Gimlet Labs raises $80M Series A for AI that runs on many chips

In short: Gimlet Labs raised $80 million to build software that lets the same AI app run across many kinds of computer chips.

What happened

Gimlet Labs announced an $80 million Series A funding round, led by Menlo Ventures. The raise brings the company’s total funding to $92 million.

Gimlet makes a platform for running “agentic AI” systems, meaning AI that can take steps and use tools to complete a task, like a travel planner that searches, compares options, and books. The company says its software can run these AI workloads across different hardware from different vendors without developers having to change their code.

The platform breaks an AI workload into smaller parts and then assigns each part to the most suitable hardware. It supports GPUs (the chips often used for AI) from NVIDIA, Intel, and AMD, plus CPUs (general purpose chips) and other specialized accelerators. In additional reporting, the company also described support for chips from ARM, Cerebras, and d-Matrix.

Gimlet says this approach helps with a common problem in AI deployment. Many AI systems are easiest to run on the same type of hardware they were trained on, because moving them to other chips can be slow and complicated while keeping performance high. Gimlet aims to make that move more like plugging an appliance into different outlets (same device, different power sources) rather than rebuilding the appliance.

Why it matters

AI services are getting expensive to run, and chip supply is uneven. If AI software can use whatever hardware a company already has, including a mix of laptops, servers, and cloud machines, it could reduce costs and make AI features easier to scale.

Source: TechCrunch AI