Frontier Rounds— AI funding intelligence

Gimlet Labs Funding History

Gimlet Labs provides cloud infrastructure for running AI agent inference. Its platform distributes workloads across chips from vendors including Nvidia, AMD, Cerebras and d-Matrix.

Company Overview

Gimlet Labs logo

Gimlet Labs

  • Model Inference & Deployment

Founded 2023 · San Francisco, California, US

Official Website

Gimlet Labs Funding Timeline

  1. Announced Sep 4, 2026

    Series B

    US$300 million

    Gimlet framed the investment around making model inference more efficient as AI workloads and infrastructure demands grew.

    Led by Andreessen Horowitz. Participation included Sapphire Ventures, Menlo Ventures, Eclipse, Emergence Capital, Factory, M12, Prosperity7 Ventures, Samsung Venture Investment, Tiger Global, Triatomic Capital.

    Source · gimletlabs.ai ↗

  2. Announced Oct 22, 2025

    Seed

    US$12 million

    Gimlet Labs disclosed the seed financing when it launched a platform that distributes AI-agent workloads across different kinds of hardware.

    Led by Factory.

    Source · globenewswire.com ↗

Gimlet Labs Founding Team

Gimlet Labs reunited members of the Pixie Labs team to work on running inference across different kinds of chips. Co-founder Michelle Nguyen led engineering, Omid Azizi led hardware platforms, and James Bartlett worked on system architecture.

  • Zain Asgar became CEO of Gimlet Labs after leading Pixie Labs. He has also taught at Stanford.
  • Natalie Serrino led work on KForge, a system for generating GPU kernels.

Gimlet Labs Products

  • Gimlet Cloud

    Distribute model-serving workloads across more than one chip architecture.

Gimlet Labs Company Development

Serving Agent Workloads Across Different Chips

Gimlet emerged from stealth in October 2025 with a system for distributing agent workloads across different kinds of chips. Its Cerebras announcement illustrates the next step: expanding the silicon available to the same inference stack rather than requiring customers to rebuild their applications for each accelerator.

Gimlet Labs Competitive Landscape

Inference Across Chips

Modal gives developers programmable access to cloud execution. Gimlet works at the inference-routing layer, where a single workload can be split across accelerators with different capabilities. The comparison matters when hardware utilization, rather than the packaging of Python jobs, determines deployment cost.