Tesla would cover about $830 million of the expansion of a Samsung Electro-Mechanics plant in Sejong, South Korea. In return it would receive, from 2029 to 2033, the substrates that link its AI chips to the rest of the electronics in its self-driving cars and Optimus robot.
Falling share prices, withdrawn IPOs and wider spreads on credit default swaps marked the week for AI infrastructure players. The strain sets the stage for upcoming earnings from the big cloud providers.
Nvidia has moved from selling chips to helping finance the AI build-out. With neocloud providers under market pressure, some investors think the company has taken on outsized risk.
The AI boom is setting local residents against corporations building energy-hungry data centers. In Newport, South Wales, people living beside a towering new facility are fighting back.
Enterprise AI is entering a more cost-conscious phase as inference spending rises. Business leaders are weighing where models run, which ones they use and how much data they are prepared to entrust to third parties, and those choices are reshaping purchasing decisions.
OpenAI explains how it scaled Habitat, its storage system for ChatGPT, with reliable multi-tenancy, global caching and Azure Cosmos DB. The company says reads became 12 times faster.