What happened with CBRS today?
No read yet
Uses news and trading volume to explain the day’s move and how it relates to the market.
Stop guessing. Get the call, a three-month target price, the best and worst cases, and a source for every claim.
Cerebras Systems supplies computing infrastructure to organizations that train AI models and run them to generate responses, a process called inference. Its Wafer-Scale Engine keeps computing, memory and communications on one wafer-sized chip, which it combines with CS-3 systems and software into supercomputers. Hardware sales provide most revenue: customers buy systems for their own data centers and pay separately for renewable software subscriptions and support.
Customers that do not want to build or maintain that infrastructure can use Cerebras Cloud or partner platforms. They pay for usage or reserve dedicated capacity, generally paying for that commitment whether they use it or not. Cloud offerings remain at an early commercial stage. Revenue is highly concentrated: Mohamed bin Zayed University of Artificial Intelligence (MBZUAI) supplies most, and G42 contributes another substantial share. Cerebras has also begun delivering inference capacity under a multi-year purchase commitment from OpenAI; further deliveries depend on expanding data-center capacity and meeting contractual deadlines and performance requirements.
Cerebras relies on TSMC to fabricate all its processor wafers, without long-term supply commitments. Overseas sales face export-license requirements, including security and compliance conditions on licensed deployments for G42 and MBZUAI in the United Arab Emirates. Some customer contracts, including OpenAI’s, also restrict sales or collaboration with named competitors, limiting whom Cerebras can serve.
The risk section is the longest, dullest part of the filing. This keeps only what can really hurt you: where it hits, how hard, when it plays out.
Every company claims an edge. Each one is checked against hard numbers like margins and returns: which hold, and which line gives way first.
Cerebras’ results depend heavily on demand from G42 and MBZUAI and on delivering the large inference-capacity commitment bought by OpenAI.
NVIDIA competes for the AI training and inference budgets Cerebras targets, selling GPU systems and software against Cerebras’ wafer-scale systems. Watch NVIDIA’s data-center AI business and CUDA software ecosystem, rather than its gaming GPUs.
AMD offers data-center customers AI GPUs and ROCm software as an alternative to Cerebras’ wafer-scale hardware and software. Its ROCm improvements for generative AI show how that competing option is developing.
CoreWeave competes with Cerebras for AI labs and enterprises paying for cloud capacity to train and run models. Watch its committed cloud contracts for the demand those customers are locking in with a rival.
Cerebras depends on TSMC to manufacture all its processor wafers from Cerebras’ designs, so TSMC’s capacity, yields and process transitions constrain system production. Watch whether TSMC can support next-generation Cerebras products on schedule and in sufficient volume.
Amazon’s AWS both competes with Cerebras for cloud AI customers and buys from it as a significant customer. Watch deployment under AWS’s binding term sheet to put Cerebras systems in its own data centers, alongside the partnership expanding access to Cerebras inference.
Oracle competes with Cerebras Cloud for customers paying for AI computing and committing to cloud capacity. Oracle’s cloud infrastructure business is the relevant contest, rather than its broader hardware and services businesses.
Alphabet’s Google Cloud competes with Cerebras Cloud for customers buying AI computing capacity. Watch Google Cloud separately: advertising remains Alphabet’s main revenue source and says less about this rivalry.
Microsoft Azure competes with Cerebras Cloud for customers paying for cloud AI computing; that competition is separate from Microsoft Marketplace’s distribution role. Watch Azure’s AI compute offering, where Microsoft’s large data centers and pooled customer demand support lower unit costs and better utilization.
Broadcom develops custom AI chips for large cloud providers, giving them an alternative to Cerebras’ wafer-scale processors for large-scale AI computing.
Cerebras’ cloud expansion depends in part on Zone Frontier delivering the 40-megawatt Minnesota data-center campus it is leasing exclusively to Cerebras for ten years. Watch whether the campus brings 40 megawatts of critical IT load online by the end of the second calendar quarter of 2027, as planned.
Intel sells data-center customers AI accelerators that compete with Cerebras’ specialized AI systems. Watch its AI accelerator business rather than its broader CPU sales, which make Intel a less focused comparison.
Penguin Solutions builds and manages AI and high-performance computing clusters for customers, competing with Cerebras’ integrated supercomputers and cluster-operation services.
Amazon.Com Inc earningsAMZN
Amazon’s report can show how AWS’s cloud AI business is developing; AWS is both a Cerebras customer and a rival for cloud AI computing customers.
Look at changes in AWS’s cloud AI business and related demand for computing capacity. If they strengthen, that points to more AI-computing activity at AWS and supports the customer and distribution opportunity for Cerebras systems and inference; if they weaken, there is less support for that opportunity. Amazon’s results alone cannot confirm a Cerebras order or deployment.
CoreWeave, Inc. earningsCRWV
CoreWeave’s report can show how much AI labs and enterprises are committing to cloud computing capacity, revealing the demand and competition Cerebras Cloud faces.
Look at CoreWeave’s customer commitments and the capacity demand tied to them. More commitments would show AI labs and enterprises locking more demand into a rival’s cloud capacity; weaker commitments would mean less evidence of that competitive pull. This reflects customers’ choices with a rival, not Cerebras’s own orders.
Cerebras Systems Inc. earningsThis company
Cerebras Systems’ report will show whether CS system sales and cloud-services revenue are growing together, and whether the data-center costs of serving cloud customers are squeezing gross margin.
Read hardware revenue alongside cloud and other services revenue, then check gross margin. Growing cloud-services revenue with a stable or rising margin would show the service business expanding without costs swallowing the growth; growing cloud revenue with a falling margin would show data-center costs offsetting it. If hardware remains the main driver while cloud services lag, revenue still depends chiefly on system deliveries.
Nvidia Corp earningsNVDA
NVIDIA’s report can show demand for data-center AI systems, where its GPU systems and software compete with Cerebras for training and inference budgets.
Focus on NVIDIA’s data-center AI business, not gaming GPUs. Strength there would show substantial training and inference budgets, while also showing an established rival capturing that demand; weakness would mean less evidence of spending in the same market, not proof that customers are switching to Cerebras. The report helps readers judge both the size of AI-system demand and how much a major rival is drawing from it.
