Skip to content

News, guides and free tools for the SaaS stack

Press release. This announcement was issued by Gimlet Labs via GlobeNewswire. Below is our summary; claims are the company’s own. Read the original release

Press release

Gimlet Labs raises $300M Series B led by Andreessen Horowitz at a $3B valuation

The company says its multi-silicon inference cloud runs AI workloads across GPUs, accelerators and CPUs with up to 10× throughput gains.

Network cables in a server rack
Network cables in a server rack · Photo: Taylor Vick / Unsplash

Gimlet Labs announced a $300 million Series B led by Andreessen Horowitz, valuing the company at $3 billion.

Sapphire Ventures, M12 (Microsoft's venture fund), Arm, Menlo Ventures and Factory also took part.

What Gimlet Labs does

Gimlet Labs sells inference cloud software that runs AI workloads across different kinds of hardware at once, including GPUs, other AI accelerators and CPUs. The goal is to get more speed and efficiency out of each chip. Gimlet calls it the first multi-silicon inference cloud built for agentic AI.

Its approach combines three things:

  • Automated GPU kernel generation
  • Workload orchestration
  • Running work across mixed hardware

Claims made in the announcement

  • The company says it has tripled its customer base since March 2026 and has billions of dollars in contracted revenue.
  • It reports up to 10× gains in throughput and interactivity.
  • The release notes that AI inference now runs to quadrillions of tokens a month, and cites an industry projection of $765 billion in AI capital spending in 2026.

What the executives said

  • Raghu Raghuram of Andreessen Horowitz, who joins Gimlet's board, argued that the answer to AI's compute demand is better architecture, not just more infrastructure. He pointed to Gimlet getting more throughput from every watt.
  • Zain Asgar, CEO and co-founder, said customers can serve very large volumes of tokens at low latency by orchestrating work across different hardware types.

Why SaaS teams should care

Most SaaS companies won't buy inference infrastructure directly. But the cost and speed of running AI models now shape the pricing of almost every "AI feature" in the tools you use. Competition in inference, including running models on cheaper mixed hardware, is one of the forces that could push those costs down.

About Gimlet Labs

Gimlet Labs describes itself as an applied AI research and product company. It works across the stack — automated GPU kernel generation, workload orchestration and execution across mixed hardware — to improve AI inference performance.

Media contact

Amber RowlandThe Rowland Agencyamber@therowlandagency.com
Sources

Views and claims in press releases are the issuing company’s own. Report an issue