Gimlet Labs, which helps customers divide AI tasks across multiple chip types, raised $300M led by a16z at a $3B valuation, six months after an $80M Series A
The company, which helps divide AI tasks among different chips, raised $300 million in its latest round
Context & Ripple Effects
Gimlet Labs introduced its multi-silicon inference-cloud positioning with an $80 million Series A in March 2026, framing its product around running AI workloads across different hardware types. The new financing sharply increases the resources behind that approach.
a16z’s lead role fits its stated focus on AI infrastructure investing, while Etched’s $300 million July round shows that investors are also funding companies tied to the inference-hardware layer.
First-order effects
- Gimlet Labs gains $300 million to build and sell software that allocates customer AI tasks across multiple chip types, rather than tying deployments to one hardware platform.
- a16z deepens its exposure to AI infrastructure through a company whose product sits between customers’ workloads and their underlying chips.
Second-order effects
- Inference-chip vendors and cloud providers face greater pressure to demonstrate that their hardware can be incorporated into multi-chip deployments, where workload placement can become a purchasing lever.
- Gimlet’s better-funded platform approach heightens competition with inference-focused startups such as Etched for investor attention and customer budgets tied to serving AI models.
Third-order effects
- If multi-silicon inference platforms gain adoption, the value in AI infrastructure may shift toward software that brokers among chips rather than solely toward the chip vendors themselves.
- The financing points to an AI hardware market in which customers seek optionality across accelerators, making interoperability a strategic feature of inference infrastructure.
The trend: AI-infrastructure funding is expanding beyond chip design into software layers that let customers use heterogeneous compute more flexibly.