognition: The first customer to run production workloads on CoreWeave's Vera Rubin NVL72 systems, from September 2026.^[49]
• NVIDIA: Both a strategic investor and a customer; it had paid CoreWeave $320 million by the end of 2024 and signed a $6.3 billion capacity order in September 2025.^[29][41]
• Others named in CoreWeave's earnings releases, including Cursor, Midjourney, Mercado Libre, CrowdStrike, Databricks, Caterpillar and Isomorphic Labs.^[8][36]
• Various VFX and media customers carried over from the pre-AI rendering business.

NVIDIA relationship

The relationship between CoreWeave and NVIDIA is one of the most important dynamics in the AI infrastructure industry. NVIDIA is simultaneously a strategic investor, technology partner, key supplier, and customer of CoreWeave.

NVIDIA first invested in CoreWeave during the 2023 Series B funding round.^[7][53] In March 2025, during the IPO, NVIDIA anchored the offering with a $250 million order at the $40 IPO price.^[29] On September 9, 2025, CoreWeave and NVIDIA signed an order form with an initial value of $6.3 billion under which NVIDIA is obligated to purchase any residual unsold capacity through April 13, 2032.^[41] In January 2026, NVIDIA made a $2 billion private placement investment at $87.20 per share, which Reuters calculated would nearly double the 6.3% stake that had made it CoreWeave's third-largest shareholder and make it the second-largest.^[25][44] The two companies said NVIDIA would use its financial strength to help CoreWeave procure land, power and shell to build more than 5 GW of AI factories by 2030, and that CoreWeave would deploy NVIDIA's Rubin platform, Vera CPUs and BlueField storage systems.^[25]

CoreWeave has consistently been among the first cloud providers to deploy new NVIDIA GPU architectures. It says it was the first cloud provider to make NVIDIA GB200 NVL72 instances generally available (February 2025) and the first AI cloud provider to deploy NVIDIA GB300 NVL72 systems for customers (July 2025).^[79][9] In January 2026 it said it expected to be among the first cloud providers to deploy the NVIDIA Vera Rubin platform in the second half of the year, and it made Vera Rubin NVL72 available on September 30, 2026.^[26][49] CoreWeave was also one of the cloud partners named when NVIDIA launched its DGX Cloud Lepton marketplace in May 2025.^[89]

CoreWeave uses NVIDIA BlueField DPUs in its data centers for network and resource management, and its network infrastructure relies on NVIDIA InfiniBand and Quantum-2 networking technology.^[15][84] Its customers have contractually specified the use of NVIDIA GPUs.^[35]

How does CoreWeave differ from its competitors?

CoreWeave operates in the rapidly growing GPU cloud infrastructure market. In October 2026, the research firm Gartner forecast that spending on AI-specific infrastructure as a service would grow from $66 billion in 2027 to $174 billion in 2030.^[40]

Hyperscale cloud providers

The traditional "Big Three" cloud providers (Amazon Web Services, Microsoft Azure, and Google Cloud) all offer GPU instances for AI workloads. CoreWeave differentiates itself through GPU specialization, faster provisioning, and competitive pricing. The hyperscalers offer broader service portfolios, while CoreWeave focuses on GPU-intensive compute. Microsoft has been both a customer and a competitor: it rents CoreWeave capacity at scale, much of it for OpenAI workloads, through finance leases that supplement the NVIDIA GPUs in its own data centers.^[7][32]

Specialized GPU cloud providers (neoclouds)

CoreWeave competes with a growing category of GPU-focused cloud providers:

CoreWeave is positioned at the enterprise end of the GPU cloud market, requiring multi-year contract commitments for exclusive access to large GPU clusters. Lambda Labs targets more flexible, growth-stage companies, while Together AI focuses on startups with pay-per-token pricing. Analysts often compare CoreWeave with Nebius; in October 2026, DA Davidson's Gil Luria said the two "literally sell the same thing to the same companies," while Nebius carries almost no financial leverage.^[40]

Employees

CoreWeave has experienced rapid headcount growth. At the time of the Conductor Technologies acquisition in January 2023, the combined company had 95 employees.^[59] By the end of 2024, headcount had grown to 881.^[29] At the end of 2025 CoreWeave had 2,189 employees, 1,967 of them in the United States and 222 in 19 other countries.^[35] Michelle O'Rourke, who joined the company's human resources team in 2021, was named Chief People Officer in October 2024 to lead talent acquisition, employee development and culture.^[67]

Stock performance

CoreWeave trades on the Nasdaq under the ticker CRWV. The stock has been volatile since its IPO at $40.00 per share in March 2025:

Prices are Nasdaq closing prices.^[64] Market capitalization is approximated from each closing price and the Class A and Class B shares outstanding reported on the nearest SEC filing cover page.^{[34][35][37][105][106][107]}

After the first-quarter 2026 report, Wells Fargo analyst Michael Turrin raised his price target to $155 from $135 on May 8, 2026, maintaining an Overweight rating; TheStreet, citing Stock Analysis, put the median analyst target at $125 at the time.^[101] Opinion turned more cautious in October 2026. On October 14, the yield to maturity on CoreWeave's 9.625% notes due 2032 reached 14.8% and that on its 9.75% notes due 2031 about 14.9%, and Paul Meeks of Freedom Capital Markets called its balance sheet "relatively bad" compared with its neocloud peers.^[39] The next day the 2032 bond's yield rose to 15.77% and five-year credit default swaps on CoreWeave debt traded at a spread of 1,140 basis points.^[40] In a memo to employees, Intrator wrote: "When you are building a new type of technology company there will be periods of time when the market misunderstands and acts irrationally. We are confident in our backlog which more than supports all of our debt obligations."^[39]

Risks and challenges

Debt load

CoreWeave's aggressive expansion has been funded primarily through debt. Total debt reached about $35 billion at June 30, 2026, up from $21.6 billion of total indebtedness at the end of 2025, and net interest expense was $640 million in the second quarter of 2026, against $267 million a year earlier.^[35][36] Much of the borrowing is tied to specific customer contracts: DDTL 3.0 funded equipment for the OpenAI contract, and DDTL 4.0 was secured by HPC infrastructure and an associated customer contract.^[11][45] NVIDIA's $6.3 billion capacity order obliges it to buy any reserved capacity that CoreWeave leaves unsold through April 2032.^[41] Still, Gavan Nolan of S&P Global Market Intelligence said in October 2026 that CoreWeave "has capital expenditure levels that are almost hyperscaler-like, but it doesn't have the hyperscaler balance sheet."^[39] Servicing this debt requires sustained revenue growth and presents a significant financial risk if demand for GPU compute were to slow; in October 2026, S&P Global Ratings rated CoreWeave B+, a non-investment-grade rating.^[39]

Customer concentration

CoreWeave's historical reliance on a small number of large customers, particularly Microsoft (which accounted for about 67% of 2025 revenue), represented a concentration risk.^[35] The wave of 2025-2026 deals with OpenAI, Meta, Anthropic, Perplexity, and Jane Street brought the largest customer's share down to 36% of revenue in the second quarter of 2026.^[37] However, the customer base is still heavily skewed toward the handful of frontier AI labs and the largest internet platforms.

Capital intensity and GPU depreciation

The GPU cloud business is extremely capital-intensive, requiring continuous investment in the latest GPU hardware and data center infrastructure. GPU technology evolves rapidly, and older hardware can become less competitive quickly. CoreWeave must continually invest billions to maintain its technological edge. It depreciates its technology equipment, including GPUs, over six years, having lengthened the schedule from five years in early 2023.^[102] Depreciation schedules on GPU assets have become a point of disagreement between management and short sellers, with bears arguing that the useful life of high-end GPUs is shorter than CoreWeave's depreciation assumptions imply, which would reduce reported earnings power over time.^[102]

Competition

The GPU cloud market is attracting significant investment from both hyperscale providers expanding their GPU offerings and new entrants building specialized infrastructure. CoreWeave's ability to maintain its competitive advantages in pricing, performance, and provisioning speed will be tested as the market matures. Nebius, in particular, has emerged as a fast-growing direct competitor. In October 2026 Gartner predicted that by 2030 "cost pressure and a lack of a differentiating business model will force over 80% of neocloud providers to cease independent operations."^[40]

GPU supply dependency

CoreWeave's business depends almost entirely on NVIDIA GPUs, creating a single-supplier dependency; its IPO prospectus said that all of the GPUs in its infrastructure at the time came from NVIDIA.^[7] Any disruption in NVIDIA's supply chain, changes in NVIDIA's allocation priorities, or the emergence of competitive chip alternatives such as those from AMD or hyperscaler-built accelerators could affect CoreWeave's operations. The deep equity, commercial, and capital alignment with NVIDIA partially mitigates this risk but also concentrates strategic exposure on a single partner.

Failed Core Scientific acquisition

The October 2025 termination of the Core Scientific acquisition removed an opportunity to vertically integrate roughly 1.3 gigawatts of power into CoreWeave's footprint and left the company paying rent to a hosting provider whose stockholders rejected the offer; CoreWeave had said the deal would eliminate about $10 billion of future lease obligations.^[17][18] CoreWeave has since said it will accelerate development of its own self-build sites.^[27]

Power and siting constraints

CoreWeave depends on partners and its own projects to deliver powered buildings on time. In November 2025, after a delay at one third-party data center developer forced it to cut its 2025 revenue forecast, Intrator said the company's constraint was not power itself but the availability of "powered-shell" data centers.^[100] Bringing most of its contracted power online by the end of 2027 depends on securing power, transmission, and site rights at unprecedented scale, in some cases competing with hyperscalers and other neoclouds for the same substations and utility interconnections.^[27] Opposition to data center construction has also grown: New York's governor signed an executive order in July 2026 establishing a moratorium on new large-scale data centers, and Intrator said in August 2026 that "when parts of the U.S. become unwilling to even engage in those conversations," expansion "becomes more challenging," although he said the company's guidance was not affected.^[38]

Litigation

CoreWeave's annual report for 2025 disclosed a putative securities class action filed against the company and two stockholder derivative actions filed on its behalf.^[35]

See also

• IREN
• List of AI compute deals

References

1. ^1 ^2"CoreWeave Announces Pricing of Initial Public Offering." CoreWeave, 2025-03-27. coreweave.com/...ricing-of-initial-public-offering
2. ^1 ^2 ^3"CoreWeave Secures $1.1 Billion in Series C Funding." PR Newswire, 2024-05-01. prnewswire.com/...g-for-the-future-of-ai-302133328
3. ^1 ^2 ^3 ^4 ^5 ^6"CoreWeave co-founder explains how a closet of crypto-mining GPUs led to a $1.5B IPO." TechCrunch, 2025-03-29. techcrunch.com/...to-mining-gpus-led-to-a-1-5b-ipo
4. ^1 ^2 ^3 ^4"CoreWeave Secures $7.5 Billion Debt Financing Facility led by Blackstone and Magnetar." Blackstone, 2024-05-17. blackstone.com/...y-led-by-blackstone-and-magnetar
5. ^1 ^2"CoreWeave Expands Agreement with OpenAI by up to $6.5B." CoreWeave, 2025-09-25. coreweave.com/...reement-with-openai-by-up-to-6-5b
6. ^1 ^2"CoreWeave stock closes up 12% after company lands $14 billion deal with Meta." CNBC, 2025-09-30. cnbc.com/...coreweave-meta-deal-ai
7. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14 ^15 ^16 ^17 ^18 ^19 ^20 ^21 ^22"Nvidia turned CoreWeave into a major player in AI years before helping to save its IPO." CNBC, 2025-03-30. cnbc.com/...-to-ipo-wound-through-crypto-before-ai
8. ^1 ^2 ^3 ^4 ^5 ^6"CoreWeave Reports Strong Fourth Quarter and Fiscal Year 2025 Results." BusinessWire, 2026-02-26. businesswire.com/...r-and-Fiscal-Year-2025-Results
9. ^1 ^2"CoreWeave Becomes First Cloud Provider to Deploy NVIDIA GB300 NVL72 Platform." CoreWeave, 2025-07-03. coreweave.com/...eploy-nvidia-gb300-nvl72-platform
10. ^1 ^2 ^3 ^4"CoreWeave Announces Significant European Expansion; Commits an Incremental $2.2 Billion." PR Newswire, 2024-06-05. prnewswire.com/...tructure-in-the-region-302164149
11. ^1 ^2 ^3 ^4"CoreWeave Closes $2.6 Billion Secured Debt Financing Facility." BusinessWire, 2025-07-31. businesswire.com/...et-Position-as-AI-Cloud-Leader
12. ^"CoreWeave's European Expansion: Let's Power Tomorrow's AI Innovations." CoreWeave Blog, 2024-06. coreweave.com/...ts-power-tomorrows-ai-innovations
13. ^1 ^2"In just 3 months, CoreWeave CEO, once a crypto-mining bro, becomes a deca-billionaire." TechCrunch, 2025-06-26. techcrunch.com/...g-bro-becomes-a-deca-billionaire
14. ^"CoreWeave (CRWV) Revenue 2022-2025." Stock Analysis. stockanalysis.com/...revenue
15. ^1 ^2 ^3"CoreWeave Launches NVIDIA GB200 Grace Blackwell Systems at Scale." CoreWeave, 2025-04-15. coreweave.com/...-grace-blackwell-systems-at-scale
16. ^"CoreWeave Completes Acquisition of Weights & Biases." PR Newswire, 2025-05-05. prnewswire.com/...ion-of-weights--biases-302445966
17. ^1 ^2 ^3"CoreWeave to acquire Core Scientific in $9 billion all-stock deal." CNBC, 2025-07-07. cnbc.com/...scientific-in-9-billion-all-stock-deal
18. ^1 ^2"Core Scientific Announces Termination of Merger Agreement with CoreWeave." Core Scientific Investor Relations, 2025-10-30. investors.corescientific.com/...ent-with-coreweave
19. ^1 ^2 ^3 ^4 ^5"CoreWeave and Meta Announce $21 Billion Expanded AI Infrastructure Agreement." CoreWeave Investor Relations, 2026-04-09. investors.coreweave.com/...default
20. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10"CoreWeave stock pops 11% on deal to power Anthropic's Claude." CNBC, 2026-04-10. cnbc.com/...coreweave-anthropic-claude-ai-deal
21. ^1 ^2 ^3 ^4 ^5 ^6 ^7"Jane Street Signs $6 Billion AI Cloud Agreement With CoreWeave." CoreWeave Investor Relations, 2026-04-15. investors.coreweave.com/...default
22. ^1 ^2 ^3 ^4 ^5"CoreWeave Announces Agreement to Power Perplexity's AI Inference Workloads." CoreWeave Investor Relations, 2026-03-04. investors.coreweave.com/...default
23. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9"CoreWeave Reports Strong First Quarter 2026 Results." BusinessWire, 2026-05-07. businesswire.com/...ong-First-Quarter-2026-Results
24. ^"Applied Digital Announces 250MW AI Data Center Lease With CoreWeave in North Dakota." Applied Digital Investor Relations, 2025-06-02. ir.applieddigital.com/...ai-data-center-lease-with
25. ^1 ^2 ^3 ^4"NVIDIA and CoreWeave Strengthen Collaboration to Accelerate Buildout of AI Factories." NVIDIA Newsroom, 2026-01-26. nvidianews.nvidia.com/...-buildout-of-ai-factories
26. ^1 ^2 ^3"CoreWeave Extends Its Cloud Platform with NVIDIA Rubin Platform." CoreWeave, 2026-01-05. coreweave.com/...atform-with-nvidia-rubin-platform
27. ^1 ^2 ^3 ^4 ^5"CoreWeave surpasses 1GW of data center capacity, pushes more into self builds." Data Center Dynamics, 2026-05-08. datacenterdynamics.com/...es-more-into-self-builds
28. ^"CoreWeave Q1 2026 slides: revenue surges 112% as profit margins compress." Investing.com, 2026-05. za.investing.com/...-margins-compress-93CH-4263906
29. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9"Nvidia-backed CoreWeave closes flat at $40 after biggest U.S. tech IPO since 2021." CNBC, 2025-03-28. cnbc.com/...-starts-trading-on-nasdaq-at-per-share
30. ^1 ^2 ^3 ^4 ^5 ^6"CoreWeave stock sinks 10% on weak revenue guidance, increased spending forecast." CNBC, 2026-05-07. cnbc.com/...coreweave-crwv-q1-earnings-report-2026
31. ^1 ^2 ^3 ^4 ^5 ^6 ^7"CoreWeave Announces Multi-Year Agreement With Anthropic." CoreWeave, 2026-04-10. coreweave.com/...lti-year-agreement-with-anthropic
32. ^1 ^2 ^3"CoreWeave surges after top customer Microsoft reaffirms spending plans." CNBC, 2025-05-01. cnbc.com/...ter-microsoft-sticks-to-spending-plans
33. ^"CoreWeave Expands OpenAI Deals to as Much as $22.4 Billion." Bloomberg, 2025-09-25. bloomberg.com/...openai-to-as-much-as-22-4-billion
34. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10"CoreWeave, Inc. Prospectus (Form 424B4)." CoreWeave via SEC EDGAR, 2025-03-31. sec.gov/...d899798d424b4
35. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14 ^15 ^16 ^17 ^18 ^19 ^20"CoreWeave, Inc. Annual Report on Form 10-K for 2025." CoreWeave via SEC EDGAR, 2026-03-02. sec.gov/...crwv-20251231
36. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10 ^11 ^12 ^13 ^14"CoreWeave Reports Strong Second Quarter 2026 Results (Exhibit 99.1)." CoreWeave via SEC EDGAR, 2026-08-11. sec.gov/...coreweave2q26earningspress
37. ^1 ^2 ^3 ^4 ^5 ^6"CoreWeave, Inc. Quarterly Report on Form 10-Q for the quarter ended June 30, 2026." CoreWeave via SEC EDGAR, 2026-08-12. sec.gov/...crwv-20260630
38. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8"CoreWeave stock pops 14% as revenue doubles on accelerating AI infrastructure demand." CNBC, 2026-08-11. cnbc.com/...coreweave-crwv-q2-earnings-report-2026
39. ^1 ^2 ^3 ^4 ^5 ^6 ^7"CoreWeave stock sinks as bond yields surge near 15%." CNBC, 2026-10-14. cnbc.com/...coreweave-stock-bond-yields-ai-debt
40. ^1 ^2 ^3 ^4 ^5"What's behind the sell-off in AI neocloud stocks like CoreWeave and Nebius." CNBC, 2026-10-15. cnbc.com/...loud-stocks-like-coreweave-and-nebius-
41. ^1 ^2 ^3"CoreWeave, Inc. Current Report on Form 8-K (order form with NVIDIA)." CoreWeave via SEC EDGAR, 2025-09-15. sec.gov/...crwv-20250909
42. ^1 ^2"CoreWeave, Inc. Current Report on Form 8-K (order form with Meta)." CoreWeave via SEC EDGAR, 2025-09-30. sec.gov/...crwv-20250925
43. ^1 ^2"CoreWeave, Inc. Current Report on Form 8-K (new order form with Meta)." CoreWeave via SEC EDGAR, 2026-04-09. sec.gov/...crwv-20260331
44. ^1 ^2"Nvidia invests $2 billion in CoreWeave to boost data center build-out." Reuters via Yahoo Finance, 2026-01-26. finance.yahoo.com/...2-billion-coreweave-130850099
45. ^1 ^2 ^3 ^4"CoreWeave Closes Landmark $8.5 Billion Financing Facility, Achieving First Investment-Grade Rated GPU-Backed Financing (Exhibit 99.1)." CoreWeave via SEC EDGAR, 2026-03-31. sec.gov/...ex991
46. ^1 ^2 ^3 ^4"CoreWeave Closes $2.6 Billion Loan Facility, Expanding Financing Flexibility for AI Infrastructure (Exhibit 99.1)." CoreWeave via SEC EDGAR, 2026-08-10. sec.gov/...ex991pr
47. ^1 ^2"CoreWeave, Inc. Current Report on Form 8-K (2.875% convertible senior notes due 2033)." CoreWeave via SEC EDGAR, 2026-09-22. sec.gov/...crwv-20260917
48. ^1 ^2 ^3 ^4"CoreWeave Continues to Contract New Compute Capacity at Higher Prices." CoreWeave Investor Relations, 2026-09-17. investors.coreweave.com/...default
49. ^1 ^2 ^3 ^4 ^5 ^6"CoreWeave Delivers NVIDIA Vera Rubin NVL72 Performance at Production Scale, Starting With Cognition." CoreWeave, 2026-09-30. coreweave.com/...ion-scale-starting-with-cognition
50. ^1 ^2 ^3"Hudson River Trading to Build Next-Gen Research Platform Powered by NVIDIA Vera Rubin NVL72 on CoreWeave Cloud." CoreWeave, 2026-08-20. coreweave.com/...ra-rubin-nvl72-on-coreweave-cloud
51. ^"Hudson River Signs Multibillion-Dollar CoreWeave (CRWV) Deal for AI Cloud." Bloomberg, 2026-08-20. bloomberg.com/...-coreweave-crwv-deal-for-ai-cloud
52. ^1 ^2"CoreWeave Enters India With 240 MW AdaniConneX Deployment." CoreWeave, 2026-10-06. coreweave.com/...i-cloud-platform-with-adaniconnex
53. ^1 ^2 ^3 ^4 ^5"CoreWeave, a GPU-focused cloud compute provider, lands $221M investment." TechCrunch, 2023-04-20. techcrunch.com/...e-provider-lands-221m-investment
54. ^1 ^2"CoreWeave Secures $200 Million Series B Extension, Bringing Total Round to $421 Million." Business Wire via FinancialContent, 2023-05-31. markets.financialcontent.com/...und-to-421-million
55. ^1 ^2 ^3"CoreWeave Secures $2.3 Billion Debt Financing Facility led by Magnetar Capital and Blackstone to Meet Surging Demand and Ongoing Expansion of Specialized Cloud Infrastructure to Power AI." PR Newswire, 2023-08-03. prnewswire.com/...astructure-to-power-ai-301892706
56. ^1 ^2 ^3"CoreWeave Valuation At $7B After Secondary Sale." Crunchbase News, 2023-12-04. news.crunchbase.com/...orn-secondary-sale-fidelity
57. ^1 ^2 ^3 ^4"AI startup CoreWeave closes $650 million secondary share sale, with Cisco joining as investor." CNBC, 2024-11-13. cnbc.com/...oreweave-in-650-million-secondary-sale
58. ^1 ^2 ^3 ^4"CoreWeave Expands Investor Base with $650 Million Secondary Sale." CoreWeave Investor Relations, 2024-11-13. investors.coreweave.com/...default
59. ^1 ^2"CoreWeave acquires Conductor Technologies, enhancing its offerings to VFX studios." Computer Graphics World, 2023-01-25. cgw.com/...-acquires-Conductor-Technologies-enhanc
60. ^"NVIDIA H100 GPUs Set Standard for Generative AI in Debut MLPerf Benchmark." NVIDIA Blog, 2023-06-27. blogs.nvidia.com/...generative-ai-debut-mlperf
61. ^"MLPerf Results: CoreWeave and NVIDIA Showcase Record-Breaking, Cloud Native AI Supercomputer." CoreWeave, 2023-06. coreweave.com/...ing-cloud-native-ai-supercomputer
62. ^1 ^2"CoreWeave to open $1.6B data center in Plano." Community Impact, 2023-07-26. communityimpact.com/...en-16b-data-center-in-plano
63. ^1 ^2"CoreWeave prices IPO lower than expected, and its stock still sags on its debut before recovering." SiliconANGLE, 2025-03-27. siliconangle.com/...ected-40-per-share-valuing-19b
64. ^1 ^2 ^3 ^4 ^5"CoreWeave (CRWV) Historical Quotes." Nasdaq, 2026-10-15. nasdaq.com/...historical
65. ^1 ^2 ^3 ^4 ^5 ^6 ^7 ^8 ^9 ^10"CoreWeave, Inc. Definitive Proxy Statement (Schedule 14A) for the 2026 Annual Meeting." CoreWeave via SEC EDGAR, 2026-04-22. sec.gov/...crwv-20260422
66. ^1 ^2"About Us." CoreWeave, 2026-10. coreweave.com/about-us
67. ^1 ^2"CoreWeave Announces Appointment of Michelle O'Rourke as Chief People Officer." CoreWeave, 2024-10-18. coreweave.com/...e-orourke-as-chief-people-officer
68. ^"Nvidia-backed cloud firm CoreWeave to acquire AI developer platform Weights & Biases." Reuters via Yahoo Finance, 2025-03-05. finance.yahoo.com/...loud-firm-coreweave-080719983
69. ^1 ^2"CoreWeave to Acquire Weights & Biases: Industry Leading AI Developer Platform for Building and Deploying AI Applications." CoreWeave, 2025-03-04. coreweave.com/...ing-and-deploying-ai-applications
70. ^1 ^2 ^3 ^4"CoreWeave and Weights & Biases Announce New Products and Capabilities Helping AI Developers Iterate Faster on Models and Agents." CoreWeave Investor Relations, 2025-06-18. investors.coreweave.com/...default
71. ^1 ^2"Core Scientific shareholders dismiss $9B CoreWeave deal over valuation worries." SiliconANGLE, 2025-10-30. siliconangle.com/...reweave-deal-valuation-worries
72. ^1 ^2"CoreWeave acquires agent-training startup OpenPipe." TechCrunch, 2025-09-03. techcrunch.com/...-agent-training-startup-openpipe
73. ^1 ^2"CoreWeave to Acquire Monolith, Expanding AI Cloud Platform into Industrial Innovation." CoreWeave, 2025-10-06. coreweave.com/...atform-into-industrial-innovation
74. ^1 ^2"CoreWeave Acquires Marimo to Unify the Generative AI Developer Workflow." CoreWeave, 2025-10-30. coreweave.com/...-generative-ai-developer-workflow
75. ^"CoreWeave to offer compute capacity in Google's new cloud deal with OpenAI, sources say." Reuters via Business Reporter, 2025-06-12. business-reporter.co.uk/...penai-sources-say-12879
76. ^"Microsoft to spend $10bn on CoreWeave by end of decade." SDxCentral, 2024-11-04. sdxcentral.com/...bn-in-coreweave-by-end-of-decade
77. ^1 ^2"CoreWeave Signs Multibillion-Dollar Deal With Anthropic, CEO Says." Bloomberg, 2026-04-10. bloomberg.com/...-dollar-deal-with-anthropic-video
78. ^"CoreWeave Lands Perplexity in New AI Cloud Deal, Stock Jumps 5.7% Pre-Market." 24/7 Wall St. via Yahoo Finance, 2026-03-04. finance.yahoo.com/...perplexity-ai-cloud-135930966
79. ^1 ^2"CoreWeave First Cloud Provider to Announce General Availability of NVIDIA GB200 NVL72 Instances." CoreWeave, 2025-02-04. coreweave.com/...y-of-nvidia-gb200-nvl72-instances
80. ^1 ^2 ^3 ^4 ^5"Thousands of NVIDIA Grace Blackwell GPUs Now Live at CoreWeave, Propelling Development for AI Pioneers." NVIDIA Blog, 2025-04-15. blogs.nvidia.com/...ve-grace-blackwell-gb200-nvl72
81. ^"CoreWeave, NVIDIA and IBM Set MLPerf Record with Largest NVIDIA GB200 Blackwell Cluster, Achieving Over 2x Faster Training." CoreWeave, 2025-06. coreweave.com/...achieving-over-2x-faster-training
82. ^"CoreWeave Sets New AI Training Records in MLPerf Training v6.0, Training DeepSeek-V3 in Approximately Two Minutes." CoreWeave, 2026-06-16. coreweave.com/...k-v3-in-approximately-two-minutes
83. ^1 ^2"CoreWeave Leads the Charge in AI Infrastructure Efficiency, with up to 20% Higher GPU Cluster Performance than Alternative Solutions." CoreWeave, 2025. coreweave.com/...rmance-than-alternative-solutions
84. ^1 ^2 ^3 ^4"CoreWeave Kubernetes Service." CoreWeave, 2026-10. coreweave.com/...coreweave-kubernetes-service
85. ^"Slurm on Kubernetes (SUNK)." CoreWeave Docs, 2026-10. docs.coreweave.com/...sunk
86. ^"Straggler detection." CoreWeave Docs, 2026-10. docs.coreweave.com/...straggler-detection
87. ^1 ^2"Pricing." CoreWeave, 2026-10. coreweave.com/pricing
88. ^"CoreWeave homepage (archived copy)." CoreWeave via Internet Archive, 2023-06-03. web.archive.org/...www.coreweave.com
89. ^"NVIDIA Announces DGX Cloud Lepton to Connect Developers to NVIDIA's Global Compute Ecosystem." NVIDIA, 2025-05-19. investor.nvidia.com/...default
90. ^"CoreWeave begins $1.8 billion data center project in Kenilworth, landing first award under new EDA tax credit program." Real Estate NJ, 2025-11-25. re-nj.com/...ward-under-new-eda-tax-credit-program
91. ^"Applied Digital Finalizes Additional 150MW Lease with CoreWeave in North Dakota (Exhibit 99.1)." Applied Digital via SEC EDGAR, 2025-08-29. sec.gov/...ex99-1
92. ^"Applied Digital Completes Phase II Ready for Service at Polaris Forge 1, Fully Energizing First 100 MW Building." Applied Digital, 2025-11-24. ir.applieddigital.com/...e-ii-ready-for-service-at
93. ^"Applied Digital Brings an Additional 75 MW of AI Infrastructure Online at Polaris Forge 1." Applied Digital, 2026-10-02. ir.applieddigital.com/...an-additional-75-mw-of-ai
94. ^"CoreWeave announces $6B+ commitment to AI infrastructure in Pennsylvania." ROI-NJ, 2025-07-16. roi-nj.com/...to-ai-infrastructure-in-pennsylvania
95. ^1 ^2"CoreWeave expands data center footprint in Flexential facilities." Lightwave, 2023-10-12. lightwaveonline.com/...-in-flexential-data-centers
96. ^"Bloom Energy Bringing Fuel Cell Power to CoreWeave Data Center Project in Illinois." Microgrid Knowledge, 2024-07-17. microgridknowledge.com/...enter-project-in-illinoi
97. ^"Blue Owl, Chirisa Technology Parks and PowerHouse Data Centers announce first phase of $5B JV development partnership." IREI, 2024-08-29. irei.com/...phase-of-5b-jv-development-partnership
98. ^1 ^2"CoreWeave's first two UK data centers are now operational." ITPro, 2025-01-14. itpro.com/...o-uk-data-centers-are-now-operational
99. ^1 ^2"CoreWeave Announces £1.5 Billion Commitment to Power UK AI Innovation and Growth Through Sustainable Computing." CoreWeave, 2025-09-16. coreweave.com/...power-uk-ai-innovation-and-growth
100. ^1 ^2"CoreWeave's stock slides on weak guidance even as revenue more than doubles." CNBC, 2025-11-10. cnbc.com/...coreweave-crwv-q3-earnings-report-2025
101. ^"Wells Fargo revamps CoreWeave stock price target for 2026." TheStreet via Yahoo Finance, 2026-05-11. finance.yahoo.com/...mps-coreweave-stock-200700036
102. ^1 ^2"Michael Burry's Latest Warning Could Be Bad News for CoreWeave." The Motley Fool, 2025-11-11. fool.com/...ys-latest-warning-could-be-bad-news-fo
103. ^1 ^2"CoreWeave Secures $100M to Expand NVIDIA HGX H100 GPU Offering, Usher in New Era of AI Innovation." Business Wire, 2022-12-06. businesswire.com/...en
104. ^1 ^2"CoreWeave Turns $50 Million Convertible Notes Into $12.5 Billion AI Jackpot For This Hedge Fund." Benzinga via Yahoo Finance, 2025-10-10. finance.yahoo.com/...million-convertible-013120134
105. ^"CoreWeave, Inc. Quarterly Report on Form 10-Q for the quarter ended March 31, 2025." CoreWeave via SEC EDGAR, 2025-05-15. sec.gov/...crwv-20250331
106. ^"CoreWeave, Inc. Quarterly Report on Form 10-Q for the quarter ended June 30, 2025." CoreWeave via SEC EDGAR, 2025-08-13. sec.gov/...crwv-20250630
107. ^"CoreWeave, Inc. Quarterly Report on Form 10-Q for the quarter ended March 31, 2026." CoreWeave via SEC EDGAR, 2026-05-08. sec.gov/...crwv-20260331
108. ^1 ^2 ^3"CoreWeave Launches the CoreWeave Zero Trust AI Framework, Keeping Customers in Control of Access to Their Models, Data and Runtime Environments." CoreWeave (press release), 2026-10-21. coreweave.com/...els-data-and-runtime-environments
109. ^Higgins, James; Sanders, Corey. "CoreWeave Zero Trust AI Framework: Your Model. Your Data. Your Trust." CoreWeave Blog, 2026-10-21. coreweave.com/...k-your-model-your-data-your-trust