By Naba 
Nvidia recently struck a $20 billion deal to acquire the brains of a fierce rival, marking its biggest move to date in the AI arms race. However, the massive chip manufacturer may soon have to respond to more difficult inquiries about the future of its chips.
CEO Jensen Huang offered further detail in an internal email obtained by CNBC.
“We plan to integrate Groq’s low-latency processors into the NVIDIA AI factory architecture, extending the platform to serve an even broader range of AI inference and real-time workloads.”
The move comes as Nvidia is getting more attention in Asia. Megaspeed International, a fast-growing Singapore-based importer of Nvidia chips, is being looked into for possibly smuggling banned H100 and H200 chips into China, according to a recent Bloomberg report.
The world’s most valuable chipmaker is now stuck between two extremes: It has to deal with rising geopolitical threats and regulatory difficulties in one of its hottest regions while still dominating AI infrastructure around the world.
Groq, a nine-year-old firm started by former Google TPU engineers, was never for sale — at least not in public. But Nvidia came in with a $20 billion deal that includes licensing Groq’s cutting-edge AI inference technology and bringing on its top leaders, such as CEO Jonathan Ross.
Sources informed CNBC that Nvidia is really buying all of Groq’s assets, except for a small GroqCloud firm, even though Groq called the purchase a “non-exclusive licensing agreement.”