Perplexity AI $20 Billion Nvidia Masterstroke: Groq Duo + LPUs
8 snips
Jan 6, 2026 A major $20 billion licensing deal marks Nvidia's strategic move with Groq’s CEO-President duo. The conversation dives into Groq's language processing units, boasting impressive speed and energy efficiency. The podcast also explores Nvidia's competitive stance against Google’s TPU, revealing the dynamics in the AI chip market. Additionally, Social Capital's successful early investment is highlighted, projecting an astonishing return. Public reactions to the deal offer insight into the tech community's sentiments, making for an engaging discussion.
AI Snips
Chapters
Transcript
Episode notes
Acquihire-Like Licensing Play
- Nvidia paid $20B to license Groq's tech and hire its leadership without a formal acquisition.
- Jaeden Schafer frames this as a strategic move to gain LPU tech while avoiding acquisition scrutiny.
LPUs Promise Major Efficiency Gains
- Groq claims LPUs are ~10x faster and use one-tenth the energy of GPUs for inference.
- Jaeden Schafer highlights LPUs as a direct performance and efficiency threat to Nvidia's GPU dominance.
Defensive Move Against Cloud TPU Threat
- Nvidia likely viewed Groq as the main challenger that could enable Google-like TPU advantage at scale.
- Jaeden Schafer suggests Nvidia moved to neutralize a competitor that could be added to cloud offerings.