
Share
CoreWeave built an AI compute empire on a web of hyperscaler contracts and private credit. Now its bonds and shares are sliding as investors price in rising default risk across the leverage that fueled the buildout.
CoreWeave is the clearest case study yet of how the AI infrastructure boom has been financed, and the clearest warning sign of what happens when markets start to question that financing.
The Nvidia-backed cloud provider has stitched together tens of billions of dollars in customer contracts with major technology firms, then borrowed heavily against that revenue to fund the data centers and chips needed to deliver it. This week, that model came under pressure. CoreWeave's shares and bonds slumped as investors reassessed the debt load behind the company's rapid expansion. The cost of insuring CoreWeave against default has climbed to around 1,000 basis points, implying a nearly 63% chance of default within five years, according to LSEG data. A year ago, that probability sat closer to 30%.
The numbers behind the business are not small. CoreWeave carries roughly $50 billion in debt. Microsoft alone accounted for about 67% of its 2025 revenue, up from 62% the year before. That concentration is the thesis and the risk in one line: a handful of hyperscaler relationships underwrite a mountain of borrowing.
CoreWeave's growth has followed a repeatable pattern. Land a large compute contract with a hyperscaler or AI lab, then use that contract as collateral or justification to raise debt and fund the buildout needed to service it.
The Microsoft relationship anchors the model. OpenAI has layered on top of it. In March 2025, CoreWeave signed a deal worth up to $11.9 billion with OpenAI, which also became a CoreWeave shareholder through a $350 million stock issuance. Two more deals followed: $4 billion in May 2025, payable through 2029, and $6.5 billion in September 2025. Together, the OpenAI relationship now totals about $22.4 billion.
Meta has followed a similar trajectory. In September 2025, Meta committed to $14.2 billion in cloud services from CoreWeave through December 2031, with an option to renew. By April 2026, that agreement had expanded to $21 billion, running through December 2032. Anthropic signed on in April 2026 for undisclosed terms, supporting its Claude model family. Nvidia, an early CoreWeave investor since a 2023 Series B round, has agreed to buy any unsold CoreWeave compute capacity through 2032 for $6.3 billion, effectively backstopping demand.
The financing side of the ledger is just as dense. A $2.3 billion raise in August 2023 was led by Magnetar Capital and Blackstone Tactical Opportunities, with Coatue, DigitalBridge Credit, BlackRock, Pimco and Carlyle participating. That was followed in May 2024 by a $7.5 billion facility led by Blackstone, with Magnetar as co-lead and a syndicate that included CDPQ, Eldridge Industries and Great Elm Capital Corp.

Since mid-2025, CoreWeave has leaned on a series of delayed-draw term loans to keep pace with contract obligations: $2.6 billion in July 2025 to accelerate OpenAI service delivery, led by Morgan Stanley and MUFG; $8.5 billion in March 2026, anchored by Blackstone Credit & Insurance; $3.1 billion in May 2026; and $2.6 billion more in August 2026. Morgan Stanley, MUFG, Goldman Sachs and JPMorgan appear repeatedly across these facilities, alongside a wider bank syndicate that includes Wells Fargo, BBVA, Crédit Agricole, SMBC, PNC and Société Générale.
A revolving credit facility, originally $650 million in October 2024, has been expanded twice. It grew to $1.5 billion by May 2025, then to $2.5 billion in November 2025, with maturity pushed out from May 2028 to November 2029. On the bond side, CoreWeave has issued senior notes carrying coupons between 8.5% and 9.75%, alongside three tranches of convertible senior notes at 1.75% and 2.875%, including a $3.5 billion convertible offering in April 2026 and a $3.7 billion convertible offering in September 2026.
Retail-adjacent exposure shows up in fund holdings too. Fidelity Convertible Securities Fund held $39.82 million face value of CoreWeave's 1.75% convertible notes as of August 31, 2026, representing 1.64% of its portfolio. Nomura High Income Fund held about $11.68 million across four CoreWeave bond issues as of the same date. These are not outsized positions, but they illustrate how CoreWeave credit has worked its way into mainstream fixed-income portfolios.
The coupons on those senior notes, several north of 9%, tell their own story. That is distressed-adjacent pricing for a company whose backers include Nvidia, Microsoft and some of the largest names in private credit. High-yield borrowing of this kind is typical for capital-intensive infrastructure buildouts. It is less typical when default insurance is pricing in a near coin-flip outcome over five years.
CoreWeave's financing structure is not inherently flawed. Contract-backed lending against hyperscaler revenue is a reasonable model when the counterparties are Microsoft, Meta and OpenAI, and when demand for AI compute keeps growing. The risk is concentration and leverage moving together. Two-thirds of revenue tied to one customer, $50 billion in debt, and credit default swap pricing that has more than doubled in a year is not a crisis signal on its own, but it is a flashing amber light.
Investors holding CoreWeave paper, directly or through funds like Fidelity's convertible fund or Nomura's high-income fund, are making a bet that AI infrastructure demand stays durable enough to service this debt load for years. The broader market reaction this week suggests that bet is no longer unanimous. Private credit's record fundraising and deployment pressure have made deals like CoreWeave's attractive to lenders chasing yield. Whether that yield compensates for the risk now being priced into credit default swaps is the question the market is starting to ask out loud.
Tags
Original Sources
CoreWeave's AI contracts and the lenders financing its build-out
↗ https://www.reuters.com/legal/transactional/coreweaves-ai-contracts-lenders-financing-its-build-out-2026-10-15
About the author
Marcus began tracking AI's market implications in 2016, noticing AI-related patent filings accelerating ahead of earnings upgrades before most of the sell-side had caught on. A former fixed-income quantitative analyst, he spent two decades building models that priced risk across emerging markets before pivoting to cover the economic impact of AI full-time. His writing translates opaque technical developments into clear risk/reward terms — and he's rarely diplomatic about the gap between AI valuations and underlying fundamentals. He believes most market participants still underestimate AI's long-run deflationary effect on knowledge work.
More from The Analyst →This Week's Edition
16 October 2026
29 articles
Related Articles
Related Articles
More Stories
© 2026 Cedar & Bloom. All rights reserved.