Cerebras Systems Rebounds as Freedom Capital Says Selloff Is Overdone
Cerebras Systems (NASDAQ: CBRS) shares rebounded on October 2 after Freedom Capital Markets upgraded the artificial intelligence chipmaker from Hold to Buy and argued that the recent selloff had overstated concerns surrounding its relationship with OpenAI.
The stock gained nearly 2% in Friday trading after analyst Paul Meeks raised his rating and set a $209 price target. The move followed a roughly 17% decline in Cerebras shares over recent sessions, which had pushed the stock close to its lowest closing level since its 2026 public-market debut.
The immediate issue for investors has been reports that OpenAI is using Nvidia GPUs for an ultrafast version of one of its AI models. That raised questions about how much of OpenAI's future inference workload will actually run on Cerebras hardware.
However, the market reaction needs to be viewed against Cerebras' broader business. The company has an existing multi-year OpenAI commitment, rapidly expanding cloud revenue, partnerships with AMD and AWS, and more than $25 billion of remaining performance obligations.
Why Cerebras stock is rebounding
The main catalyst is the Freedom Capital upgrade.
Meeks moved Cerebras to Buy from Hold while keeping a $209 price target. According to the analyst, investor concerns about the OpenAI development have been exaggerated because the broader AI infrastructure market remains supply constrained and Cerebras can potentially serve demand from multiple customers and partners.
That argument helped stabilize the stock after its sharp decline.
Cerebras shares had fallen about 17% in recent sessions amid concerns that OpenAI's reported use of Nvidia hardware for an ultrafast workload could reduce the importance of Cerebras within OpenAI's infrastructure strategy.
For investors, the distinction is important: a change in hardware used for one workload does not necessarily mean the broader OpenAI-Cerebras relationship has ended.
OpenAI remains a major part of the Cerebras story
Cerebras disclosed in its regulatory filings that OpenAI agreed to purchase 750 megawatts of AI inference compute capacity and related services under a master relationship agreement.
The committed capacity is scheduled to be deployed in tranches during 2026 through 2028. OpenAI also has an option to purchase another 1.25 gigawatts of capacity for deployment through 2030, subject to the terms disclosed by Cerebras.
The commercial relationship has also been visible at the product level.
In August, OpenAI introduced an Ultrafast mode for GPT-5.6 Sol, powered by Cerebras. OpenAI said the service could deliver up to 750 output tokens per second, with the initial launch available to a limited group of customers.
Cerebras separately said the same service could operate at up to 14 times the speed of standard processing.
That makes recent reports about Nvidia hardware relevant, but not sufficient on their own to establish that Cerebras has lost its strategic role at OpenAI.
Cerebras' business is growing beyond OpenAI
One reason Freedom Capital sees the recent selloff as potentially excessive is that Cerebras is building a customer and partner ecosystem beyond a single relationship.
The company's second-quarter 2026 results showed GAAP revenue of $180.1 million, up 74% from a year earlier. Core revenue, which excludes certain items, reached $209.9 million, up 103%.
Cloud and other services revenue was particularly strong, reaching $126 million on a GAAP basis, up 281% year over year. Core cloud revenue increased 287% to $127.7 million.
The company also reported $25.4 billion in remaining performance obligations at the end of June.
Remaining performance obligations represent contracted revenue that has not yet been recognized. It does not mean Cerebras can immediately count the entire amount as sales or profit, but it provides an important indicator of contracted future business.
Cerebras expects to deploy that demand as it expands its manufacturing and data-center capacity.
AMD and AWS partnerships add another growth avenue
Cerebras is also positioning its technology as a complement to other AI chips rather than necessarily trying to replace every accelerator in a data center.
In July, AMD and Cerebras announced a disaggregated inference architecture combining AMD's Helios systems with Cerebras' Wafer-Scale Engine.
In this setup, AMD hardware handles high-throughput processing while Cerebras technology focuses on the decode stage, where extremely fast token generation can reduce response latency. The companies said the combined architecture could deliver up to five times higher tokens per second per watt for targeted workloads.
AWS is pursuing a similar approach. Amazon Web Services and Cerebras announced a collaboration that combines AWS Trainium for prefill with Cerebras CS-3 systems for fast inference.
Cerebras said in August that its AWS partnership is expected to bring its disaggregated inference technology to Amazon Bedrock in the first quarter of 2027.
That matters because the AI infrastructure market is increasingly becoming heterogeneous. Different chips can be used for different stages of an AI workload depending on cost, speed, memory and power requirements.
The growth numbers are strong, but profitability remains a key issue
Cerebras' revenue growth is significant, but investors should not overlook its margin profile.
For the second quarter, GAAP gross margin was 14%, while core gross margin was 41%. Core operating margin remained negative at 16%, although that represented a substantial improvement from the year-earlier period.
The company expects third-quarter 2026 core revenue of approximately $214 million to $216 million, with core gross margin of 38% to 40%.
For the full year, Cerebras raised its core revenue outlook to $880 million-$890 million and expects core gross margin of 41% to 43%. Core operating margin is still expected to remain negative at 19% to 17%.
This is an important part of the investment case. Cerebras is pursuing rapid expansion, but translating that growth into sustainable profitability will depend on manufacturing scale, data-center utilization, pricing and operating efficiency.
Capacity expansion is another critical test
Cerebras said it had more than 600 MW of data-center capacity live or under contract for delivery by the end of 2027.
It is also expanding manufacturing capacity through contract manufacturers including Flex, Sanmina and Rocket EMS, with the company expecting manufacturing capacity to increase more than tenfold during 2026.
That expansion is necessary if the company is going to fulfill its large contracted commitments.
But rapid infrastructure expansion creates execution risk. Data centers need power, equipment and deployment capacity, while chip manufacturing depends on a complicated supply chain.
Cerebras has said it secured the TSMC wafer supply required for its planned growth and has worked with other component suppliers to support its 2027 and later expansion.
A new customer adds to the evidence of demand
Cerebras' growth story received another recent boost through its partnership with Gimlet Labs.
In September, Cerebras announced that it would supply its CS-4 systems to Gimlet for a cloud infrastructure deployment requiring approximately 100 megawatts of power. Deployment is expected over one to two years, with Gimlet targeting availability of Cerebras hardware through its cloud infrastructure in 2027. Financial terms were not disclosed.
The agreement does not replace the importance of OpenAI, but it demonstrates that Cerebras is attempting to build demand across multiple AI infrastructure customers.
What are the biggest risks for CBRS investors?
The OpenAI issue remains the most immediate risk because of the size and importance of that relationship.
If OpenAI ultimately shifts a significant amount of inference demand away from Cerebras, the company's revenue expectations could be affected. Investors therefore need to distinguish between temporary workload allocation and a structural change in OpenAI's procurement strategy.
There is also competition from Nvidia, AMD and custom AI accelerators.
Cerebras' wafer-scale architecture is differentiated, particularly for low-latency inference, but customers can choose different combinations of chips depending on workload requirements.
Profitability is another concern. High revenue growth does not automatically produce strong earnings if infrastructure, manufacturing and research costs rise rapidly.
Finally, the $209 price target from Freedom Capital is an analyst estimate, not a guaranteed future share price. The stock remains highly sensitive to expectations about AI spending, customer contracts and execution.
What investors should watch next
For Cerebras, the most important developments over the coming quarters include:
Progress on the 750 MW OpenAI commitment
Customer demand beyond OpenAI
Revenue growth and cloud revenue
Gross-margin improvement
Manufacturing capacity expansion
Data-center deployment and utilization
AMD partnership production rollout
AWS Bedrock integration expected in 2027
New hyperscaler and enterprise contracts
Evidence that AI inference demand continues to outpace available infrastructure
The September Gimlet agreement and the company's AMD and AWS relationships provide additional avenues for growth, but the market will ultimately want to see those partnerships translate into recognized revenue and improving economics.
Conclusion
Cerebras Systems stock rebounded after Freedom Capital upgraded the shares to Buy and argued that the recent selloff over concerns about OpenAI was excessive. The analyst set a $209 price target after the stock had fallen roughly 17% over recent sessions.
The underlying story remains more complicated than a single OpenAI hardware decision. Cerebras has a disclosed 750 MW OpenAI commitment, rapidly growing cloud revenue, more than $25 billion in remaining performance obligations, and partnerships with AMD, AWS and other customers.
At the same time, the company remains in an aggressive investment phase, with negative operating margins and substantial capacity expansion ahead. For investors, the key indicators will be whether Cerebras can diversify its customer base, execute its infrastructure expansion and convert strong AI inference demand into sustainable revenue and profitability.
Follow the blog for more updates on AI stocks, semiconductor companies, data-center infrastructure and global technology markets.
This article is for informational and educational purposes only and should not be considered investment advice.