CoreWeave CEO Michael Intrator unpacked the cloud computing company’s first day on the market in a Friday interview with CNBC’s Jim Cramer, defending its decision to raise a hefty debt load.
“The debt is the engine, it’s the fuel for this company,” Intrator said. “We go out, we find great contracts with great counterparties that need massive scale computing to drive their business, and then we go ahead and we go back to our syndicate of lenders, and they give us the debt to stand up the clusters that will deliver revenue to the company.”
The IPO debuted on a tough day for the indexes, especially for tech stocks, whose losses helped the Nasdaq Composite plunge 2.7%. CoreWeave, which sells artificial intelligence technology in the cloud, opened at $39 and closed flat at $40, raising $1.5 billion in its share sale. It’s the biggest tech IPO in the U.S. since 2021, even as the company set its share price at $40, lower than the previously expected range of $47 to $55. Intrator told CNBC the lower pricing was “where the buying interest was” and claimed there are “a lot of headwinds in the macro.”
CoreWeave has raised almost $13 billion in debt, CNBC reported, much of which is for GPUs in the company’s leased data centers in the U.S. and abroad. Intrator told Cramer that debt on the balance sheet is offset by a larger revenue contract.
Ahead of its market debut, the AI outfit bought 250,000 of Nvidia‘s graphics chips. Many of them are from the Hopper generation, models that were scarce and in demand over the past few years. There are concerns that these products will lose relevancy in the quickly-advancing world of AI – and Nvidia has already started shipping out the model’s successor, Blackwell.
Intrator refuted those concerns and highlighted the company’s recent deal with OpenAI for just under $12 billion, which he said was executed for five years with two one-year extensions. Deals like this, he said, indicate that companies believe the infrastructure will have value far into the future.
“Those same buyers will come back, they will buy new infrastructure that is the most cutting edge for their next models,” he said. “And then they’ll take this, this earlier infrastructure and use it for other use cases in their company that requires really large bulk compute.”
