Research Report Interpretation: Breakthrough in the Wafer-Scale Chip Track, Cerebras' Market Debut Exceeds Expectations

Author: Rita

Tide Trend Guide

On June 24, Morgan Stanley released its first follow-up report on Cerebras Systems after its IPO, raising the target price from $250 to $273 and maintaining an Overweight rating. This chip company holds a unique position in fast inference. Its latest financial report showed both revenue and gross margin exceeding expectations. Analysts believe that the 750MW capacity agreement and the cloud service self-build plan will drive high-speed growth over the next two years. Crucially, market doubts about the company’s technology scalability have been eliminated.

First-Quarter Performance Exceeds Expectations

Cerebras delivered solid performance in the quarter ended March 2026. Revenue reached $193.4 million, up 13% quarter over quarter, beating expectations. Non-GAAP gross margin was 46.5%. While it may not sound particularly high, given that the company currently relies on leased servers to deliver its services, this result is already quite strong.

Morgan Stanley noted that the management’s outlook is somewhat conservative. This also means there is potential for upward revisions in the future. FY2026 revenue guidance is $864 million, far above the same period last year. Expected FY2027 revenue is $2.714 billion, nearly three times.

Gross margin is expected to quickly rise from 39.4% in FY2026 to 51.1% in FY2027, and further to 57.9% in FY2028. The key driver is that the company is shifting from leasing other companies’ hardware to building its own cloud services. Leased servers have a gross margin of only 37%, while the long-term gross margin target for self-built cloud services is 57.9%. This is a fundamental change to profitability.

The 750MW Agreement Sets the Growth Base

Cerebras has signed a formal agreement for 750MW of capacity. A 750MW capacity agreement means customers have provided clear capacity requirements, which essentially locks in the company’s revenue floor for the next 2 to 3 years. Many chip companies are still struggling to secure orders, but Cerebras has locked in long-term capacity in advance. The agreement covers multiple customers, and its partnership with Amazon has moved from intent to a formal agreement. Although analysts’ estimates of Amazon’s contribution may be somewhat conservative, this still leaves room for future growth.

The market previously had doubts about Cerebras’ wafer-scale processors, worrying that its architecture might not truly support large models. This question has been answered clearly. The company successfully validated its real-world performance on trillion-parameter models (e.g., Kimi K2.6), proving the scalability of this architecture. With technical risk removed, investors naturally shift their focus to commercialization progress. The 750MW capacity agreement directly demonstrates the level of customer confidence in this technology.

Inference Chips Become the New Battlefield

Why focus on fast inference rather than training? Because inference is becoming the fastest-growing area of AI infrastructure spending. Training is a one-time investment, while inference is continuous and repeated. As large-model applications roll out, inference costs will gradually come to dominate total spending across the AI chip market.

Cerebras’ leadership in fast inference gives it a special position in the market. Most competitors focus on training chips, while this company has chosen a differentiated path with inference. That specialization is, in fact, an advantage.

Three Scenarios and Target Price

Morgan Stanley provided three scenarios. In the base case, adjusted revenue in 2028 reaches $6.0 billion, corresponding to a target price of $273. This is the scenario analysts are most confident in, based on the 750MW capacity being deployed on schedule. In the bull case, capacity deployment exceeds expectations, reaching above 1GW, and 2028 revenue could surpass $10.0 billion, corresponding to a target price of $400. In the bear case, capacity deployment is delayed, and 2028 revenue could be only $2.7 billion, corresponding to a target price of $90.

As of June 23, 2026, Cerebras’ share price closed at $226.72, implying 20% upside from the $273 target price. Even compared with the market consensus target of $250, there is still about 10% upside. Morgan Stanley’s upward revision reflects its recognition of the company’s certainty.

Risks and Opportunities

Capacity deployment progress is the biggest risk. If large data center infrastructure construction is delayed, Cerebras’ revenue growth will also be dragged down. Customer concentration is another risk—if demand from key customers changes, the impact on the company would be significant. However, if deployment goes smoothly, the chances of winning new customers are actually quite high. The fast-expanding inference chip market is likely to attract more customers with Cerebras’ differentiated products.

Cerebras has transformed from a controversial emerging company into a firm with confirmed orders, supported by its first-quarter results and technology validation. This shift is crucial. Morgan Stanley’s upward revision reflects recognition of this certainty. While the wafer-scale chip track is still full of variables, Cerebras has already established a leading position in fast inference. Over the next two years, whether the company’s revenue and profit margins improve as expected will determine investors’ returns.

Disclaimer

This article is Tide Research’s compilation and interpretation of a third-party brokerage research report. The ratings, target prices, earnings forecasts, and related judgments cited in the text are all the views of Morgan Stanley analysts and represent only the position of their affiliated institution; they do not represent Tide Research’s views, nor do they constitute any investment advice.

When reading, please note three points: First, the target price is an analyst’s expectation of the next approximately 12 months; it is a forecast rather than a commitment and will be adjusted repeatedly in response to performance and market conditions. Second, sell-side research reports are naturally bullish, and some covered companies have investment banking relationships with the brokerage. Third, the value of a research report lies in its core logic and underlying assumptions—not in any single target price. Look at the logic, not just the price.

There are risks in the market, and decisions must be made independently. This article should not be used as a basis for buying or selling any securities.

Data Source: Morgan Stanley research report (Joseph Moore, June 24, 2026) · Public market data

Tide Research · TideResearch · June 2026