AI Won't Kill the Smartphone — It Will Kill Its Monopoly as the Gateway

THE THESIS. AI is not going to kill the smartphone this decade. The data is unambiguous: 1.25 billion phones were produced in 2025, the highest figure since 2021 according to TrendForce, and the iPhone posted its best year ever with 240.6 million units. What is at stake — and it is worth far more — is the smartphone's monopoly as the gateway to digital life. This war is not about replacing one gadget with another. It is about who controls the AI agent's interface: the direct relationship with the user that today must pass through two companies, Apple and Google. That is why Apple did not sue OpenAI over a speaker. It sued over the succession to the throne.

THE GRAVEYARD OF PIONEERS. The first generation of “smartphone killers” already has headstones, and the epitaphs are worth reading carefully. The Humane AI Pin promised to free us from the screen: a $699 badge plus a $24 monthly subscription that projected information onto your palm. By mid-2024, daily returns were outpacing new sales. HP bought its assets for $116 million in February 2025, per TechCrunch — a fraction of the near-$1 billion valuation the company once courted. The Rabbit R1, a $199 orange box, sold around 100,000 units; five months later, only 5,000 people used it daily, as its own founder admitted. A 95% abandonment rate.

The lesson is not that voice is a bad interface. It is that both devices did fewer things than the phone, more slowly, and charged money for the privilege. The bar for any new hardware is brutal: it must beat a device you already carry in your pocket, that does everything “well enough,” and whose marginal cost to the user is zero. Humane and Rabbit did not die for dreaming; they died by showing up with a worse product to compete against a good one already paid for.

THE LAWSUIT AND THE REAL DEVICE. Let's separate facts from leaks. Facts: OpenAI bought io, Jony Ive's firm, for $6.4 billion in May 2025. On July 10, 2026, Apple filed a 41-page complaint in federal court in California against OpenAI, io Products and two former employees. It accuses the company of systematically stealing trade secrets — confidential product information — through ex-Apple staff, and points at Tang Tan, OpenAI's hardware chief and a former Apple design VP. More than 400 former Apple employees now work at OpenAI, according to Fortune. OpenAI calls the allegations “baseless,” moved to dismiss the case on August 5, and a hearing is set for October 1. Nothing has been proven: these are claims by the parties, in both directions.

And the device? What is verified comes, paradoxically, from court filings: it will be pocket-sized, screenless, contextually aware, and it will not be a wearable or an earbud. OpenAI describes it as a “third core device,” sitting alongside your laptop and phone — not in their place. The date has slipped: it will not ship before the end of February 2027, according to the company's own filings as reported by MacRumors. Leaks describe a smart-speaker-like object priced at $300–400; they are unconfirmed. And Sam Altman's “coolest piece of technology the world will have ever seen” is, until demonstrated, marketing. Our usual yardstick applies: demonstrated capability versus aspiration. Today, everything that exists of this product is legal paperwork and adjectives.

GLASSES: THE ONLY PARTIAL SUCCESS. Meanwhile, one AI device actually sells, with audited numbers: Meta's glasses with EssilorLuxottica shipped 7 million units in 2025 alone, triple the previous year, according to the manufacturer's results — and the partners are weighing production capacity of 20 to 30 million a year, per Bloomberg. Why do they work where Humane failed? Because they do not try to replace the phone: they extend it. They are normal glasses with a camera and microphone exactly where an assistant needs them — at your eyes and mouth — at premium-sunglasses prices. Google and Samsung have copied the playbook: their Gemini glasses, built with Warby Parker and Gentle Monster, arrive this fall and are explicitly billed as a “companion device” to the phone. The market's message is clear: the transition advances by accessory, not by assault.

THE 30% TOLL, THE REAL ENGINE. Why does OpenAI insist on building hardware after touring the graveyard? Money and control. Apple and Google take commissions of up to 30% in their app stores and decide which assistant answers when the user presses a button. Every ChatGPT subscription bought from an iPhone pays rent to the landlord. OpenAI is already moving on the software front — it launched “apps inside ChatGPT” in October 2025 to become a platform itself — but without its own hardware there will always be a button, an operating system and a store standing between its agent and the user. Apple, meanwhile, is playing its classic card: silent integration. It pays Google roughly $1 billion a year for Gemini to power the new Siri unveiled in June 2026, according to CNBC, and has relegated ChatGPT to a secondary role on the iPhone. Translation: the phone is the moat, and AI gets embedded into it without surrendering the door. The lawsuit and the Google deal are the same strategy through two channels: keeping the interface theirs.

OUR READ: THE PRECEDENT AND THE HORIZON. History justifies Cupertino's paranoia. Every platform transition dethroned the previous kings: IBM ruled the mainframe and lost the PC; Microsoft ruled the PC and lost mobile; Nokia and BlackBerry, masters of the phone in 2007, vanished from the map within five years. Incumbents almost never survive an interface change. But the precedent has a second, less-quoted reading: the iPhone won because it did more than what it replaced — phone, iPod and internet in one device — not less. No current AI device passes that test. Our short-term forecast: the smartphone remains the center, glasses keep growing as an accessory, and OpenAI's device will face the same wall Humane hit — with one decisive difference Humane never had: the distribution of hundreds of millions of ChatGPT users, per OpenAI's own figures. That card can change the game; the Apple lawsuit will decide when it can be played.

Long term, our optimism is measured but firm. Computing dissolving into the environment — voice, cameras, agents working on your behalf — means technology finally adapting to humans rather than the reverse: fewer hours staring at glass, more free hands and eyes. That horizon fits AI's deeper promise: freeing time and capacity for what actually matters. The risk is not the gadget but repeating the original sin: whoever escapes the 30% toll ends up building their own walled garden and charging their own. The right question is not whether AI will kill the smartphone. It is whether the next gateway will be more open than the current one. That will not be decided by a Jony Ive render — it will be decided by regulators, competition and, above all, users.