Groq's bet on inference-only silicon is finally showing up in enterprise contracts

Groq has spent years arguing that the AI hardware market misjudged its own future by fixating on training FLOPs while underinvesting in inference latency, and the company's recent run of enterprise deployment announcements suggests that thesis is starting to pay off commercially. Groq's Language Pro

Source: TechCrunch

Read the full story on Venture OS