Week in AI: Chips, Copilot and control

This week’s AI news centered on a practical question for companies building with AI: how much of the stack can you really control? The biggest stories touched compute supply, model partnerships, legal exposure, regulation, and the growing push toward owned AI infrastructure.

SK Hynix raises $26.5B

SK Hynix raised $26.5B in the biggest foreign IPO in US history and was urged to build new US fabs. For AI builders, the story matters because compute supply continues to shape what can be built, scaled, and delivered reliably. Chips are not just a cost center. They are becoming a strategic dependency.

OpenAI says GPT 5.6 is preferred for Copilot 365

OpenAI said GPT 5.6 is the preferred model for Microsoft Copilot 365 amid breakup chatter. The news reinforces how much enterprise AI still depends on a few major model providers and platform partnerships. For companies adopting AI, model roadmaps and vendor relationships remain core planning risks.

Apple sues OpenAI

Apple sued OpenAI over alleged trade secret theft. The headline is important because it highlights rising legal risk around AI, hardware, and intellectual property. As AI development moves closer to custom chips, devices, and proprietary systems, companies will need stronger governance around talent, data, and confidential technical work.

The EU warns Meta on autoplay and infinite scroll

The EU told Meta to disable autoplay and infinite scroll or risk massive fines. This matters beyond social media. It signals that product mechanics designed to maximize engagement can become regulatory concerns. AI product teams should expect design choices, recommendation systems, and user experience patterns to face more scrutiny.

Hugging Face argues companies are done renting AI

Hugging Face’s CEO said open source AI matters more than ever and argued that companies are moving away from simply renting AI. The point reflects a broader industry shift. More teams want flexibility, portability, and ownership across models, data, and deployment environments.

The takeaway

The week’s pattern is clear. AI advantage is moving beyond model access alone. Companies now need to think about compute, legal risk, regulation, and ownership of the stack. The winners will be the teams that treat AI as infrastructure, not just software.