At $56.4B fully diluted, Cerebras enters the market with a $20B OpenAI capacity deal and a customer concentration that shifted from G42 to a single UAE university.
Cerebras Systems priced its initial public offering at $185 per share Wednesday, above the already-raised expected range, pulling in at least $5.55 billion and setting a fully diluted valuation of $56.4 billion. The pricing lands during a silicon renaissance. Intel, Advanced Micro Devices, and Micron are each up more than 80% in the past month. Investors have been spreading their chip bets from Nvidia to the wider universe of semiconductor companies benefiting from the AI boom. The deal is the largest U.S. tech IPO since Uber raised about $8 billion in 2019, eclipsing Snowflake's $3.8 billion offering in 2020. The stock will trade on the Nasdaq under ticker CBRS.
For traders, the print is not just a headline number. It reflects a business that has pivoted from selling hardware systems to a cloud-service model, swapped one concentrated customer risk for another, and anchored its growth narrative to a massive OpenAI capacity deal.
Cerebras initially filed to sell 28 million shares at $115 to $125 per share on May 4. A week later, it bumped the offering to 30 million shares and lifted the expected range to $150 to $160. The final $185 price signals demand that exceeded even the revised range.
The sequence of upward revisions tells a story of a heavily oversubscribed book. Underwriters rarely raise both share count and range unless they see enough orders to absorb the additional supply without breaking the price. The $185 final print, 15.6% above the top of the revised range, suggests the deal could have priced even higher. Banks typically leave some money on the table to ensure a positive first-day trade.
Co-founder and CEO Andrew Feldman now holds a stake worth about $1.9 billion at the IPO price. That level of insider skin-in-the-game aligns management with shareholders. It also creates a large overhang when lock-up periods expire. Traders will track the lock-up expiry date as a potential volatility event.
Cerebras started life selling physical systems built around its Wafer Scale Engine 3 chips. The company has since shifted its focus toward providing a cloud service based on those chips. This puts it in direct competition with cloud providers Google, Microsoft, Oracle, and CoreWeave, all listed as competitors in its prospectus.
A cloud-service model generates recurring revenue. It typically carries lower gross margins than hardware sales in the early years. The company must build out data center capacity before the recurring revenue scales. Cerebras claims its chips offer speed and price advantages over Nvidia's GPUs. If those claims translate into real workload migration, the cloud business could scale faster than hardware ever could. The risk: cloud competitors with deeper balance sheets can subsidize AI compute to defend market share.
Nvidia's Alpha Score of 71/100 (Moderate) on AlphaScala reflects a mature AI franchise with dominant data-center revenue. Cerebras enters as a challenger with a differentiated architecture. The Wafer Scale Engine integrates an entire wafer into a single chip, reducing communication latency between cores. That architecture advantage is the technical bet. The commercial bet is whether enterprises will adopt a non-CUDA ecosystem at scale. Nvidia's current price is $225.83, and its stock page is NVDA stock page.
Cerebras has been dogged by customer concentration risk since its first IPO filing in September 2024. That filing was withdrawn after heavy scrutiny of the company's reliance on G42, a Microsoft-backed AI firm in the United Arab Emirates. In the refreshed prospectus, Cerebras disclosed that G42 accounted for 24% of revenue last year, down from 85% in 2024.
The reduction in G42 dependence is real. The concentration simply migrated. Mohamed bin Zayed University of Artificial Intelligence (MBZUAI), also in the UAE, accounted for 62% of revenue last year. A single academic institution representing nearly two-thirds of sales is an unusual risk factor for a company valued at $56.4 billion. Any change in that relationship, whether from budget cycles, geopolitical shifts, or technology choices, would hit the income statement hard.
Both G42 and MBZUAI are based in the UAE, which sits at the intersection of U.S. technology export controls and Chinese AI ambitions. The U.S. government has tightened restrictions on advanced chip exports to certain Middle Eastern countries. Cerebras will need to navigate those controls carefully. Any new export license requirements could disrupt the revenue base.
In January, Cerebras signed a deal with OpenAI worth over $20 billion for 750 megawatts of computing capacity. That single contract dwarfs the company's historical revenue and provides a multi-year demand anchor that few pre-revenue cloud startups can claim.
To put 750 megawatts in context, a large hyperscale data center campus might draw 100-200 megawatts. Cerebras is effectively committing to build and operate multiple campuses' worth of AI compute for one customer. The capital expenditure required is enormous. The company will likely need to raise additional debt or equity to fund the buildout. The IPO proceeds are a down payment on that infrastructure.
During Elon Musk's trial against OpenAI, testimony revealed that OpenAI considered merging with Cerebras in 2017. Greg Brockman, OpenAI's co-founder and president, wrote in an email:
"Exclusive access to Cerebras hardware would give OpenAI an overwhelming hardware advantage over Google."
Brockman held about 78,000 shares of Cerebras at the end of 2025, worth $14.4 million at the IPO price. The historical link adds a layer of strategic logic to the capacity deal: OpenAI has been evaluating Cerebras hardware for nearly a decade.
At $56.4 billion fully diluted, Cerebras is pricing at a premium that assumes the OpenAI deal converts into sustained, high-margin revenue and that customer concentration risk fades over time. Nvidia trades at roughly 20 times forward sales. Nvidia has a diversified customer base and $60 billion in quarterly data-center revenue. Cerebras is a story stock with a single-digit revenue base and a single-customer anchor.
Pre-IPO investors hold significant paper gains. Fidelity's stake is valued at about $3.8 billion, Benchmark at $3.3 billion, Foundation Capital at $2.8 billion, and Eclipse at $2.5 billion. These venture investors will eventually distribute shares to limited partners or sell into the market. The pace and timing of those distributions will influence the stock's float and trading dynamics in the first year.
Bloomberg reported, citing unnamed sources, that weeks before the IPO, Arm and SoftBank both attempted to acquire Cerebras. Cerebras declined to comment. An acquisition offer from strategic buyers suggests that the company's technology is viewed as a scarce asset in the AI infrastructure stack. The fact that Cerebras chose to proceed with the IPO instead indicates that management and the board believe the public market will assign a higher value than what strategic buyers offered.
Cerebras gives public-market investors a new way to bet on non-Nvidia AI silicon. The stock will likely trade as a high-beta name, sensitive to every AI demand headline and every Nvidia earnings print. For traders building a semiconductor watchlist, CBRS adds a pure-play alternative to the NVDA stock page and the broader stock market analysis landscape. Microsoft (MSFT stock page) is both a competitor and a backer of G42, adding another layer of interconnected exposure.
Nvidia's current price of $225.83 and Alpha Score of 71 reflect a mature franchise. Cerebras is the opposite: early-stage, concentrated, and betting on architectural differentiation. If Cerebras executes on the OpenAI buildout and diversifies its customer base, it could carve out a meaningful niche. If the buildout hits delays or the MBZUAI relationship shifts, the valuation compression would be severe.
Traders approaching the CBRS open should focus on three variables: the first-day price relative to $185, the volume of shares traded as a percentage of the float, and any lock-up expiry disclosures in the prospectus. A strong open above $200 with high volume would confirm institutional demand. A fade back toward $185 on light volume would suggest that the deal was priced to clear, with limited aftermarket support. The OpenAI capacity deal provides a fundamental backstop. The customer concentration keeps the risk elevated.
Drafted by the AlphaScala research model and grounded in primary market data – live prices, fundamentals, SEC filings, hedge-fund holdings, and insider activity. Each story is checked against AlphaScala publishing rules before release. Educational coverage, not personalized advice.