If Cerebras can grow beyond its initial beneficiary, G42, this IPO will be a rocket ship. If not,…
It may be hard to believe, but there hasn’t been a single IPO for an AI hardware company. Period. Not One. Why? Because none of the startups have yet reached the escape velocity needed to stand alone again like Nvidia. That may be about to change. Cerebras has filed with the SEC to offer shares to the public; Their IPO could happen later this year. Let’s dive in.
Cerebras co-founder and CEO Andrew Feldman plans to take his company public this fall.
OK, who is Cerebras again?
If you follow me, you know that I’ve long been a fan of Cerebras Systems, their CEO Andrew Feldman, whom I’ve known for 12 years, and their Wafer Scale Engine (WSE) approach to AI computing. The idea is so straightforward and so compelling. Instead of splitting a wafer into chips, packaging those chips with high-bandwidth memory, installing them on a board in a computer, and then recombining hundreds of chips to talk to each other over a network expensive, why not just interconnect the chips resident on the wafer? This is wafer-scale computing, which I think will become a big trend over the next 5 years. Oh, and they don’t need HBM because Cerebras use on-chip SRAM supplemented by a stream-weighted server. Yes, it’s hard to power and cool an entire wafer, and you have to deal with inevitable wafer defects, but the resulting math is too compelling not to succeed. That is, if you get the software right.
Why an IPO, and why now?
The timing can be good. Cerebra’s revenue grew more than 14 times to $136.4 million in the first half of 2024, and the company’s net loss fell from $77.8 million to $66.6 million. This makes a good background to try to raise more money to accelerate growth. But instead of turning to (expensive) venture capital, co-founder and CEO Andrew Feldman thinks the public markets are ready for a generative AI offering, and he’s going for it.
Customer focus is a problem
While Nvidia has a similar problem, with four customers accounting for over 40% of revenue last quarter, Cerebra’s customer focus may be a matter of where they are in their lifecycle. It takes a lot of selling and convincing to bring a potential customer to the table, especially if you’re another startup with a slide deck. This IPO could allow Cerebras to find more G42s flying under the hyperscaler’s radars.
Cerebra’s prospectus said the G42 represented 87 percent of its revenue in the first half of the year. G42 has committed to buy $335 million in Cerebra shares by April next year, which will give it a stake of more than 5 percent. G42 and its affiliates are already buying and building a monster 36-exaflop distributed supercomputer with Cerebras.
As an analyst, investors often ask me how Cerebras can grow, given that their biggest potential customer is already building their own AI chips and of course buying hundreds of thousands of Nvidia GPUs. I ask them if they ever heard of G42 before they started buying Cerebras WSE, and I claim that there are a dozen more G42s being funded to build local AI factories like sovereign AI farms and tier 2 public clouds. Maybe they just need permission to buy.
conclusions
While we’ll have to wait to see the market’s reception to Cerebra’s planned IPO, we’d be willing to bet it will be strong. Stock market investors are anxious to invest in an alternative to Nvidia, and many consider AMD to be overvalued. The AI market is also anxious to have an alternative they can decide on for AI deployment, and Cerebra’s value proposition is extremely compelling for both training and termination.
If they can pull off a successful IPO, it can help them grow from a small base; many enterprises will not buy from startups. And nothing breeds success like success.
FINDINGS: This article expresses the opinions of the author and
should not be taken as advice to buy or invest in the companies mentioned. Cambrian-AI Research is fortunate to have many, if not most, semiconductor firms as our customers, including Blaize, BrainChip, Cadence Design, Cerebras, D-Matrix, Eliyan, Esperanto, GML, Groq, IBM, Intel, Micron, NVIDIA, Qualcomm Technologies, Si-Five, SiMa.ai, Synopsys, Ventana Microsystems, Tenstorrent and many investment clients. We have no investment positions in any of the companies mentioned in this article and do not plan to initiate any in the near future. For more information, please visit our website at https://cambrian-AI.com.