Through 2027, no rival AI assistant will be a hands-free voice default on a stock EU Android phone despite the DMA order

The prediction

By December 31, 2027, there will still be no way to set a non-Google AI assistant — ChatGPT, Claude, Perplexity, or any other — as a hands-free, voice-invoked default on a stock, out-of-the-box EU Android phone (no rooting, no sideloading, no developer flags). You will be able to install a rival assistant and tap to open it, and by late 2027 it may be able to act inside some apps. But summoning it the way you summon Gemini today — by speaking to the device with the screen off — will not yet work as a shipped, default-eligible capability anywhere in the EU.

If, before the end of 2027, a mainstream EU user can buy an ordinary Android phone and set a non-Google assistant to answer its own wake word out of the box, this prediction is wrong.

Why I think this happens

The prediction is really a bet on the European Commission's own sequencing, read against Google's incentives. On July 16, 2026 the Commission adopted two binding specification decisions under Article 6(7) of the Digital Markets Act. The bulk of the Android interoperability obligations — voice invocation plumbing, cross-app action, context access — is tied to Android 18 and due no later than August 1, 2027. But the specific capability that makes a rival wake word work as a live alternative, concurrent wake-word access, carries a separate and later deadline of August 1, 2028, tied to Android 19.

That one-year gap is the whole basis for this call. The headline benefit — a rival you can talk to hands-free — is the last piece to land, and it lands in 2028, not 2027. Even the August 2027 tranche is a floor Google is unlikely to beat voluntarily, and Google has both the incentive and the tool to hold to the floor: five of the eleven features carry an objective-criteria carve-out that lets Google set security conditions on the most sensitive capabilities. A firm defending the most valuable surface it owns does not ship the crown-jewel capability early; it ships to the deadline and litigates the criteria. I expect the practical experience through the end of 2027 to be tap-to-open rivals plus partial in-app action, with the reflexive spoken default still effectively Google's — exactly the year-long window the staged remedy leaves open.

This is the falsifiable edge of a broader argument I made about why the assistant slot, not the model, is the real moat in consumer AI, in The Wake Word Is the Moat.

What would make me wrong

Three things could break this. First, Google could ship concurrent wake-word support ahead of its 2028 deadline — possible if it decides the reputational cost of slow-walking exceeds the competitive cost of opening early, but against its revealed behavior. Second, the Commission could accelerate the timeline through a fresh enforcement action if it judges Google is complying too slowly. Third, a rival could reach hands-free voice invocation through an OEM partnership or its own hardware rather than through the DMA remedy, sidestepping the schedule entirely. If any of those produces a shipped, default-eligible rival wake word on a stock EU Android phone before 2028, the prediction fails — and the "invocation is the moat" thesis gets stronger, not weaker, because it would mean the surface was contestable sooner than even the regulator planned.