Ed Zitron has spent two years telling anyone who’ll listen that the AI industry burns cash it will never recoup. In a new interview with MacRumors, he turns that argument on Apple. The answer isn’t the one Cupertino’s marketing team wants circulating.
Apple spent roughly $14 billion on AI infrastructure this year. Hyperscalers spent north of $650 billion. That gap used to read as Apple falling behind. Zitron’s read is blunter: Apple refused to play a game with no winners, and it’s still getting billed for everyone else’s table stakes — the same gap between AI spending and AI payoff that keeps widening across the industry.
The Bill Comes Due Either Way
Memory prices have roughly doubled in 2026. Hyperscalers are hoovering up DRAM supply for data centers that, per Zitron, mostly serve OpenAI and Anthropic rather than any broader wave of enterprise demand. Tim Cook has already called Apple’s price increases “unavoidable.” Macs and iPads went up first. iPhones are next.
That’s the part worth sitting with. Apple avoided the capex arms race. It kept Apple Intelligence thin and rented its AI dependency to Google for about a billion dollars a year to run Siri. It still passes a bubble-inflated hardware markup to customers who never asked for any of it. Restraint on the spending side didn’t buy insulation on the pricing side.
Nobody Can Say What Apple Is Behind On
Zitron’s sharpest line isn’t really about money. It’s about the premise that Apple was “falling behind” in the first place. Ask what capability Apple is missing, and nobody has an answer. Apple Intelligence’s early stumble — summaries that became a punchline, a Siri rebuild that shipped worse than the one it replaced — did something useful almost by accident. It soured Apple’s own userbase on AI features before the company sank real money into them.
Compare that to OpenAI, which Zitron’s own reporting pegs at a $20.9 billion loss on $13.07 billion in revenue in 2025. Or to Uber, whose COO admitted the company blew through its entire annual AI token budget in a single quarter after switching to usage-based billing. Apple’s problem was never technical debt. It was a UX misfire that happened to double as a hedge.
Where the Contagion Actually Lands
The interview’s most useful contribution isn’t about Apple at all. It’s about who eats the loss when the buildout doesn’t pay for itself. Private credit funds finance most of these data centers through project-level debt, and pension funds like CalPERS or San Francisco’s teachers’ fund back those private credit funds in turn — retirement money quietly exposed to a bet it never chose to make.
Zitron singles out Oracle as the clearest domino. The company has committed $340 billion-plus in AI infrastructure bets, financed on the assumption that OpenAI becomes the most profitable company on Earth by 2030. Taiwanese ODMs like Quanta and Foxconn ride the same wave up, and in his telling, down too — though Foxconn’s Apple revenue gives it more cushion than most, part of the wider billion-dollar exposure a bubble pop would trigger across the supply chain.
The Vision Pro Tell
Ask Zitron what Apple should do with its restraint, and he doesn’t say “more AI.” He says Vision Pro — not as a hit, but as the one genuinely new interface idea to come out of Silicon Valley in years. The device needs to be functionally weightless to work, and it isn’t yet.
That’s the real thesis buried in the interview. The AI bubble exists, in his framing, because the industry ran out of new interfaces and needed a story to tell investors instead. Apple’s edge was never spending less. It was staying solvent and skeptical enough to still be standing, and maybe even shopping, once the story stops working.
Related: OpenAI’s First AI Device Hits a $6.5 Billion Legal Battle With Apple
