Cerebras to Supply AI Systems to Cloud Computing Startup Gimlet Labs

Cerebras Systems is set to supply AI chips and computing systems to cloud computing startup Gimlet Labs as demand for faster artificial intelligence inference continues to accelerate.

The hardware supplied by Cerebras will have the capacity to consume roughly 100 megawatts of power, according to the companies. Gimlet plans to use the systems to expand its cloud computing business and provide customers with access to high-performance infrastructure for running advanced AI models.

The agreement centers on Cerebras’ CS-4 systems, which the chipmaker unveiled earlier this year. Cerebras CEO Andrew Feldman said the systems will be delivered over a period of one to two years, while Gimlet plans to make the hardware available through its cloud platform beginning in 2027. Financial terms of the deal were not disclosed.

Unlike AI model training, which involves creating or refining models, inference is the process of using trained models to generate responses and perform tasks. Faster inference can be particularly important for applications such as cybersecurity, voice-based services and financial analysis, where users and businesses may require rapid responses.

Gimlet CEO Zain Asgar said the company plans to target businesses and startups building products around AI. Gimlet will use the Cerebras systems to run so-called frontier models, the increasingly sophisticated AI systems that require significant computing resources to deliver responses quickly.

The partnership also reflects Gimlet’s strategy of building a cloud platform that can operate different types of AI chips rather than relying on a single processor architecture. The company describes its infrastructure as a multi-silicon inference cloud, allowing workloads to be matched with different processors depending on performance, latency and efficiency requirements.

Gimlet recently raised $300 million in a Series B funding round led by Andreessen Horowitz, bringing its valuation to about $3 billion. The company said it has secured billions of dollars in contracted revenue and is scaling its managed infrastructure to hundreds of megawatts.

The Cerebras deal comes as AI companies and cloud providers increasingly seek alternatives and additions to traditional GPU infrastructure. OpenAI has also agreed to use Cerebras chips, while Nvidia has pursued partnerships involving competing AI accelerator technology.

For Gimlet, adding Cerebras systems expands the range of hardware available through its cloud. For Cerebras, the agreement provides another avenue for deploying its processors as the AI industry increasingly shifts toward infrastructure optimized for fast, large-scale inference.