Gruber's Provocation: AI Is a Technology, Not a Product

There’s a sentence ricocheting through tech Twitter right now: “AI is a technology, not a product.” It came from John Gruber, the longtime Apple-watcher behind Daring Fireball. Sounds like wordplay. It isn’t. It’s a precision strike at the entire Big Tech AI playbook.

The Distinction That Matters

Gruber’s argument is almost embarrassingly simple. Electricity isn’t a product. Neither is the internet, or a database. We don’t run ads telling people to “buy electricity.” We sell refrigerators and lamps that run on it.

Yet here we are in 2026, watching every keynote brag that their app “has AI in it.” The pitch leads with the ingredient, not the meal. Gruber’s point lands because nobody seems to have noticed how strange that is.

Why Apple Made This Hit Harder

The reason this critique stings is that it’s coming from inside the house. On April 1, 2026, the Wall Street Journal published prototype footage of Apple’s reworked Siri — it cleared 1.39 million views. Around the same time, a YouTube essay titled “Apple’s AI Delay: Credibility in Crisis” started making the rounds.

Apple Intelligence was announced at WWDC 2024 with the swagger only Apple can muster. Then the Siri overhaul slipped. And slipped again. Features quietly turned into vaporware. When Gruber says AI isn’t a product, he’s partly aiming at the company he’s spent two decades defending.

The “AI-Powered” Trap

Run the tape on any recent launch event — Microsoft Copilot, Google Gemini, Meta AI. More than half the runtime goes to how big the model is, how many benchmarks it tops, how clever the architecture is.

That’s not what users are shopping for. People want to write emails faster. Nobody, anywhere, wants a 175-billion-parameter model. The second your demo turns into a spec sheet, you’ve lost the room. That’s the gap Gruber is pointing at — and it’s why most AI marketing feels like it’s shouting into a void.

What an Actual AI Product Looks Like

Here’s the tell: the AI products that actually work barely mention AI. GitHub Copilot sells itself as “code faster.” ChatGPT led with conversation — a thing humans already understood — not with transformer architectures. The framing was the verb, not the engine.

Compare that to “AI phones,” “AI PCs,” and “AI search” — categories where the technology name got slapped directly onto the product name. Most are drifting. Users don’t buy technology. They buy solutions. It’s a sentence worth printing on every product manager’s wall.

The Question Every Roadmap Should Answer

Strip Gruber’s provocation down and it becomes a single question: what problem does your AI feature actually solve for a real person? If you can’t answer that in one sentence, you don’t have a product yet. You have a press release.

The back half of 2026 is shaping up to be another AI launch parade. The interesting question isn’t who ships the biggest model. It’s which company stops saying “now with AI” and starts saying “here’s the problem we solved.” That pivot might be the one that decides who wins the next cycle.