Skip to content
Companies

Groq

Inference cloud running open AI models on its own chips

Raised to date

$672.2M

of $672.2M offered across 1 Form D filing

Investors
144
Last filed
Jul 24, 2026

About

Groq operates an AI inference cloud, describing itself as a neocloud built for fast, low-cost inference. It runs a global fleet of data centers across North America, Europe, the Middle East and Asia-Pacific serving several million developers and processing trillions of tokens per week. The platform is sold in tiers spanning bare-metal infrastructure, hosted inference on open models, and enterprise-governed deployments. In 2026 the company raised roughly $650 million led by Disruptive and Infinitum and licensed its inference technology non-exclusively to Nvidia while continuing to operate GroqCloud.

Groq built its own inference silicon (the LPU) and runs the full stack itself, so it sells latency and cost-per-token characteristics that general-purpose GPU clouds do not match, packaged as a plain API rather than a hardware purchase.

Funding

  • EquityFiled Jul 24, 2026Form D
    $672.2M/ $672.2M

Product

GroqCloud

AI infrastructure / inference platform

A hosted inference service that runs popular open AI models on Groq's own accelerator hardware, accessible to developers through an API with production-grade capacity across 13 data centers.

  • Inference on Groq's custom LPU hardware rather than general-purpose GPUs
  • Global data center footprint across North America, Europe, the Middle East and Asia-Pacific
  • Developer API used by millions of developers running trillions of tokens weekly

Built for

AI application developers and enterprises deploying large language models in production, plus sovereign and regional AI infrastructure buyers

People on file