Cerebras Refiles for IPO; A Real-World Test of Whether “Price Matters Least” in AI Infrastructure

This week The Information reported that Cerebras Systems, the Silicon Valley chip designer built around its wafer-scale computing architecture, has confidentially refiled for a U.S. IPO that could come as soon as April 2026.

If it proceeds, the offering will do more than reopen a paused listing. It will offer public markets a clean test case: how should investors approach a capital-intensive, early-scale AI infrastructure company in the current cycle. Will “price matter least,” as one early backer recently put it, when the strategic opportunity is large enough?

The Path Back to Market

Cerebras initially filed to go public in 2024 but withdrew its S-1 in 2025 amid scrutiny from the Committee on Foreign Investment in the United States (CFIUS) tied to investor G42. The national security review slowed the process and ultimately led the company to reset.

Since then, Cerebras has reshaped its narrative.

In fall 2025, days before the withdrawal, it raised over $1B of capital at an $8.1 billion valuation, double its 2021 Series F valuation of $4 billion. Then in January this year, it announced a major multiyear agreement to provide up to 750 megawatts of AI compute capacity to OpenAI through 2028, and just weeks later raised another $1B at $23B, nearly ~3x the valuation in four months. Also in that time, key peer Groq was acqui-hired by Nvidia for a healthy $20B, validating the emerging space. 

Early Revenue, Large Expectations

The revenue base for Cerebras remains early but is clearly scaling. The prior S-1 disclosed revenue of approximately $25 million in 2022 and $79 million in 2023. First-half 2024 revenue reached about $136 million, implying potential annual revenue north of $200 million, albeit with meaningful lumpiness.

Even under aggressive assumptions—say revenue reaching $500 million in 2025—a ~$23 billion valuation implies more than 40x trailing sales.

For comparison, NVIDIA trades around ~25x trailing EV/sales and the broader semiconductor peer group averages closer to ~11x.

That premium suggests investors are underwriting not just rapid growth, but durable architectural differentiation and sustained participation in large-scale AI training and inference workloads.

The OpenAI Deal: Scale Without Full Clarity

The OpenAI agreement provides headline scale but limited transparency. The widely cited ~$10 billion figure appears to reflect cumulative commitments through 2028. However, disclosure remains sparse.

Spread over multiple years, the implied annual opportunity is meaningful. But beyond just the headline number, the quality of that revenue  also matters to investors.

Capital Intensity Changes the Equation

Unlike asset-light AI software companies, Cerebras operates in a deeply capital-intensive segment of the stack. Scaling wafer-scale systems and delivering compute at the 750MW level requires sustained investment in silicon design, manufacturing partnerships, systems integration, and infrastructure.

For such a growing business, public capital is not just liquidity, it is strategic fuel.

The structural backdrop is supportive. AI training and inference demand remain capacity constrained. Hyperscalers are increasingly open to architectural alternatives as they seek performance gains and supply diversification. Cerebras positions itself not as a marginal chip vendor, but as a differentiated compute architecture.

The question is durability. Competing against NVIDIA’s entrenched hardware-software ecosystem and developer network requires consistent execution, ecosystem development, and economic discipline.

“Price Matters Least” Meets a Real-Time Pricing Engine

This week, Lucas Swisher, co-lead of the growth fund at longtime Cerebras backer Coatue, argued price matters least when a “platform company” controls a strategically scarce asset in a massive market, particularly when that market is pulling them into hyper-growth stage.

That thesis often works in private markets, where capital is patient and optionality is prized.

Public markets, however, are a real-time pricing engine, a function also praised by Lucas in the episode. As seen with recent declines to public SaaS valuation multiples, the stock market will quickly translate shifts in expected terminal value into today’s share price.

Cerebras’ IPO will be an excellent test of what happens when these frameworks converge.

We’re still awaiting the latest financial details that will come when the updated S-1 filing becomes public. That said, if the deal prices successfully at or above its latest private valuation, it could signal continued willingness to fund early, capital-intensive AI platforms primarily on strategic positioning and future potential.

If it prices meaningfully lower or struggles post-IPO, it may suggest that even in AI infrastructure, price still matters. For market watchers, the outcome will say as much about the current IPO environment as it does about Cerebras itself. 

The company’s valuation swing from ~$8 billion to ~$23 billion in less than six months underscores how quickly capital markets can reprice strategic AI assets when a large hyperscale counterparty validates the platform. Whether that repricing proves durable will help us answer not only whether the IPO window is open, but also what public markets are willing to reward in the age of booming AI infrastructure.

To stay up-to-date on our fund and research, fill out the form below.

THEMATIC RESEARCH DISCLOSURE: This article is for informational purposes only and represents the thematic analysis of Prepublic Equity Partners ("PEP"). This content is intended to discuss industry trends and does not constitute investment advice, a recommendation, or a solicitation to buy or sell any securities. PEP is not a registered investment adviser or broker-dealer.

Private company data and "pre-IPO" mentions are based on available market reports and have not been independently verified. Investing in late-stage venture capital involves high risk and illiquidity. PEP or its affiliates may hold financial interests in the companies or sectors discussed, and we reserve the right to trade these positions at any time without notice.