G
Infrastructure
Groq
Fast, low-cost AI inference for everyone (LPU).
Valuation$6.9B
Funding$3B+ raised
FlagshipLPU Inference Engine / GroqCloud
Founded2016 · Mountain View
71
Bright Sentience Score
Rank #21 of 23
What they're building
Inference economics may matter almost as much as model quality. Groq's Language Processing Unit delivers ultra-fast, lower-cost inference for open-weight models, and the company has raised billions to build out GroqCloud data centers in the US, Europe and the Middle East.
Latest developments
- Raised $750M at a $6.9B valuation (September 2025).
- GroqCloud serves open-weight models from Meta, Mistral, Moonshot and DeepSeek at record token speeds.
- Multi-billion-dollar infrastructure deal in Saudi Arabia.
Why it matters
When every product runs on tokens, whoever serves them fastest and cheapest wins a share of everything. Groq is the leading challenger to Nvidia on inference.
Key facts
Leader
Jonathan Ross
FoundersJonathan Ross (inventor of Google's TPU)
PeopleJonathan Ross (CEO)
HQMountain View, USA
Latest modelLPU v2 inference chips
Valuation note$750M round (September 2025)
InvestorsBlackRock, Disruptive, Cisco, Samsung Catalyst, Neuberger Berman, D1
WebsiteNearby on the leaderboard