Cerebras Systems (CBRS): The AI Chip Challenger That Just Shook Wall Street

The Dinner-Plate Chip That's Taking On Nvidia

In the fiercely competitive world of AI semiconductors, a Silicon Valley upstart just made one of the boldest entrances in recent memory. Cerebras Systems — the company behind a chip literally the size of a dinner plate — priced its IPO at $185 per share on May 13, 2026, above its already-elevated range of $150–$160, raising a staggering $5.55 billion and valuing the company at approximately $56.4 billion on a fully diluted basis. It is, by any measure, the biggest IPO of 2026 so far, and the largest U.S. tech IPO since Snowflake's blockbuster debut in 2020.

When shares opened on the Nasdaq the following morning under ticker CBRS, they launched at $350 — nearly double the IPO price — and rocketed to an intraday high of $385 before closing at $311.07 on its first day. Demand for the offering exceeded available shares by more than 20 times, a testament to the feverish appetite investors have for anything AI-related right now.

What Exactly Does Cerebras Build?

Founded in 2016 and headquartered in Sunnyvale, California, Cerebras is not your typical chip company. Its flagship product — the Wafer Scale Engine 3 (WSE-3) — is unlike anything else in the industry. While conventional AI chips (like Nvidia's GPUs) are small dies cut from a silicon wafer, Cerebras uses the entire wafer as a single chip, packing roughly four trillion transistors onto one piece of silicon.

The result? A chip that is 58 times larger than any chip previously built, and according to CEO Andrew Feldman, more than 15 times faster than the competition when it comes to AI inference — the real-time work of responding to prompts.

"We built a chip the size of a dinner plate. It's 58 times larger than any chip previously built," — Andrew Feldman, CEO of Cerebras Systems

This speed advantage is particularly compelling for AI inference workloads, where latency matters enormously. Rather than stitching together thousands of GPUs in a data centre cluster, Cerebras offers a single, massively parallel alternative.

Revenue Momentum — and the Concentration Problem

Cerebras' growth story is genuinely impressive. Revenue surged 76% in 2025 to $510 million, up from $290 million in 2024 and a mere $25 million back in 2022. The company also swung from a net loss of $482 million in 2024 to net income of $238 million in 2025 — though analysts note that much of that profit reflects a one-time accounting gain, and the business is still running at an operating loss on a recurring basis.

The elephant in the room, however, is customer concentration. In 2025, approximately 86% of Cerebras' revenue came from just two UAE-linked customers: Microsoft-backed G42 (24% of revenue) and the Mohamed bin Zayed University of Artificial Intelligence (62% of revenue). This heavy reliance on Middle Eastern clients had previously derailed Cerebras' first IPO attempt — the company originally filed to go public in September 2024 but withdrew its submission over a year later amid regulatory scrutiny.

The Game-Changing Deals: OpenAI & AWS

What changed the narrative — and arguably made this IPO possible — were two landmark partnerships announced in 2026.

In January 2026, Cerebras signed a multi-year compute agreement with OpenAI worth over $20 billion at full expansion, covering 750 megawatts of AI inference capacity, expandable to 2 gigawatts by 2030. OpenAI has already launched its first AI model running on Cerebras chips. Then in March 2026, Amazon Web Services (AWS) signed a binding term sheet to deploy Cerebras systems inside its own data centres.

These deals gave investors the revenue visibility they needed. The OpenAI contract alone transforms Cerebras from a UAE-dependent hardware vendor into a serious cloud inference player — going head-to-head with Google, Microsoft Azure, Oracle, and CoreWeave.

Acquisition Overtures — and the Road Not Taken

The IPO drama had one more twist. Just weeks before listing, both Arm Holdings and SoftBank reportedly made last-minute acquisition attempts for Cerebras — both of which were declined. The company chose independence and the public markets instead.

It was also revealed during Elon Musk's trial against OpenAI that back in 2017, OpenAI itself had considered merging with Cerebras, viewing the chip company as a potential asset in the pursuit of artificial general intelligence (AGI).

Should Investors Be Excited — or Cautious?

The post-IPO euphoria cooled quickly. On the Friday after its debut, CBRS fell roughly 10%, a reminder that IPO pops are often followed by painful corrections. History is not particularly kind to IPO buyers: research from University of Florida finance professor Jay Ritter shows that newly public companies have underperformed comparable firms by an average of 3.6% per year over their first five years, with IPOs since 2010 showing an even sharper first-year gap of ~9 percentage points.

The Snowflake parallel is instructive: investors who bought at Snowflake's first-day close are still underwater today, despite the company growing revenues many times over. Arm Holdings followed a similar pattern — a strong pop, then months of sideways trading.

At current prices, CBRS trades at over 130 times sales — a valuation that prices in near-perfection. The risks are real: customer concentration remains severe, operating losses persist, and the competitive moat against Nvidia (which is also investing heavily in inference speed) remains to be proven at scale.

Yet the bull case is equally compelling. The OpenAI and AWS deals provide genuine, contracted revenue visibility. If Cerebras can diversify its customer base and grow into its valuation, it could be one of the defining tech stories of the decade. The AI chip market is not a winner-takes-all game — there is room for challengers.

The Bottom Line

Cerebras is a genuinely fascinating company: technically audacious, commercially ambitious, and now very publicly accountable. Its wafer-scale approach to AI chips is a real technological differentiator, and its partnerships with OpenAI and AWS signal that the industry's biggest players are taking it seriously. But at 130x sales, with concentrated customers and widening operating losses, this is a stock for investors with high conviction — and high risk tolerance.

The dinner-plate chip has arrived on Wall Street. Whether it becomes the main course or just an appetiser remains to be seen.