Cerebras announced a 100MW AI infrastructure deployment deal with Gimlet Labs, positioning itself as a key inference accelerator provider.

Cerebras Systems will supply approximately 100 megawatts of AI computing capacity to cloud startup Gimlet Labs under a new hardware agreement. While the companies did not disclose the financial terms, Cerebras CEO Andrew Feldman stated that the company will provide its new CS-4 systems over the next one to two years, with Gimlet planning to make the hardware available through its cloud platform in 2027.

The agreement emphasizes inference—the stage of AI where a trained model produces its output. When someone queries an AI chatbot, for instance, inference is the computational work that generates the response.

Speed is critical in this domain. Faster inference makes AI products feel more responsive and can be especially valuable for cybersecurity and voice applications, according to Gimlet CEO Zain Asgar. Financial analysis is another area where quicker responses deliver meaningful advantages. To support these demanding use cases, Gimlet plans to deploy Cerebras systems to run frontier models—some of the most advanced and computationally intensive AI models available.

Gimlet will operate and maintain the Cerebras equipment itself once delivered. Feldman noted that the agreement also demonstrates that Cerebras hardware can interoperate with systems from other vendors, an important capability since cloud providers typically rely on multiple hardware suppliers rather than a single source.

This interoperability approach means Cerebras does not require customers to replace existing infrastructure; instead, its systems can be integrated where companies need faster AI inference capabilities.

The deal strengthens Cerebras' position among major AI customers. Earlier this year, the company signed an agreement to supply systems to OpenAI, while Nvidia has similarly emphasized inference's growing importance through its own investments and licensing agreements.