New Jersey-based AI and cloud computing company CoreWeave made its market debut on Friday to little applause.
The stock was poised for success by all means: It was the biggest tech IPO since software company UiPath (PATH-3.74%) in 2021, and a rare IPO of a pure-play AI company.
Plus, it was backed by market-favorite tech giant Nvidia (NVDA-1.83%), which ended up having to anchor the IPO last minute with a $250 million order. But things didn’t pan out as expected.
The company had initially targeted a $2.7 billion raise but ended up coming in at $1.5 billion instead. The shares priced at $40, well below the $47 to $50 price range expected previously, and started trading on Friday at $39. On Monday, the stock’s second day of trading, the shares were down almost 10% in midday trading.
CoreWeave is a cloud computing company that provides GPU infrastructure to major tech names like Microsoft (MSFT-1.36%) and Meta (META-0.28%). The company primarily uses Nvidia’s GPUs to run these data centers. Due to the costly nature of the business, CoreWeave relies on accumulating debt to build its capacity, relying on future demand for AI computing power.
CoreWeave is considered a pure-play AI company because the company’s success bets directly on the future of AI demand growth. Most pure-play AI companies — like OpenAI, which kicked off the AI craze in 2022 with ChatGPT — are privately held and don’t publicly disclose finances, which makes it tough for investors to have a gauge on the industry as a whole. CoreWeave’s market debut was seen by many as a way to understand how the AI trade was going amidst rising fears of an AI bubble.
Some industry experts think the stock’s disappointing IPO should be an indicator that the hype is subsiding.
“[If] a year ago CoreWeave had gone public, the words ‘AI’ and ‘Nvidia’ would have been enough to carry it over the line. The fact that it’s not able to do it now is an indicator that the momentum of just trading based on the words AI or Nvidia are not working anymore,” NYU Stern School of Business professor of finance Aswath Damodaran told CNBC last week, adding that he thinks that’s healthy and that AI has been hyped a bit too much.
“The AI product and service business, which ultimately is what has to pay for all of this, has not taken off in any substantial way. I mean, I’m hard pressed to think about any company making significant money from the AI product and service business,” Damodaran said.
For its part, OpenAI expects revenue to more than triple by the end of this year, according to revenue expectations reported by Bloomberg last week, but the company doesn’t anticipate to be cash-flow positive until 2029.
Market skepticism of the AI trade began late last year, Damodaran said, and accelerated when Chinese AI startup DeepSeek rattled markets by unveiling a reasoning model that rivaled OpenAI’s for less cost and energy earlier this year.
What happened last week with the CoreWeave IPO, Damodaran says, is the market recognizing that “too much money” has been spent “too far in advance of the actual evidence that the market exists.”
“I think we’re at that moment in the AI business where we’re looking for some evidence that AI product and service business will be trillions of dollars rather than hundreds of millions of dollars,” he said.
Not all experts agree that the IPO’s disappointment has a direct correlation with the power of the AI trade, though.
“While the CoreWeave IPO clearly did not go well, it should not be taken as an indication of the strength of the market for AI,” D.A. Davidson analyst Gil Luria told Quartz.
Luria thinks that CoreWeave is “a little more than a highly leveraged off balance sheet arrangement for Nvidia,” pointing to Nvidia’s decision to bail out the company’s IPO. CoreWeave’s leveraged spending helps fuel Nvidia’s GPU sales without putting any direct financial risk on Nvidia.
In an initiation note last week, Luria said CoreWeave’s debt-based business model can only be sustainable as long as demand for AI grows “faster than hyperscalers are able to build data centers,” making CoreWeave a risky investment as is.
Regarding speculations of whether or not AI is in a bubble, Luria makes an analogy to the 2008 financial crisis, in which collateralized debt obligations (CDOs) — particularly, subprime mortgages — played a huge role in creating a housing bubble.
“If AI was a CDO, we see CoreWeave as the sub-prime tranche,” Luria wrote. D.A. Davidson currently has a neutral rating on the stock.
“The importance of AI to the future path of technology transcends any one company, and is definitely bigger than CoreWeave,” Luria told Quartz.
The post Does CoreWeave’s disappointing IPO signal an AI bubble? appeared first on Quartz.
