Cerebras Systems priced the year's biggest US initial public offering at $185 a share on Wednesday night, raising $5.55 billion and arriving on the Nasdaq the next morning with an indicated opening price of $350 — a 90% jump that briefly implied a market capitalisation north of US$80 billion. Chief executive Andrew Feldman, speaking live to Bloomberg Technology as the trading desks tried to balance the order book, said the offering had been more than 25 times oversubscribed and that he had been forced to ration allocation across institutional and retail buyers.
"It's the biggest tech, one of the biggest tech IPOs in history, and it's the biggest semi in history," Feldman said. "We couldn't be more proud. This is the culmination of a decade of work, of countless late nights and long weekends. We are enormously proud and excited and ready to get back to work."
The price tape itself was the headline. Cerebras had launched the deal at an indicative range that Bloomberg's Bailey Lipschultz noted was already aggressive. "More than 20 times oversubscribed. We reported before it even launched, there was $10 billion of demand. So these are huge numbers," Lipschultz said. "This is the hottest stock, at least from an IPO perspective this year, bar none." Buy-side sources told Bloomberg they were modelling a path to nearer US$100 billion in market value if the company executed on its inference backlog. "If Nvidia is worth $5 trillion, who's to say that this company doesn't easily trade closer to a hundred billion?" Lipschultz said one investor told him.
What Cerebras sells, and how, is core to the bull case. The company makes a wafer-scale processor the size of a dinner plate — the only one of its kind to reach production — and packages it inside a full vertically integrated supercomputer rather than selling chips to be assembled by third parties. Feldman defended that model in the interview by comparing it to high-performance automotive engineering. "You don't just get 15 or 18 or 20 times faster than the competition because you built a good chip," he said. "You can build a great chip, and the system vendor, the ODM, can nibble away at your performance by not delivering the right amount of power or the right amount of I/O. There's a reason why Nvidia then sought to control the IO, the NVLink — because they didn't want others to nibble away at their performance."
The commercial book underpinning the IPO has also moved beyond the single-customer concentration that delayed Cerebras's first attempt to list two years ago. "In the last four months, we announced a deal with OpenAI that's north of $20 billion dollars for 750 megawatts of compute," Feldman said. "We also announced a major engagement with AWS where our equipment would be deployed in their data centers." He confirmed a binding term sheet with Amazon Web Services was signed and a master agreement is being papered. "I think they are one of the preferred cloud providers for just about every enterprise on Earth," he said. "An opportunity to have your solution embedded in their offering as part of their Bedrock offering — that's a huge win for us."
Margin and scale economics were the other thread investors wanted on the record. Feldman said Cerebras' gross margin sits around 40-41% today, well above Dell's "low teens" on server assembly but below Nvidia's "mid-70s" on chips, and pointed to operating leverage as revenue compounds. "We did half a billion last year in sales. That means we put 250 million in the supply chain. Obviously, that's not an efficient spot. As we grow, we will have more leverage in the supply chain and our cost of goods will come down."
He also told Bloomberg there is room to push pricing. "I think we have an opportunity to increase prices. The demand for fast inference is overwhelming this minute. In the long run we'll be really proud of our gross margins and where they will sort of wash out as we hit scale."
On retail participation, where many late-stage tech IPOs have leaned heavily, Feldman was unapologetic. "We were more than 25x oversubscribed. There were a lot of hard decisions that needed to be made and nobody got what they wanted. And we did our best."
Asked how the $5.5 billion of fresh capital would be deployed, the answer was simple: more capacity. "I think we'll use it to increase capacity. We are excited to bring many new customers on board. There's tremendous demand for what we're doing. I think we can be aggressive on that front."