Marketplace Tech Bytes: Week in Review – New chip exports for China, Microsoft to pay electricity for AI data centers, and Gemini will power Apple’s AI
5 snips
Jan 16, 2026 Anita Ramaswamy, a tech journalist and columnist at The Information, joins to discuss cutting-edge developments in AI and data centers. They delve into Nvidia's new export rules for its H200 chips to China, highlighting the 25% fees and security requirements involved. The conversation shifts to Microsoft’s significant commitment to cover energy costs for AI data centers, and the political backlash surrounding these massive power demands. Finally, they explore Apple's decision to leverage Google’s Gemini for Siri, raising questions about AI competition.
AI Snips
Chapters
Transcript
Episode notes
New Limits On Nvidia Chip Exports
- The U.S. allows NVIDIA to export H200-class chips to China under strict rules and a 25% fee to the U.S. government.
- The rule limits China to less than half of chips sold to U.S. customers and aims to control advanced AI hardware flows.
China Both Buys And Builds AI Chips
- NVIDIA CEO Jensen Huang has pushed to keep China reliant on American AI infrastructure because China remains a major market.
- China is simultaneously accelerating domestic chip development and restricting some imports to build local capacity.
AI Data Centers Strain Local Grids
- Hyperscaler AI data centers consume power comparable to entire cities and often trigger new transmission and generation projects.
- Those infrastructure costs typically get shared with local ratepayers, raising community concerns about bills and resources.