After Nvidia’s $20B not-aqui-hire, AI chip startup Groq reportedly raising $650M - TechCrunch

Following a $20 billion 'not-a-qui-hire' deal with NVIDIA, AI chip startup Groq is reportedly raising $650 million in new funding. The round aims to support Groq’s expansion into its inference neoclou... Read original →

NVIDIA’s $2B ‘non-acquisition talent deal’ with Groq is a strategic hedge against surging AI inference demand, effectively decoupling inference-optimized silicon from training-centric architectures. This accelerates a full-stack shift toward heterogeneous compute, especially benefiting compiler and runtime co-design. Groq’s pivot to an inference neocloud—backed by Infinitium and Disruptive—sidesteps immediate advanced-node geopolitical exposure but exposes long-term vulnerability: its LPU architecture, built on mature nodes, may hit energy-efficiency ceilings just as U.S.-China tech restrictions threaten cross-border IP licensing. Competitors like AMD could counter with FPGA-based inference solutions via Xilinx, while Taiwan, China foundries may capture more ASIC outsourcing. Within 18 months, ‘inference-as-a-service’ will become the new battleground; without breakthroughs in model quantization-aware hardware, Groq’s early lead risks erosion by hyperscaler in-house chips.