About Groq
Groq is a US AI chip and inference company that designed the Language Processing Unit (LPU), a deterministic processor built for fast, low-latency inference of large language models. It offers GroqCloud, an API platform serving open models, and GroqRack on-premises systems. Groq was founded in 2016 by Jonathan Ross, a co-creator of Google's TPU, and is headquartered in Mountain View, California.
Is this your company? Claim this profile
What you can check about Groq
Reviews and answers are moderated before they appear. Vendor staff can’t review their own company.
Building (how this is worked out)
Well evidenced: 5 or more published reviews plus 5 or more answered questions or community mentions. Some evidence: at least 1 published review, or 3 or more answered questions and community mentions combined. Building: anything less. Counts include published, moderated material only.
- Published reviews
- None yet
- Answered questions
- None yet
- Community mentions
- None yetThreads that name this company
- Latest story
- None yet
Evidence is still building for Groq. Used it in production? A short, moderated review helps other teams decide with more confidence.
Profile not yet verified by our editors. How we check profiles
See alternatives to Groq → Other companies in HPC & Accelerated Compute, busiest first.
More about Groq
Groq sells AI inference as a service through GroqCloud, where developers call open models such as Llama, Qwen and others through an OpenAI-compatible API, and supplies LPU-based systems for data centre and sovereign AI deployments.
In HPC & Accelerated Compute, Groq's portfolio centres on the LPU, a software-defined, compiler-scheduled architecture using on-chip SRAM, packaged into GroqCard, GroqNode and GroqRack systems.
Groq was founded in 2016 by Jonathan Ross and other former Google engineers and is privately held. In December 2025 Groq entered a non-exclusive licensing agreement with NVIDIA for its inference technology, under which Ross and other staff joined NVIDIA while Groq continued to operate GroqCloud as an independent company.
Frequently asked questions
What does Groq do?
Groq designs the LPU AI inference processor and operates GroqCloud, an API service for running large language models at low latency.
What is the Groq LPU?
The Language Processing Unit is Groq's inference processor, which uses a deterministic, compiler-scheduled design and on-chip SRAM to deliver fast token generation.
Is Groq related to xAI's Grok?
No. Groq, the chip and inference company, is unrelated to Grok, the chatbot developed by xAI.
Who competes with Groq?
Groq competes with NVIDIA and other AI accelerator and inference providers such as Cerebras, SambaNova and cloud inference services.
Something wrong or out of date? Suggest a correction. We reply within 5 business days.
