After the stock price of "NVIDIA Challenger" Cerebras nearly halved in three weeks since its listing, Wall Street collectively shouted "buy."

According to Zhitong Finance APP, on June 8, Cerebras Systems (CBRS.US) saw its stock price rise by 3.02% in early trading to about $207, marking the second consecutive trading day of moderate rebound for the stock. However, even so, this "AI chip newcomer," which just landed on NASDAQ about a month ago, has seen its stock price evaporate nearly 42% from the intraday high of $350 on its first trading day, with its market capitalization dropping from over $100 billion to about $44.1 billion today.

In stark contrast to the secondary market, Wall Street research reports have collectively turned bullish. On this day, multiple Wall Street institutions including Morgan Stanley, UBS, Wedbush, Rosenblatt, Mizuho, Barclays, and Needham "clustered" to release their initial coverage reports on Cerebras, almost uniformly giving "buy," "overweight," or "outperform" ratings, with target prices concentrated in the range of $250 to $300.

Why is Wall Street collectively bullish?

The reason Cerebras can receive "buy" ratings from Wall Street investment banks is not fundamentally due to how impressive its financial reports are, but rather because it occupies a unique "narrow gate" in the AI chip market.

The core difference lies in the product form of Cerebras. Currently, the AI computing power market is dominated by NVIDIA GPUs, while Cerebras has taken a different path by manufacturing the world's only commercially deployed wafer-scale processor—the Wafer Scale Engine. According to technical data, the latest WSE-3 integrates 40 trillion transistors and 900,000 AI-optimized cores, with each core equipped with 48KB of local SRAM, resulting in a total on-chip SRAM of 44GB, providing 21PB/s of on-chip memory bandwidth, which is thousands of times that of typical HBM solutions. This architecture fundamentally eliminates the bottleneck of the "memory wall," and Cerebras claims that its system is 15 times faster than current leading GPU solutions when processing most workloads, especially suitable for AI inference scenarios that are extremely sensitive to latency.

Morgan Stanley pointed out that as the inference capabilities of AI workloads continue to enhance, the demand for fast, low-latency inference is rapidly growing; with a large backlog of signed orders and a commitment capacity agreement of up to 750MW, Cerebras is well-prepared to seize this opportunity.

Analyst Timothy Arcuri noted that Cerebras is the only supplier that provides goods to OpenAI on a prepayment basis and is also collaborating with Amazon AWS in the inference fieldUBS has set a target price of $300 based on conservative estimates.

Wedbush focuses on the market cycle switch—Cerebras's entry into the public market coincides with a critical moment when the AI computing cycle transitions from training to inference. The firm stated that in inference scenarios, speed rather than raw floating-point computation power directly determines the commercial value of the output. Cerebras's WSE-3 is the largest chip on the market to date, designed for rapid token generation, and its differentiated advantages provide asymmetric upside potential.

Before releasing a bullish report on Monday, Bank of America Merrill Lynch was the first to predict that Cerebras would lead all large AI chip companies with an explosive performance growth rate of 370% over the next year, a forecast that even surpasses NVIDIA's 195% revenue growth during the same period.

Morgan Stanley (Overweight rating, target price $250)

Morgan Stanley analyst Joseph Moore wrote in a report to clients: "We believe Cerebras is one of the most differentiated AI infrastructure companies, with the industry's only commercially deployed wafer-scale processor at its core. As the inference capabilities of AI workloads continue to grow, the demand for fast, low-latency inference is also rapidly increasing. With a large number of signed orders and a commitment capacity agreement of 750 megawatts, we believe Cerebras is well-positioned to seize this opportunity. This is an excellent investment opportunity in an AI processor company that has a first-mover advantage over NVIDIA and significant investment potential as the field continues to evolve."

UBS (Buy rating, target price $300)

UBS analyst Timothy Arcuri wrote in a report to clients: "The Cerebras wafer-scale engine (WSE) is the largest computing chip globally, outperforming GPUs in certain fast inference applications aimed at the high-end inference market. Importantly, the company has demonstrated strong commercial momentum, currently being the only supplier to OpenAI on a prepayment basis, with deepening collaboration with Amazon in the inference space and many other potential partnership opportunities. Our target price is based on what we consider a conservative OpenAI deployment assumption, taking into account the potential increase in options from Amazon in the coming years. As we believe the company will ultimately evolve into a more comprehensive hardware supplier, our target price of $300 is derived from a 10x enterprise value/sales (EV/Sales) calculation, based on projected sales of $11 billion in 2029, and referencing the performance of major hardware competitors such as NVIDIA, AMD, Broadcom, and Marvell Technology."

Needham (Buy rating, target price $300)

Needham analyst N. Quinn Bolton wrote in a report to clients: "Cerebras is the only wafer-scale engine (WSE) supplier whose SRAM capacity and memory bandwidth exceed those of any other AI processor by several orders of magnitude. The WSE is designed for fast inference workloads that require low latency, including real-time encoding and instant research agentsIn January 2026, the company announced a computing agreement with OpenAI worth over $20 billion, under which OpenAI will deploy approximately 750 megawatts of Cerebras computing power by 2028, with an option to add an additional 1.25 gigawatts. In March, Cerebras announced a partnership with AWS to develop decoupled inference. If OpenAI exercises the option for the additional 1.25 gigawatts of capacity, or if Amazon receives any significant number of Cerebras systems, we believe our current model has significant upside potential.

Wedbush (Outperform rating, target price $270)

Wedbush analyst Matt Bryson wrote in a report to clients: "Cerebras is the only company to commercialize wafer-scale AI chips, and as it enters the public market, the computing cycle is shifting from training to inference; in this market phase, speed rather than raw floating-point computing power determines output value. Cerebras' third-generation wafer-scale engine (WSE-3) is the largest chip sold to date, designed for a task that defines high-value inference: rapid token generation. With a differentiated architecture, a leap in contract revenue from OpenAI and AWS, and a market that has just begun to pay for speed, we see an asymmetric, upward-tilting pattern."

Barclays (Overweight rating, target price $280)

Barclays noted that the recent agreements between Cerebras and OpenAI and Amazon represent significant progress in the AI chip market, where customers are competing for limited chip supplies to scale their businesses. Barclays pointed out that Cerebras is positioned in the fast inference market, which the bank believes could reach a potential size of $300 billion by the end of this decade. Barclays stated that as agent technology becomes more prevalent, the demand for workloads requiring fast inference will increase, aligning with Cerebras' strengths.

Mizuho (Outperform rating, target price $300)

Mizuho pointed out that Cerebras is a direct beneficiary of the growth in AI capital expenditures, which are expected to grow at a compound annual growth rate of 36% to approximately $2.8 trillion by 2030. Mizuho believes the company is leading in the fastest-growing segment of "fast inference." Mizuho expects the compound annual growth rate for this segment to reach approximately 291% by 2030, with a size of about $550 billion, while the overall compound annual growth rate for the entire AI inference market is around 53%. The company has a self-developed software and hardware stack, centered around the wafer-scale engine and large-scale on-chip SRAM, thereby building a strong competitive advantageIn addition, Mizuho stated that the continuously expanding customer channels support the company's revenue to achieve a compound annual growth rate of 122% between 2025 and 2029; the target price of $300 corresponds to 15 times the expected price-to-sales ratio in 2028, which is at the midpoint of the 8 to 24 times price-to-sales ratio range for Nvidia and Broadcom before the accelerated development of their respective artificial intelligence businesses