Deterministic inference silicon built for low-latency token generation.
Groq designs language processing units and operates a cloud inference service, competing on latency and throughput for open-weight models rather than on training workloads.
Open roles moved from 69 to 92 over the trailing 30 days, with compiler engineer roles leading the increase.
A jump this size is a funded programme, not backfill — expect new teams rather than replacements.
The inference provider is serving the newly released open model at launch, publishing throughput benchmarks against GPU baselines.
Day-one availability is how inference providers win developer defaults.
Estimated employees moved from 447 to 540 across the trailing 90 days, based on public profile and directory changes.
Headcount moving this fast means new teams are being stood up, not backfilled.
Median posted compensation for compiler engineer roles moved from about $214K to $245K across 40 observed postings and reports.
Pay moving this fast usually signals a supply shortage rather than a general raise.
No investment relationships tracked.