Apple Cut 200 Jobs Across Siri and Vision Pro. The Split Tells You Which AI Skills It Thinks It Still Needs.
Apple confirmed roughly 200 layoffs on August 21, split almost evenly between two teams that have nothing obviously in common: Vision Pro and Siri. About 100 roles came out of the Vision Products Group, largely a gaming team that's being shut down along with a reduction in immersive video production. The other roughly 100 came out of Siri and adjacent software teams. Apple wasn't vague about why: the new AI-powered Siri is being built on a different technical architecture than the old one, and the engineers who built the old one aren't automatically the engineers who can build the new one.
That's a specific, dated data point about what a company with essentially unlimited AI budget decided its existing AI and ML talent couldn't carry forward, and it's worth more than another generic "AI is changing jobs" trend piece.
What actually happened, in specifics
Apple's own statement on the cuts, framed around evolving "our business to deliver the best experiences for our users": "While we will create new roles as part of this change, it will also impact a limited number of existing roles. We are grateful to these team members for their contributions, and we are committed to supporting them throughout their transition, including opportunities to apply for other roles at Apple." That's standard language for a reorganization rather than a straightforward headcount reduction, and it's worth reading closely: Apple is describing a swap, new roles alongside cut ones, not a straightforward shrink. Apple was also explicit that Vision Pro and visionOS as products are not being discontinued. What's shrinking specifically is the gaming-focused engineering team within Vision Products Group and part of the immersive video production capacity, not the platform itself.
On the Siri side, the framing matters more than the headcount. This isn't "Siri is being deprioritized." It's the opposite: Apple is investing in a new AI-powered Siri built on different technical foundations, and the people whose expertise was built around the old assistant architecture don't automatically transfer to the new one. That's a swap in required skill set, not a retreat from the product category.
Why the split is the actual story
Most coverage of AI-era layoffs treats headcount reductions as a single undifferentiated signal: company X cut Y jobs, therefore AI is eating jobs or AI spending is cooling. That framing misses what's specific and useful here. Apple didn't cut Siri and Vision Pro because it's spending less on AI. By every available signal, Apple is spending more. It cut a specific team building a specific kind of assistant on a specific architecture, while simultaneously saying it needs different people to build the replacement.
That's a much sharper signal for anyone trying to figure out what AI expertise is actually valuable right now versus what's aging out. "I worked on assistant/NLU systems" was a strong resume line for the better part of a decade. Apple's own internal decision, backed by real severance costs and real reorganization overhead, says that specific expertise wasn't sufficient to carry into whatever Siri becomes next. Companies rarely say this part out loud in public statements, but a reorg that swaps an entire team rather than just growing the new one alongside the old one says it plainly enough.
The context this sits inside
This isn't happening in isolation. Industry-wide tech layoffs crossed roughly 127,000 jobs across more than 280 companies through August 2026, already ahead of all of 2025's total tech job cuts combined, with other trackers putting cumulative 2026 job losses closer to 209,000 workers as of late August. TikTok cut roughly 75 Seattle-area roles the same week. Apple's move is one entry in a much larger pattern of companies restructuring around AI capability rather than simply cutting cost, but it's one of the clearest examples I've seen this year of a company naming the specific skill mismatch out loud instead of hiding behind "efficiency" language.
What this means if you're positioning your own skills
If part of your income depends on AI or ML consulting, or you're deciding what to specialize in next, "which specific architecture and workflow experience survived a reorg at a company that can afford to keep literally anyone it wants" is a far more useful signal than another roundup of AI trend predictions. Apple didn't need to cut the old Siri team for cost reasons. It chose to, because it judged that expertise didn't transfer to what it's building next. That's Apple, with its own resources and its own roadmap, and it doesn't generalize perfectly to every company's stack. But when a company this well-resourced makes a swap instead of an addition, it's worth asking the same question about your own toolkit: is what you know how to build actually the thing that's still being built in two years, or is it the thing a well-funded team just decided to replace wholesale.
The honest take
I'd treat this as a prompt to audit, not a verdict to panic over. If your consulting or product work leans on general LLM integration, prompt engineering, or RAG pipeline work, that's still broadly transferable across architectures in a way that deep expertise in one specific assistant's internals isn't. The narrower and more architecture-specific your expertise, the more exposed you are to exactly this kind of swap, regardless of how good you are at it. Concrete move: if you can't clearly articulate which parts of your current AI skill set would survive your own employer or client rebuilding their product from scratch on different foundations, that's the gap worth closing now, not after the reorg announcement.
The honest counter-take
Apple's Siri rebuild has already slipped multiple times over the past two years, so a team swap here is evidence of intent, not proof the new approach ships on schedule or works better than what it replaces. It's entirely possible this reorg is corrected again in six months if the new architecture doesn't deliver, and reading too much long-term signal into one company's internal staffing decision risks overfitting to a single data point. Treat this as one company's bet on one architecture, not a verdict on which AI skills matter industry-wide.
Author
Lukas
@lukcombinator