On May 14, 2026, artificial intelligence chip manufacturer Cerebras Systems officially debuted on the Nasdaq Global Select Market under the ticker symbol CBRS. Headquartered in Sunnyvale, Silicon Valley, Cerebras completed its IPO at an issue price of $185 per share, offering 30 million Class A common shares and raising a total of $5.55 billion. This marks the largest IPO in the US tech sector since Uber’s listing in 2019. On a fully diluted basis, Cerebras’s IPO valuation reached $56.4 billion. On its first trading day, CBRS opened at $350, surged to an intraday high of $386, with gains reaching 108% and triggering circuit breakers, and ultimately closed at $311.07, up 68.15% for the day. In the following sessions, the stock pulled back, dropping over 10% the next day, and has now entered the initial phase of analyst coverage. Institutions including Citi, Morgan Stanley, and Barclays have issued buy ratings, with target prices ranging from $250 to $340.
The IPO was jointly led by Morgan Stanley, Citigroup, Barclays, and UBS Investment Bank as bookrunners, with Mizuho Financial Group and TD Cowen acting as co-bookrunners, forming a top-tier Wall Street syndicate. Institutional demand exceeded available shares by more than 20 times, prompting Cerebras to raise both the price range and share count twice during the roadshow.
Cerebras’s core highlight lies in its wafer-scale engine technology. This time, Cerebras returns to the public market with stronger commercial validation and overwhelming oversubscription from Wall Street.
From Triple Price Hikes to First-Day Circuit Breakers: A Complete Cerebras IPO Data Review
Cerebras’s IPO followed a classic trajectory of price escalation. Initially, the company planned to offer 28 million shares at $115–$125 per share, implying a valuation of about $26.6 billion. The price range was then raised to $150–$160, with the share count increased to 30 million. On the eve of pricing, the issue price was pushed up again to $185 per share. In less than two weeks, the issue price jumped from the $115 range to $185, reflecting intense institutional demand for AI chip stocks.
On its debut, CBRS opened at $350, up 89.2% from the issue price, and the fully diluted market cap briefly surpassed $100 billion. The stock quickly rallied to $385, up 108%, triggering trading halts. It closed at $311.07, a 68% gain over the issue price.
Afterward, the stock entered a correction phase. As of June 8, 2026 (US market close), CBRS (Cerebras Systems Inc., NASDAQ:CBRS) traded at approximately $237.83, up about 28.6% from the $185 issue price but down roughly 38.4% from the intraday high of $386 on its first day.
With the post-IPO analyst quiet period ending, several Wall Street firms began coverage in early June: Citi initiated coverage with a "Buy" rating and a 12-month target of $340, projecting Cerebras to capture 40–50% of the high-speed inference market; Morgan Stanley rated it "Overweight" with a $250 target; Barclays also rated it "Overweight" with a $280 target; UBS gave a similarly positive outlook. The consensus target price is around $295, offering about 23% upside from current levels.
Wafer-Scale Chip Technology: How WSE-3 Challenges NVIDIA
Cerebras’s technology fundamentally diverges from traditional GPU manufacturers. NVIDIA’s approach relies on clusters built from many small chips interconnected at high speed, while Cerebras’s wafer-scale integration uses an entire 12-inch silicon wafer as a single processor. The third-generation Wafer-Scale Engine, WSE-3, is manufactured using TSMC’s 5nm process, with a chip area of 46,225 mm², over 4 trillion transistors, 900,000 AI compute cores, 44GB of on-chip high-speed SRAM, and memory bandwidth of 21PB/s. By comparison, NVIDIA’s H100 measures just 814 mm²—WSE-3 is 57 times larger, its on-chip memory is about 880 times greater, and its bandwidth is roughly 7,000 times that of the H100.
Integrating logic and memory on the same wafer eliminates the communication bottlenecks found in traditional GPU clusters, where data transfer between chips can limit performance. For many AI training and inference workloads where memory bandwidth is the bottleneck, Cerebras’s architecture delivers significant performance advantages. According to the prospectus, for inference tasks with open-source models like Llama 3.2 70B, Cerebras’s CS-3 system outperforms eight-way NVIDIA H100 HGX nodes by 8–22 times, averaging about 13.2 times faster. Compared to NVIDIA’s latest DGX B200 Blackwell system, Cerebras claims the CS-3 runs 21 times faster, uses two-thirds less power, and costs about 67% less.
CEO Andrew Feldman once described the company’s first two product generations as "technical rehearsals," noting that the explosive growth in AI inference demand is the true market entry point for Cerebras. The company’s technical advantages don’t apply to all AI workloads. Its 44GB on-chip SRAM can’t independently handle today’s mainstream large language models with over a trillion parameters, but with the MemoryX external memory technology, it can scale up to models with 24 trillion parameters, albeit with narrower speed advantages. This means Cerebras’s strongest value lies in high-speed inference scenarios sensitive to latency and throughput, rather than pre-training ultra-large models.
Financial Fundamentals: 76% Revenue Growth and Customer Concentration Risks
Cerebras’s revenue growth has been steep. According to its prospectus, revenue rose from $24.6 million in 2022 to $78.7 million in 2023, then to $290.3 million in 2024, and reached $510 million in 2025—a 76% year-over-year increase. Over four years, revenue grew more than 19-fold. However, this rapid growth comes with highly concentrated customer structure. In 2025, two Middle Eastern entities, G42 and MBZUAI, contributed about 86% of Cerebras’s revenue. This concentration was repeatedly highlighted during the IPO roadshow and was a key reason for the company’s previous withdrawal of its IPO application—in 2024, when Cerebras first applied to go public, CFIUS launched a national security review of its relationship with G42. Although the review was eventually cleared, it underscored the geopolitical risks inherent in Cerebras’s revenue structure.
Profitability shows a marked gap between accounting and operational reality. Under US GAAP, Cerebras posted a net profit of $238 million in 2025, reversing a $482 million net loss in 2024. However, GAAP profits mainly came from a non-cash debt cancellation gain. Excluding one-off items, the company remains at a non-GAAP net loss of about $75.7 million.
Gross margin in 2025 was 39%, down from 42.3% in 2024, mainly due to increased cloud infrastructure spending. Hardware accounted for about 70% of revenue, with cloud and other services making up nearly 30%. This margin is significantly lower than NVIDIA’s hardware gross margin, which exceeds 70%. The fundamental reason is the difference in business models—NVIDIA primarily sells chips, while Cerebras provides complete system integration, cooling, and software adaptation services, resulting in a heavier cost structure.
From a financial perspective, Cerebras remains in the early stages of commercialization, with an unsteady revenue base, high customer concentration, and negative operating cash flow. The $5.55 billion raised in the IPO is crucial for expanding R&D, data center, and customer deployment capabilities.
From OpenAI to AWS: Cerebras’s Commercialization Path
In early 2026, Cerebras signed a multi-year partnership with OpenAI worth over $10 billion. Under the agreement, OpenAI will use Cerebras’s wafer-scale chips to deploy 750MW of inference compute, with an additional 1.25GW expansion option. Cerebras disclosed total remaining performance obligations of $24.6 billion, much of which comes from this OpenAI deal. OpenAI also provided a $1 billion working capital loan to Cerebras and secured warrants to purchase millions of shares in the IPO, creating a dual tie between compute procurement and equity investment.
In March 2026, Cerebras signed a multi-year memorandum of understanding with AWS to deploy CS-3 systems in AWS data centers and plans to offer wafer-scale compute services to customers via the Amazon Bedrock platform. It’s important to note that this agreement is currently just an MOU, not a formal contract, and AWS’s final adoption scale and deployment timeline remain uncertain.
In Cerebras’s pre-IPO financing rounds, AMD and Benchmark Capital were key shareholders. OpenAI CEO Sam Altman personally invested in Cerebras as early as 2017 and held about 89,373 shares at the end of 2025, with a post-IPO stake valued at roughly $30 million.
Gate TradFi Real Stock Trading: How to Trade CBRS with USDT
On June 1, 2026, Gate officially launched its real stock trading service (Gate TradFi), allowing users to directly trade over 10,000 US stocks and ETFs across five major US exchanges: NYSE, Nasdaq, NYSE Arca, NYSE American, and BATS. As the latest AI chip stock listed on Nasdaq, CBRS is now included in Gate TradFi’s trading pool.
Compared to traditional stock trading channels, Gate TradFi offers a core differentiator: users can trade US stocks directly using USDT. For crypto industry participants accustomed to managing funds in digital assets, this means they can handle both crypto and traditional financial assets from a single account, avoiding the hassle of switching platforms and transferring funds across accounts.
To trade CBRS on Gate, follow these steps: log in to your Gate account and complete identity verification (KYC) to ensure your account is active and eligible for stock trading; transfer USDT to your unified trading account, with real-time exchange processing by the platform; search for "CBRS" and enter the trading interface, choose either market or limit order; once the order is filled, your position will appear in "My Holdings." Because Gate TradFi connects with licensed brokers for underlying brokerage operations, funds used to trade CBRS are protected by the formal securities clearing system, rather than by on-chain mapped assets or tokenized derivatives.
Conclusion
Cerebras’s IPO marks a significant milestone in the AI chip sector. From a technology perspective, wafer-scale integration offers a differentiated computing path for the AI inference market. Financially, rapid revenue growth, high customer concentration, and ongoing non-GAAP losses coexist, with valuation multiples far exceeding peers like NVIDIA. In terms of market performance, after a 68% first-day surge, the stock underwent a notable correction, and current target price ranges from multiple institutions suggest further upside. Investors should weigh these factors and consider their own risk tolerance before making independent decisions. Gate TradFi provides a real stock trading channel for CBRS, supporting convenient participation with USDT—worth keeping an eye on.
