The post compares two visions of personal computing after the current chatbot boom. Apple’s pitch is that the phone stays central and AI gets folded into the existing device and app model. Microsoft’s pitch is looser and more cloud-first, with users moving among thin devices that act as portals to remote agents. The piece argues Apple can still win even after fumbling its 2024 Siri and Apple Intelligence rollout, because most consumers do not want a radically new computing model. They want their phone to keep doing familiar things with less friction.
That basic framing landed better than the headline. Most people thought "last stand" was clicky nonsense and read the actual situation as Apple holding position, not facing collapse. The strongest consensus was that consumer demand for AI is narrower and more practical than the industry pitch suggests. People do use ChatGPT and coding tools when they choose to, but they do not want operating systems rebuilt around chatbots, agent demos, or a swarm of ambient devices. What they want is boring competence. Siri that understands a request every time. Password changes done automatically. Summaries, search, and writing help that stay out of the way. In that framing, Apple’s instincts still look sane even if its execution was late and messy.
A second theme was that Apple’s delay was less about not understanding AI than about refusing to ship the kind of flaky assistant behavior that competitors have already normalized. Several commenters pointed to Samsung and broader
Copilot-style rollouts as proof that shipping early mostly created annoying, low-trust features. Apple still got hammered for announcing too early, but many saw the eventual product direction as more credible precisely because Apple has more to lose from hallucinations and broken workflows than companies whose users already expect churn.
The most concrete anxiety was economic, not strategic. Apple’s new AI stack appears split across on-device models,
Private Cloud Compute, daily usage limits,
iCloud+ upsells, and special terms for smaller developers. Commenters could not pin down who pays when a third-party app uses Apple-hosted models, what gets metered, or whether users and developers will be competing for the same quota. That ambiguity made some of the platform story feel unusually un-Apple. If the cloud side is constrained, expensive, or tied to Apple’s own bundle economics, developers may default to
OpenAI,
Anthropic,
Google, or open models and use Apple only for lightweight local tasks.
Microsoft came off worse. Few people found its hardware-and-agent vision compelling in practice, and several argued that big companies always describe the future in ways that fit what they are able to sell. Apple sells integrated hardware. Microsoft sells subscriptions and cloud services. That makes its "ambient thin device" future look less like destiny and more like a sales strategy. Even people who agreed that hosted services and thin clients will grow were skeptical that Microsoft, specifically, will be the company to make consumers want them. The result was a fairly stable conclusion: the iPhone is not in a last stand. It is still the default consumer computer, and AI will likely matter there as infrastructure and workflow glue, not as a clean break from the phone itself.
stratechery.com