AI cloud startup Gimlet Cloud has signed a partnership with Cerebras to deploy 100MW of wafer-scale compute dedicated to inference workloads on the Gimlet Cloud platform. The first data center under the agreement is scheduled to come online later this year.

The deal also positions Gimlet as one of the launch partners for Cerebras' next hardware generation, the CS-4. Availability for Gimlet customers is planned for 2027.

Gimlet Cloud is building a cloud platform offering multiple silicon options rather than betting on a single architecture. The company laid out its reasoning in a blog post announcing the Cerebras partnership: "GPUs and wafer-scale systems have different strengths. The opportunity is to combine them within a single inference workload, using GPUs and Cerebras wafer-scale systems where each performs best, and in the process maximize both throughput and latency."

That hybrid approach frames the 100MW commitment. Instead of treating Cerebras' wafer-scale processors as a replacement for GPU fleets, Gimlet intends to route parts of each inference workload to the hardware best suited for it, targeting gains in both throughput and latency simultaneously.

Funding follows fast

The Cerebras agreement lands weeks after Gimlet closed a $300 million Series B round. Andreessen Horowitz led the round. The participant list includes Sapphire Ventures, Menlo Ventures, 645 Ventures, Arm, Eclipse, Emergence, Factory, Hudson River Trading, M12, OnePrime Capital, Prosperity7, QuantumLight, Samsung Ventures, Tiger Global Management, Triatomic, Wing Ventures, and XTX Markets.

The round valued Gimlet at $3 billion.

CEO Zain Asgar tied the financing directly to infrastructure buildout when the round was announced. "Since March, we've secured billions of dollars in contracted revenue and are scaling toward hundreds of megawatts of managed capacity," Asgar said. "This funding will help us build that infrastructure, expand our team, and bring high-performance inference to many more customers."

The Series B follows an $80 million Series A raised in March. In roughly half a year, Gimlet has moved from its first institutional round to a multi-billion-dollar valuation, backed by contracted revenue it measures in billions of dollars.

Founding team and Stanford ties

San Francisco-based Gimlet was founded by Zain Asgar, Michelle Nguyen, Omid Azizi, Natalie Serrino, and James Bartlett. Asgar also holds a position as a professor of computer science at Stanford University.

The company has not disclosed details of its data center footprint. Its current hiring includes an opening for a data center facilities operations lead, a signal that physical infrastructure teams are still being assembled as the Cerebras deployment approaches its first milestone.

What the 100MW commitment means

A 100MW commitment places Gimlet's Cerebras deployment among the larger dedicated inference builds announced by cloud startups to date. For Cerebras, the agreement extends a pattern of hyperscale-style capacity deals with AI cloud providers, locking in demand for current-generation wafer-scale systems while securing launch partners for the CS-4 in 2027.

For Gimlet customers, the timeline is concrete: inference capacity from the first Cerebras data center later this year, with a next-generation hardware path already on the roadmap. The two-year gap to CS-4 availability gives Gimlet time to scale its current deployment and refine workload routing between GPU and wafer-scale resources before the newer systems arrive.

The company's stated target of hundreds of megawatts of managed capacity suggests the Cerebras deal covers only part of a broader infrastructure plan, with additional silicon types and sites likely to follow.