Key Partners
- Open-weight model labs hosted on GroqCloud (Llama, GPT-OSS, Qwen, Whisper — docs).
- NVIDIA non-exclusive licensing / LPX platform path (company news Jun 2026).
- Data-center operators across North America, Europe, MEA, APAC (13 DCs).
- Bell Canada exclusive inference for Bell AI Fabric (May 2025 newsroom).
- Infra investors and growth capital (Disruptive, Infinitum — $650M claim).
- Developer ecosystem around OpenAI-compatible API and playground.
Key Activities
- Operate LPU / LPX inference clusters in 13 data centers.
- Run GroqCloud Console, OpenAI-compatible API, playground.
- Host production models — GPT-OSS, Whisper, Qwen, Llama enterprise, Compound.
- Ship Developer features — Flex, Batch, spend limits, prompt caching.
- Deliver Enterprise — custom models, regional endpoints, LoRA inference.
- Expand capacity (company — toward 200 MW by end of 2027).
Key Resources
- LPU / LPX inference accelerators and compiler stack.
- GroqCloud — public, private, co-cloud instances.
- 13 data centers · claimed >5M developers (Jun 2026 newsroom).
- Production model catalog + Compound tool-using systems.
- Enterprise controls — LoRA, SSO/SCIM, audit logs, regional endpoints.
- $650M growth capital (company news Jun 2026).
Value Propositions
- LPU inference speed + predictable cost-per-token.
- Free Console land → Developer pay-per-token PLG.
- OpenAI-compatible API for easy migration.
- Hosted open weights — GPT-OSS, Whisper, Qwen, Llama enterprise.
- Enterprise — custom models, LoRA, regional endpoints, SSO.
- GroqRack / private / co-cloud for sovereignty and air-gap.
Customer Relationships
- Self-serve Free signup at console.groq.com.
- Developer upgrade — payment method + chat support.
- Usage dashboard and spend limits for cost control.
- Docs, models list, playground for builders.
- Enterprise — dedicated support, LoRA access, SSO/SCIM.
- Expansion path — Free → Developer → Enterprise / Rack.
Channels
- Primary — groq.com + console.groq.com Free signup.
- Developer docs and OpenAI-compatible API endpoint.
- Playground / demos for speed proof.
- Direct Enterprise sales for dedicated / LoRA / Rack.
- Partner programs (e.g. Bell Canada AI Fabric).
- Company newsroom and developer community amplification.
Customer Segments
- Indie builders and students on Free Console.
- Startups / AI-native teams on Developer PAYG.
- Product orgs needing OpenAI-compatible fast inference.
- Enterprises — custom models, LoRA, SSO/SCIM.
- Regulated / air-gapped buyers on GroqRack.
- Sovereign / telco partners (e.g. Bell Canada).
Cost Structure
- LPU / LPX hardware + power / colocation COGS.
- Free-tier inference as demand-gen cost.
- R&D — compiler, GroqCloud, Compound, Enterprise features.
- Sales & marketing — PLG content + enterprise field.
- Support / success for Developer chat and Enterprise CSM.
- Compliance — SOC 2, GDPR, HIPAA, private tenancy overhead.
Revenue Streams
- Free plan — $0 demand gen (FACT — plans).
- Developer PAYG tokens — card/bank required (FACT — billing).
- GPT-OSS 20B $0.075/$0.30 · 120B $0.15/$0.60 per 1M (FACT — docs).
- Whisper Turbo $0.04/hr · Whisper V3 $0.111/hr (FACT).
- Enterprise contracts — custom models, LoRA, SSO (sales).
- GroqMetal / GroqRack dedicated and on-prem capacity.
- Partner / sovereign deals (e.g. Bell Canada).