NVIDIA’s Big Bet: Stacking Groq LPUs on Next-Gen GPUs by 2028
NVIDIA’s plan to integrate Groq’s LPU technology could reshape its AI chip roadmap. The reported strategy involves stacking separate LPU dies on next-gen Feynman GPUs using advanced TSMC packaging, similar to AMD’s 3D V-Cache. This move is a direct play to control the lucrative AI inference market.