This week wasn't about a flashy new gadget — it was about the plumbing underneath the entire tech industry: who controls compute, who controls power, and who gets to set the rules for AI as it gets embedded deeper into everyday systems.
Apple has a new boss for the first time in 15 years
On September 1, Tim Cook officially stepped down as Apple's CEO after leading the company since 2011, moving into a new role as executive chairman. His successor is John Ternus, previously Apple's head of hardware engineering, who becomes the fifth CEO in Apple's history. Cook is staying close to the business for now, largely to help manage Apple's relationships with policymakers, including its ongoing trade and tariff issues tied to manufacturing in Asia.
The timing is tight: Ternus's very first public moment as CEO is Apple's September 9 product event, rumored to include the company's first foldable iPhone. Whatever comes out of that event will effectively be the first real signal of how Ternus wants to run the company differently — especially on AI, an area where Apple has visibly lagged behind Google and OpenAI.
AI's biggest constraint right now isn't chips — it's electricity
Anthropic locked in a multi-billion-dollar, Nvidia-backed compute deal in Texas, part of a broader pattern where AI labs are now signing multi-decade infrastructure contracts instead of simply renting server time. Separately, Texas froze new data-center power hookups after a wave of speculative demand strained the grid, and reports point to Nvidia showing up on multiple sides of these deals at once — as investor, landlord, and chip supplier — a structure that's already drawing antitrust attention.
The pattern is becoming clear: the limiting factor for AI's growth right now isn't model quality, it's whether there's enough electricity and physical infrastructure to run the thing at all.
Regulators are starting to draw real lines
A few decisions this week show governments moving from talking about AI to actually constraining it:
- The EU classified ChatGPT as a search engine, which pulls it under a different, stricter set of regulations.
- New York City barred AI tools for students through eighth grade.
- OpenAI itself rated its upcoming model a critical cyber risk, an unusually blunt admission from a company about its own product.
Jobs are moving, not just disappearing
Uber announced roughly 3,300 layoffs, about a tenth of its workforce, as part of a restructuring aimed at funding its robotaxi push. Advertising giant WPP is preparing to cut close to a thousand more roles by the end of the year as it leans further into AI-driven ad production. This isn't really an "AI takes jobs" headline in the simple sense — it's companies actively shifting budget away from people and toward automation infrastructure, in real time.
Why this matters if you're building software right now
None of this is abstract for developers:
- Compute costs and availability are being shaped by deals between three or four companies — that trickles down to API pricing for everyone building on top of these models.
- Regulation is now real, not theoretical — if you're shipping AI features into a product, assume new compliance requirements are coming, not hypothetical.
- Security expectations just got heavier — a healthcare data breach disclosed this week affected over 9.5 million people, a reminder that the same AI-driven infrastructure race is happening alongside a very ordinary, very persistent security problem: companies still take months to detect and disclose breaches.
The common thread across every story this week is the same: AI stopped being a feature companies add to products, and became the thing entire corporate strategies, national budgets, and power grids are now organized around.
