Groq raises $650M to become inference cloud after Nvidia deal
Groq pivots from chip maker to cloud services after licensing its LPU to Nvidia for $20B, raising $650M from existing backers to build inference infrastructure.
Tenstorrent's $110K AI Server Challenges Nvidia's Inference Dominance
Tenstorrent launches Galaxy Blackhole inference servers at 3-5x cheaper per node than Nvidia DGX, with 16 units already deployed at Equinix and performance claims that undercut the GPU+LPU disaggregation trend.
Tesla AI5 chip tapes out, first stop is Optimus, not cars
Tesla taped out its AI5 self-driving chip on April 15, 2026, with 8x the compute of AI4, but it's deploying first to humanoid robots and data centers, not the vehicle fleet—a signal about where the real AI work actually is.
SambaNova and Intel Ship Production Agentic AI Chip Stack
SambaNova and Intel announced a signed, production-ready heterogeneous inference architecture combining GPUs, Xeon 6 CPUs, and RDUs for agentic AI, deploying in standard data centers by H2 2026.