Overview of Here’s what an AI-first phone might look like
This VergeCast episode explores the current wave of “AI devices” that claim not to be phones, but may end up looking a lot like smartphones anyway. Host David Pierce talks with David Imel about the rumored device efforts from OpenAI, SpaceX, and Amazon, and whether any of them can create a compelling AI-first product that replaces, rather than imitates, the smartphone.
What Else Happened on The Verge
90 Seconds on The Verge highlights
- Open Secure AI Alliance: NVIDIA, Microsoft, SpaceX, IBM, and others are forming a new alliance around open AI models and security.
- Amazon satellite internet: Amazon filed to deploy over 5,100 satellites and may be positioning Leo as a Starlink competitor with direct-to-satellite mobile service.
- Framework Laptop 13 Pro review: The modular, repairable laptop impressed, but the RAM shortage has pushed the price to a steep $3,100.
The Core Debate: What Is an “AI Phone”?
The big question
The episode centers on whether companies can build an AI-centric device that:
- changes how people interact with their digital lives,
- avoids becoming “just another smartphone,” and
- gives AI a front-and-center role without requiring a full app-based ecosystem.
The intended shift
The hosts frame the goal as moving from:
- an app-based model: open app → tap buttons → complete task
- to an intent-based model: express what you want in natural language → AI handles the rest
That sounds appealing in theory, but the episode repeatedly questions whether it’s actually better than tapping buttons, especially for fast, common tasks.
Why the Idea Is Hard to Pull Off
1. Apps still work well
The discussion emphasizes that phone apps remain efficient, familiar, and precise. Typing a prompt is slower than pressing buttons, and many tasks are already optimized for app flows.
2. Voice and AI are useful, but not enough on their own
AI-powered voice control and dictation are a real improvement. David Imel gives a strong example of using Siri to:
- find out when concert tickets go on sale,
- then automatically set a reminder for the exact time.
That kind of automation feels genuinely useful, but the episode argues it doesn’t necessarily justify a whole new phone.
3. A “not phone” often turns into a phone
The main critique is that once a company tries to support enough real-world use cases, it winds up needing:
- apps,
- app integrations,
- user accounts,
- notification systems,
- and familiar mobile hardware.
At that point, the “new device” is basically a smartphone.
Company-by-Company Breakdown
OpenAI
- Likely to build something ChatGPT-centric.
- Could be an Android-based device or a phone with a custom AI-first launcher.
- Biggest challenge: getting people to switch from an iPhone or Galaxy just to use ChatGPT more conveniently.
- The hosts suggest OpenAI’s best path may be better AI features on existing phones, not a whole new device.
SpaceX / Elon Musk
- Potentially the strongest shot at shipping something people will buy, largely because of Elon Musk’s ability to generate attention and loyal buyers.
- Starlink plus a phone could make sense as a combined ecosystem.
- But the episode notes that anything tied to Musk also carries brand baggage, and a “Grok phone” sounds more like a disaster than a revolution.
Amazon
- The most pessimistic case in the discussion.
- Amazon’s rumored phone appears designed to make shopping, Prime Video, Prime Music, and food ordering easier.
- The hosts are highly skeptical that people want their phone to be an always-on portal to Amazon commerce.
- Their verdict: Amazon seems to have learned very little from the original Fire Phone failure.
Important Takeaways
The real opportunity may be in accessories, not standalone phones
The hosts argue that AI-first earbuds, watches, or other add-ons may make more sense than trying to replace the phone entirely. Accessories are:
- cheaper,
- easier to adopt,
- and more realistic as AI companions to existing phones.
Better dictation is a real win
One of the most concrete AI improvements discussed is voice dictation:
- Android is currently better than iPhone for dictation in their experience.
- AI-powered transcription that understands context and apps is one of the most promising near-term uses.
The duopoly is still the problem
The episode repeatedly returns to the reality that in the U.S., iPhone and Samsung dominate. Any non-phone phone has to overcome:
- Apple’s ecosystem lock-in,
- Google’s platform control,
- app store friction,
- and user inertia.
AI brand trust is uneven
People may like specific models or features, but there isn’t strong consumer loyalty to “AI companies” in the way there is to Apple hardware or even certain phone brands. That makes it hard to sell a dedicated AI device.
Final Verdict
The episode’s conclusion is basically:
- AI-first phones are a compelling idea in theory
- But in practice, most of them will probably become smartphones or smartphone accessories
- The most plausible future is not replacing the phone, but making the phone more AI-native
The hosts end by inviting listeners to send in ideas for what the next UI paradigm should be, because that’s the real unresolved question:
How do you combine the familiarity of apps with the power of AI agents without making the device worse?