July 27, 2026
Bubble trouble in Cupertino
Apple Will 'Watch Everything Burn' When the AI Bubble Bursts
As prices climb, commenters are split between ‘finally, a reality check’ and ‘this guy’s just doomposting’
TLDR: Ed Zitron argues the artificial intelligence boom is wildly overpriced and could push up Apple device costs if the whole thing falls apart. Commenters were sharply divided, with some praising a rare skeptical voice and others calling the argument outdated, thin, and classic doomposting.
Apple’s latest price creep just got handed a full-blown villain origin story: critic Ed Zitron says the artificial intelligence gold rush is so expensive and so shaky that when it finally pops, Apple may just have to “watch everything burn.” His basic case is simple enough for anyone following along at home: these chatbot-style tools cost a fortune to run, customers don’t really want to pay the true price, and companies keep papering over the gap with subscriptions and giant spending sprees on new server farms. If he’s right, all that spending could boomerang back into the gadgets people actually buy, from Macs to iPhones.
But the real fireworks were in the comments, where readers instantly split into Team Finally Someone Said It and Team This Is Cherry-Picked Doom. One camp cheered Zitron for pushing back against what one commenter called a “gushing torrent of AI company propaganda,” basically casting him as the cranky truth-teller at the party. The other side was not having it. Critics accused him of using old numbers, making leaps about what these services really cost, and ignoring signs that some companies may already be improving their finances. Another reader nitpicked his point that AI coding tools make programmers slower, arguing the evidence is stale because the tech moved fast.
The vibe? Less calm debate, more food fight with spreadsheets. Even without big meme jokes, the thread had that classic internet energy: half the room yelling “bubble,” the other half yelling “skill issue.”
Key Points
- •The article presents Ed Zitron's view that the economics of large language model businesses are fundamentally broken because token-based costs do not align with standard software subscription pricing.
- •It says AI providers sell subscriptions that may allow usage worth far more than the subscription price, citing SemiAnalysis and Zitron's reported OpenAI financial figures for 2025.
- •The article reports that Anthropic and OpenAI shifted enterprise customers to token-based billing and uses Uber as an example of rapid budget consumption and difficult ROI tracking.
- •It argues that many AI startups, including Perplexity, Cursor, and GitHub Copilot, remain pressured by per-token costs while offering products with limited differentiation.
- •The piece says AI data centers are capital-intensive, slow to build, and dependent on a small number of major AI tenants, particularly Anthropic and OpenAI.