Wall Street’s frenzy creates an $11 billion debt market for AI groups to buy Nvidia chips

Wall Street’s largest financial institutions have lent more than $11 billion to a group of technology companies based on their ownership of the world’s hottest commodity: Nvidia’s artificial intelligence chips.

Blackstone, Pimco, Carlyle and BlackRock are among the companies that have created a lucrative new debt market over the past year by lending to “neocloud” companies, which provide cloud computing to construction technology groups. AI products.

Neocloud groups like CoreWeave, Crusoe, and Lambda Labs have purchased tens of thousands of Nvidia’s high-performance computer chips, known as GPUs, that are crucial for developing synthetic AI models. Those things Nvidia Chips are also now being used as collateral for large loans.

This trading frenzy has shed light on Silicon Valley’s rampant GPU economy, which is increasingly backed by wealthy New York financiers. However, its rapid growth has raised concerns about the potential for riskier lending, circular financing and Nvidia’s stranglehold on the market. artificial intelligence market.

The $3 trillion tech conglomerate’s chip allocation to neocloud groups has given Wall Street lenders confidence to lend the latter companies billions of dollars to buy more Nvidia chips. Nvidia itself is an investor in neocloud companies and these companies are among their largest customers.

Critics have questioned the current value of collateralized chips as new enhanced versions come to market – or if the current high level of spending on AI begins to recede.

“All the lenders coming in are pushing the narrative that you can borrow against these chips and adding to the frenzy that you have,” said Nate Koppikar, a short seller at hedge fund Orso Partners. need to get involved now.” “But chips are an asset that depreciates, not appreciates.”

Connect CoreWeave

New Jersey-based CoreWeave, the largest neocloud company, began accumulating chips when it launched in 2017 for cryptocurrency mining but pivoted to AI two years later. The company now claims to be the largest privately held supplier of Nvidia GPUs in North America, with more than 45,000 chips.

“CoreWeave’s early success was securing GPU capacity from Nvidia at the exact moment ChatGPT and AI hit the Cambrian explosion,” said an executive at one of its largest investors.

Backed by venture capitalists and Nvidia, its valuation has increased from $2 billion to $19 billion in the past 18 months. The company is planning an initial public offering in the first half of 2025 and could value it even higher, several people close to the company said.

CoreWeave has raised more than $10 billion in debt over the past 12 months from lenders including Blackstone, Carlyle and Illinois-based hedge fund Magnetar Capital. It announced an additional $650 million in credit lines from Wall Street banks including JPMorgan, Goldman Sachs and Morgan Stanley this month.

The debt was secured by CoreWeave’s Nvidia GPUs, and the capital was used to buy thousands more. It plans to have 28 data centers across the US and Asia by the end of 2024, nine times its size at the beginning of last year.

The financial resources mean CoreWeave is extremely leveraged. When it announced its first $2.3 billion in debt financing in August 2023, which included about $1 billion in loans from Blackstone, it had annual revenue of just $25 million, two people close to the company said. USD and negative ebitda of about 8 million USD. One source said revenue has increased to about $2 billion this year.

CoreWeave declined to comment on its finances.

Some of CoreWeave’s largest lenders were persuaded to invest thanks to a large deal they negotiated with Microsoft – biggest supporter of OpenAI — last year will generate more than $1 billion in revenue over several years, several people familiar with the deal said.

“The Microsoft contract is very important,” said a person close to the deal. “They won the contract and said we need $2 billion in GPUs that we can finance.”

Like traditional asset-backed lending, in the event of default, the lender will own the GPUs as well as the contracts – known as power purchase agreements – with the companies that lease them.

CoreWeave’s success in raising large amounts of private debt has prompted more lenders to enter this space. While most neocloud companies are venture-backed, they are uncommon among startups with high capital expenditures, requiring them to turn to the credit markets to fund their ventures. extend.

Macquarie loaned $500 million to Lambda Labs in April, and Crusoe raised $200 million in debt from New York investor Upper90 last year. The Financial Times reported last week that Crusoe is also raising $500 million in equity capital from investors, including Peter Thiel’s Founders Fund.

In October, Crusoe completed a $3.4 billion deal with Blue Owl Capital, an alternative asset management firm, to fund a new data center in Texas that will be leased to Oracle and OpenAI computing capacity.

“The scale of the investment will make a lot more sense when people start to recognize this as the largest capital investment in human history,” said Crusoe CEO Chase Lochmiller.

Contact Nvidia

Neocloud relies heavily on their relationship with Nvidia. For example, CoreWeave may have access to tens of thousands of H100 chips as the chip giant’s “preferred partner,” but its future growth depends on having access to the same Nvidia’s newer Blackwell chips.

Nvidia has denied that it grants priority access to its chips to any customers, including those in which it invests. “We’re not helping anyone get in line,” Mohamed Siddeek, head of Nvidia’s NVentures venture capital unit, told the FT last year.

“It’s a good thing from the lender’s perspective,” one of the bankers said of the debt deals. “They have control over the entire supply chain. That’s a good thing because Nvidia won’t let things get too bad.”

But GPU trading prices in some markets have dropped in recent months. An hour of GPU computing now costs around $2, down from $8 earlier this year.

Even as demand for chips continues to grow, supply has improved as hardware stocks have been resold, while competition for the number of companies building underlying AI models has narrowed.

Several tech giants are developing their own AI chips, while rivals such as AMD is also racing to release its own high-performance GPU to challenge Nvidia’s supremacy.

“A year ago having access to GPUs was like having a golden ticket into Willy Wonka’s factory,” said a senior executive at one of the major lenders CoreWeave. “That doesn’t happen anymore.”

The future value of Nvidia chips being used as collateral for loans has also been questioned. The leases that neocloud signs with technology groups will begin to expire in the next few years, potentially leading to a glut of available chips on the market.

Listed tech giants that have spent billions on AI infrastructure are also under pressure to generate significant profits. In June, David Cahn, partner at venture firm Sequoia Capital, said there was a $500 billion gap between revenue expectations from tech companies building out AI infrastructure and revenue growth. reality in the AI ​​ecosystem.

However, Neocloud lenders are betting on the continued advancement of AI.

“Forecasting demand is extremely challenging, but previous forecasts have underestimated future demand in many sectors,” said Erik Falk, head of strategy at Magnetar. “The urgency here is immense; And it looks like tech companies are ready to invest heavily to stay ahead.”