Remember when Nvidia was the only name anyone could say out loud when AI chips came up? Every conversation, every earnings call, every breathless tech headline — Jensen Huang’s company had a stranglehold on the narrative. That was the whole story. Then quietly, in a lab in Sunnyvale, California, a company called Cerebras started building something different. Now it’s filing for an IPO, and the numbers it’s bringing to that conversation are hard to ignore.
Cerebras Systems has officially filed for a public offering, targeting roughly $2 billion in fresh capital. For a chip startup that most casual tech observers couldn’t have named two years ago, that’s a serious statement of intent. But unlike a lot of IPO stories I’ve covered, this one actually has receipts.
The Numbers Are Real
Let’s talk about what Cerebras actually reported for 2025, because this is where things get interesting. The company posted $510 million in revenue with a net income of $87.9 million on a GAAP basis. Strip out certain one-time items and that net income figure climbs to $237.8 million. Compare that to a net loss of $484.8 million the year prior, and you’re looking at a swing that would make most CFOs emotional.
Profitability in the AI chip space, at this stage, is not a given. Most companies burning through cash to compete with Nvidia are doing exactly that — burning. Cerebras showing up to its IPO roadshow with actual profit is a different posture entirely. It doesn’t guarantee anything, but it does mean the story isn’t purely speculative.
The OpenAI Deal Changes the Calculus
The timing of this IPO filing is not accidental. Earlier this week, Cerebras sealed a $20 billion deal to supply OpenAI with servers powered by its chips over the next three years. That contract includes Cerebras making 250 megawatts of compute available annually between 2026 and 2028, with OpenAI holding the option to purchase an additional 1.25 gigawatts on top of that.
That is not a pilot program. That is not a proof of concept. That is a multi-year, multi-billion dollar commitment from the most prominent AI company on the planet. If you’re an institutional investor trying to decide whether Cerebras has a real customer base or just a good pitch deck, that contract answers the question pretty directly.
Filing for an IPO right after locking in a deal of that scale is smart sequencing. You go public when your story is at its clearest, and right now Cerebras has a clear story.
What I’m Actually Watching
Here’s where I put on my skeptic hat, because that’s what this site is for.
- Customer concentration is a real risk. If a massive chunk of your revenue is tied to one relationship with OpenAI, your fortunes are linked to theirs. OpenAI is not a stable, boring enterprise customer — it’s a company in the middle of its own intense competitive and financial pressures.
- Nvidia isn’t standing still. The reason Cerebras has carved out space is partly technical differentiation and partly timing. Nvidia’s roadmap doesn’t pause for anyone, and the window for alternatives to establish themselves is not infinite.
- The IPO market in 2026 is not the IPO market of 2021. Investors are more demanding, valuations are more scrutinized, and the appetite for “trust us, the growth is coming” narratives has cooled significantly. Cerebras is coming in with real revenue and real profit, which helps, but the $2 billion target still requires the market to be receptive.
My Take
I’ve reviewed a lot of AI companies that talk a big game and deliver thin results. Cerebras is not that. The financials are solid, the OpenAI deal is concrete, and the IPO timing shows strategic awareness rather than desperation.
That doesn’t mean this is a guaranteed win. The chip space is brutal, the competitive pressure is real, and going public introduces a whole new layer of scrutiny and quarterly pressure that private companies don’t face. But if you’re asking whether Cerebras has earned the right to be taken seriously as a public company — based on what we know right now — the answer is yes.
The “remember when Nvidia was the only game in town” era is not over. But it’s getting more complicated, and Cerebras is one of the reasons why.
🕒 Published: