Trending:
AI & Machine Learning

Apple questions its AI strategy internally as memory constraints favor training-focused rivals

Apple executives are questioning whether the company has what it takes to compete in an AI-first era, according to Bloomberg's Mark Gurman. The introspection comes as memory chip suppliers prioritize AI training infrastructure over consumer electronics—a shift that could complicate Apple's hardware roadmap.

Apple questions its AI strategy internally as memory constraints favor training-focused rivals

Apple executives are privately questioning whether the company can compete in an AI-first world, Bloomberg's Mark Gurman reports. The doubt is notable timing: Apple just posted a record quarter, but the memory supply chain is moving against it.

The real issue isn't Apple's revenue—it's where memory chips are going. Suppliers like Samsung and SK Hynix are prioritizing high-bandwidth memory (HBM) for AI training chips over consumer electronics. HBM3 and HBM4 production is maxed out serving hyperscalers and AI companies building data centers. Apple, despite ordering 19 billion chips from TSMC's US factories, competes for the same DRAM and HBM that powers training clusters.

This is a supply chain power shift. Apple built its moat on vertical integration and just-in-time inventory—2023 saw finished goods inventory drop 12% despite disruptions. That model works when you're the biggest buyer. It's less effective when AI firms outbid you for scarce components. DRAM prices surged through 2024 and 2025 on AI demand. HBM allocation went to training infrastructure first, consumer devices second.

Apple's $600 billion US manufacturing commitment and supplier diversification matter, but they don't solve the memory allocation problem. The company makes inference chips (M-series, A-series) efficiently. It doesn't need the same HBM volumes as a training cluster. But if suppliers see bigger margins in AI infrastructure, Apple's leverage shrinks.

The iPhone 14 Pro stockouts in 2022 cost Apple $1.5 billion in sales—that was China lockdowns. The current constraint is demand-driven, and suppliers have pricing power. Apple's custom silicon gives it an advantage over generic Android makers. It's less of an advantage when the bottleneck is upstream memory allocation.

Worth noting: No product delays announced yet. Apple's internal doubts predate any public supply issues. But the semiconductor supplier leverage is real, and the AI boom isn't slowing.

Gurman also reports new M5 Pro and M5 Max MacBook Pros are imminent—inventory is tightening—and Apple is exploring a clamshell-style foldable iPhone. The foldable is years out. The memory constraint is now.