The AI Chip Company That Refused to Play by NVIDIA's Rules
The latest earnings suggest Cerebras is moving from being a niche hardware company toward becoming an AI infrastructure provider. Revenue reached $194 million in Q1 FY26, up 94% year over year, with both major business lines growing rapidly. Hardware revenue increased 59% to $111 million, while cloud and services jumped 178% to $83 million. The faster growth in cloud matters because recurring usage revenue is generally more valuable than one time hardware sales
That shift is important for margins as well. Hardware businesses tend to be cyclical and tied to customer spending on new equipment. Cloud inference generates revenue every time customers run models, creating a steadier stream of income if usage continues to grow. Management is clearly leaning into this model, positioning Cerebras as a provider of AI compute rather than simply a chip vendor.
The OpenAI agreement is the biggest reason investors are paying attention. A contract worth more than $20 billion is significant for a company whose quarterly revenue is still below $200 million. While accounting rules mean not all of that spending shows up as reported revenue because of the warrants issued to OpenAI, the agreement provides long term demand that few AI hardware startups can match
The other notable development is customer diversification. Not long ago, G42 accounted for the vast majority of Cerebras' business, leaving the company exposed if that relationship weakened. Today, OpenAI and AWS have joined the list of major customers. The customer base is still concentrated, but it's becoming broader and includes some of the largest buyers of AI infrastructure
The supply chain story also deserves attention
Most AI chip companies depend on scarce high bandwidth memory and advanced packaging capacity, both of which remain industry bottlenecks. Cerebras largely avoids those constraints through its architecture, relying instead on SRAM and a mature 5 nanometer manufacturing process. That doesn't automatically make its technology better than NVIDIA's, but it does reduce one of the biggest risks facing AI hardware companies: getting enough components to meet demand
None of this means Cerebras is without challenges. NVIDIA remains the dominant player, with a mature software ecosystem, deep customer relationships, and enormous R&D resources. Many customers also prefer hardware that fits into existing GPU-based infrastructure, making adoption of an entirely different architecture a higher hurdle. Cerebras has to prove that its speed advantage is large enough to justify changing how customers deploy AI workloads.
The investment case ultimately comes down to execution. If cloud inference continues to outgrow hardware, OpenAI ramps its usage over time, and enterprise customers adopt Cerebras for latency sensitive AI applications, revenue could compound quickly from today's relatively small base. On the other hand, if customer concentration remains high or demand fails to broaden beyond a handful of large contracts, growth could prove much less durable than the headline numbers suggest
After the IPO freefall, CBRS bulls are trying to reclaim the $250 territory
For now the earnings point in the right direction. Revenue is accelerating, the mix is shifting toward recurring services, and the company has landed customers that validate its technology. The next few quarters will show whether those early wins translate into a business that can compete at scale in the AI infrastructure market.
The latest earnings suggest Cerebras is moving from being a niche hardware company toward becoming an AI infrastructure provider. Revenue reached $194 million in Q1 FY26, up 94% year over year, with both major business lines growing rapidly. Hardware revenue increased 59% to $111 million, while cloud and services jumped 178% to $83 million. The faster growth in cloud matters because recurring usage revenue is generally more valuable than one time hardware sales
That shift is important for margins as well. Hardware businesses tend to be cyclical and tied to customer spending on new equipment. Cloud inference generates revenue every time customers run models, creating a steadier stream of income if usage continues to grow. Management is clearly leaning into this model, positioning Cerebras as a provider of AI compute rather than simply a chip vendor.
The OpenAI agreement is the biggest reason investors are paying attention. A contract worth more than $20 billion is significant for a company whose quarterly revenue is still below $200 million. While accounting rules mean not all of that spending shows up as reported revenue because of the warrants issued to OpenAI, the agreement provides long term demand that few AI hardware startups can match
The other notable development is customer diversification. Not long ago, G42 accounted for the vast majority of Cerebras' business, leaving the company exposed if that relationship weakened. Today, OpenAI and AWS have joined the list of major customers. The customer base is still concentrated, but it's becoming broader and includes some of the largest buyers of AI infrastructure
The supply chain story also deserves attention
Most AI chip companies depend on scarce high bandwidth memory and advanced packaging capacity, both of which remain industry bottlenecks. Cerebras largely avoids those constraints through its architecture, relying instead on SRAM and a mature 5 nanometer manufacturing process. That doesn't automatically make its technology better than NVIDIA's, but it does reduce one of the biggest risks facing AI hardware companies: getting enough components to meet demand
None of this means Cerebras is without challenges. NVIDIA remains the dominant player, with a mature software ecosystem, deep customer relationships, and enormous R&D resources. Many customers also prefer hardware that fits into existing GPU-based infrastructure, making adoption of an entirely different architecture a higher hurdle. Cerebras has to prove that its speed advantage is large enough to justify changing how customers deploy AI workloads.
The investment case ultimately comes down to execution. If cloud inference continues to outgrow hardware, OpenAI ramps its usage over time, and enterprise customers adopt Cerebras for latency sensitive AI applications, revenue could compound quickly from today's relatively small base. On the other hand, if customer concentration remains high or demand fails to broaden beyond a handful of large contracts, growth could prove much less durable than the headline numbers suggest
After the IPO freefall, CBRS bulls are trying to reclaim the $250 territory
For now the earnings point in the right direction. Revenue is accelerating, the mix is shifting toward recurring services, and the company has landed customers that validate its technology. The next few quarters will show whether those early wins translate into a business that can compete at scale in the AI infrastructure market.
Похожие публикации
Отказ от ответственности
Информация и публикации не предназначены для предоставления и не являются финансовыми, инвестиционными, торговыми или другими видами советов или рекомендаций, предоставленных или одобренных TradingView. Подробнее читайте в Условиях использования.
Похожие публикации
Отказ от ответственности
Информация и публикации не предназначены для предоставления и не являются финансовыми, инвестиционными, торговыми или другими видами советов или рекомендаций, предоставленных или одобренных TradingView. Подробнее читайте в Условиях использования.