The American Financial Times newspaper said that with the growing demand for chips NVIDIA (Nvidia) in the market artificial intelligenceWall Street entered into a major lending frenzy worth $11 billion to companies known as “new cloud,” which provide high-performance computing cloud services to artificial intelligence startups.
The list of leading lenders – according to the newspaper – includes companies such as Blackstone, Pimco, Carlyle, and BlackRock, which have launched a new debt market that relies on providing huge loans guaranteed by Nvidia chips, which are considered one of the most important resources required in the artificial intelligence market.
Who is the market for?
CoreWave, based in New Jersey, is among the largest companies in the “new cloud” sector and has raised more than $10 billion in loans in the past 12 months alone, including $1 billion from Blackstone and $500 million from Magenter Capital. In addition to loans from JP Morgan, Goldman Sachs, and Morgan Stanley.
The company uses this money to purchase tens of thousands of “NVIDIA” chips, and the number of chips it owns has reached about 45,000 chips, making it the largest operator of “NVIDIA” chips in North America.
According to a Financial Times report, the company’s value has increased significantly from $2 billion to $19 billion within 18 months, and it plans to go public in the first half of 2025.
One of the company’s major investors describes, “CoreWave’s initial success was due to it providing computing power from NVIDIA at a time when artificial intelligence was witnessing an explosion in uses,” adding, “Switching to artificial intelligence at the right moment contributed significantly to raising its value.”
The risks of Nvidia’s dominance
But what raises concerns – according to the newspaper – is the large focus on Nvidia, as many borrowers use its chips as collateral for loans, which makes the market highly vulnerable to any changes in the value of these chips.
Despite the high demand for chips, their prices have fallen in some markets, as the cost of a GPU computing hour fell from $8 to about $2 due to increased supply.
Companies’ reliance on Nvidia also constitutes a concern in the market, as the company is said to grant preferential access to its chips to companies that invest in them, according to what the newspaper reported.
However, Muhammad Siddiq, a director at the company, denies the existence of such preferences, stressing to the Financial Times: “We do not help any customer bypass the queue, and everything is organized in terms of supply.”
Borrowing challenges
Sources explain to the newspaper that “Core Wave” succeeded in obtaining these large loans, due to its contracts with “Microsoft“(Microsoft), which is the largest supporter of the “Open AI” artificial intelligence company, as the company signed a contract with it estimated at more than one billion dollars over several years, which motivated lenders to provide additional financing worth two billion dollars to purchase more Nvidia chips.
A source close to the deal also pointed out to the newspaper the importance of this contract, saying, “The Microsoft contract was essential for financing loans worth two billion dollars to purchase GPU chips.”
In addition to CoreWave, Lambda Labs was able to obtain a $500 million loan from Macquarie, while Cruso raised $200 million from the New York Upper 90 Fund. Last October, Cruso concluded a $3.4 billion deal with Blue All Capital to build a new data center in Texas, serving Oracle and Open AI.
The newspaper explains that the heavy reliance on Nvidia chips is considered a double-edged sword, as it may constitute a large surplus of chips in the market if the demand for artificial intelligence decreases or they are replaced by newer versions.
“All lenders encourage borrowing against chips, but we must remember that chips are assets that decline in value over time, not investment assets that increase in value,” Nat Kobikar of Urso Partners told the newspaper.
Investors’ optimism is based on the increasing future demand in the artificial intelligence sector, as Eric Falk, head of strategy at Maginter, believes: “Predicting demand is difficult, but historically, forecasts underestimate future demand.”
