NVIDIA just made a $20 BILLION move to stay on top of the AI market

While most of us were distracted by the holidays, NVIDIA quietly struck a massive deal with Groq, the startup behind ultra-fast LPUs (Language Processing Units). These chips are designed specifically for text generation, allowing AI models to run up to 10x faster and using 10x less energy than standard GPUs.

But the twist is that this is NOT technically an acquisition. To sidestep regulatory scrutiny, NVIDIA is calling it a “non-exclusive licensing agreement” while essentially hiring Groq’s top talent, including founder Jonathan Ross, who built Google’s TPUs that are in intense competition with these new LPUs. 

It's a whole new level of the growing AI web of companies, which I've covered before. 

The whole story is in my latest YouTube video.

#NVIDIA #AI #Groq #AInews

Matt Wolfe

Nvidia structured a $20 billion deal with AI chip startup Groq as a licensing agreement rather than an acquisition to avoid regulatory scrutiny. Groq's LPUs (language processing units) run AI models 10x faster with 10x less energy than traditional GPUs. The deal brings Groq founder Jonathan Ross, who previously created Google's TPU chips, to Nvidia. This move comes after Google demonstrated it could train Gemini 3 entirely on its own TPU chips without Nvidia hardware, threatening Nvidia's 90% market dominance. The licensing structure maintains the appearance of competition while giving Nvidia access to Groq's technology and key talent.

Nvidia's $20B Loophole Explained