Zitron: Apple's $14B AI spend bets on a bubble collapse

Reading Ed Zitron's interview in MacRumors about Apple's AI stance, the number that stops you is the ratio. The hyperscalers, in his framing, are collectively spending north of $650 billion on AI infrastructure this year. Apple is spending about $14 billion. That is not a rounding difference, it is a completely different bet about where the value in this cycle finally lands.

Zitron, whose newsletter Where's Your Ed At and podcast Better Offline have made him one of the loudest skeptics of the boom (Politico's Digital Future Daily called him the AI boom's 'acerbic gadfly'), thinks Apple 'will sit on the sidelines and watch everything burn'. His argument leans on unit economics. Citing SemiAnalysis research he points out that a heavy user can 'burn hundreds of dollars on a $20-a-month subscription', and he notes that 'OpenAI lost $20.9 billion on $13.07 billion in revenue in 2025'. If per-token cost never drops under the per-month price, the whole subscription structure has to reprice or fail.

Against that backdrop, Apple's choices look less lazy than deliberate. It is 'paying Google around a billion a year for Gemini to run Siri' while doing what it can on-device. And on Apple Intelligence itself, Zitron is brutal, calling it 'a mass-radicalization of its users against AI', an attempt to cram a barely-functional series of add-ons at people who did not ask for them. If that reading is right, Apple's visible pullback on deeper AI commitments was a reaction to how badly the first push landed.

The honest caveat is that this is one critic's read, and Zitron's prior is that the whole thing is a bubble. If inference costs keep falling and enterprise revenue scales, the same $14B versus $650B gap flips from prudent to a multi-year capability deficit, and the Google-for-Siri dependency hardens into a permanent tax. What the reporting does not give you is the actual shape of that Google contract, or what Apple's originally-deeper AI plan looked like before it was pulled.

The forward-looking bit is where the argument gets interesting. Zitron 'could see them doing some choice acquisitions as things begin to collapse'. Whether or not you buy the bubble thesis, a company with Apple's balance sheet gets optionality that the hyperscalers, locked into announced capex, structurally do not. Restraint, if the timing works, is a form of dry powder.