Apple is suing the hardware programme, not the software company

Apple's complaint against OpenAI is not really a trade-secret case. It is Apple treating OpenAI's hardware programme as an existential competitor, and using litigation to slow it while the first device is still unshipped. The interesting question is not whether Apple wins. It is what the filing tells us about where the moats have moved.

On Friday, Apple filed a federal complaint in California ¹ naming OpenAI and two former Apple employees as defendants. The allegations are unusually vivid: stolen laptops, data exfiltration, informants left inside Apple after departure, and a pattern of recruiting that Apple frames as espionage rather than hiring. The target of the alleged theft is specific — OpenAI's hardware programme, led by Jony Ive, Apple's former chief design officer. OpenAI's response was categorical and unspecific: the company "has no interest in other companies' trade secrets." ¹

Read that carefully. It is a denial of interest, not a denial of conduct.

The hire that made the lawsuit inevitable. When OpenAI brought Ive in to run a consumer hardware project, the industry read it as a signalling move — Sam Altman buying the aesthetic credibility of the person most identified with Apple's physical identity. That reading undersold it. Ive did not arrive alone. Design leadership, industrial engineering, supply-chain relationships, and the tacit knowledge of how to actually ship a hundred-million-unit consumer device — these travel with a team, not a name. Apple has watched that team assemble in a competitor for roughly two years.

The lawsuit is what happens when the observation crosses a threshold Apple's general counsel can defend in court.

Hardware is the new moat, and both sides know it. The software-platform era rewarded whoever controlled the OS and the app store. Distribution was the moat; the device was the delivery vehicle. That framing has been the received wisdom for fifteen years, and it is the framing under which almost every current AI valuation is priced. Frontier labs are treated as software companies whose costs are compute and whose margins will eventually resemble SaaS (software-as-a-service, sold per seat).

The Ive hire, and now this lawsuit, are evidence that at least two serious actors no longer believe it. OpenAI is not building a device because software distribution is going well; it is building one because software distribution runs through Apple and Google, and the terms of that distribution are set by the owners of the glass. Apple is not suing because a designer left; it is suing because the specific combination of Ive-plus-team-plus-OpenAI-capital is the first credible attempt in a decade to build a new consumer hardware category outside the iPhone gravitational field.

The Siri-to-Gemini switch ¹ earlier this year now reads differently in light of the complaint. Apple did not sue while it was a commercial partner of OpenAI. It terminated the ChatGPT integration, moved Siri onto Google Gemini (Google's flagship model, architecturally closer to Apple's on-device inference approach), and then filed. The sequence matters. Commercial entanglement was cleared first. The lawsuit is the second move, not the first.

Discovery is the actual weapon. A trade-secret case, if it proceeds past motions to dismiss, compels both sides to disclose internal documents to opposing counsel and, in many cases, to the public record. This is where Apple's calculation gets interesting.

OpenAI's hardware roadmap, if surfaced, damages its launch. The device has not shipped. A competitor, any competitor, not only Apple, reading the industrial design specifications, the component sourcing, the target price point and margin structure, gets months of lead time to respond. For a private company preparing a category-defining consumer launch, that is not a nuisance. That is a genuine strategic wound.

But Apple's own development pipeline enters discovery too. What Apple was building, where it perceived its vulnerabilities, which of its own projects it feared OpenAI was tracking — all of it becomes visible to OpenAI's lawyers, and some of it, eventually, to the industry. Apple is betting that the asymmetry of damage favours it: OpenAI has one hardware bet, Apple has a portfolio. A leak that costs OpenAI its launch is fatal; a leak that reveals Apple's next AirPods roadmap is embarrassing.

That bet may not hold. Apple is not just a hardware portfolio; it is a company whose competitive advantage rests on the surprise of its product cadence. Discovery that reveals how Apple currently thinks about on-device AI, the very question OpenAI is trying to answer with its device, is a different order of exposure than the "portfolio" framing suggests. Apple's lawyers will have modelled this. Whether they modelled it correctly is a question the case itself will answer.

The precedent is bigger than the case. The 2024-2026 period has been characterised by aggressive talent movement from big-tech incumbents into frontier labs. Meta poaching Scale AI's leadership. Anthropic hiring across Google DeepMind. OpenAI recruiting from Apple, Tesla, and Meta. The going rate for a senior researcher has been reported in ranges that would have been considered fabricated three years ago.

All of this operates under an implicit assumption: California employment law makes non-competes unenforceable ², and the resulting talent fluidity is a feature of the ecosystem, not a bug. If Apple wins, specifically on the recruiting-as-espionage characterisation, that assumption breaks. Every large incumbent gets a template: characterise the hiring pattern as systematic, name the recruiter, allege coordination with the departing employee before their notice period, and the case need not be as extreme as this one to survive a motion to dismiss.

The frontier labs' hiring model depends on being able to move quickly, quietly, and in teams. A precedent that makes team-moves litigable at the recruiter level, not just the employee level, changes the economics of every offer letter written after the ruling. This is where I would watch most carefully. The Apple-OpenAI headline is loud. The precedent-setting is quieter and, over five years, more consequential.

The strongest counter-case. I have written the case for Apple as if the complaint has substance. It may not. The vivid allegations, laptops, moles, espionage, are the parts that survive contact with a press release. The parts that will survive contact with a court are the employment agreements, the NDAs, the specific documents alleged to have been taken, and the chain of custody. None of that is public yet.

California's public policy strongly favours employee mobility. Section 16600 of the Business and Professions Code renders most non-compete clauses unenforceable, and courts have historically been sceptical of trade-secret claims that operate as de facto non-competes. Apple's "recruiting-as-espionage" framing has to clear a bar that California courts have set high, precisely to prevent large incumbents from using litigation to punish employee departures.

And Apple's own history invites scrutiny. Much of Apple's design and engineering heritage was built by attracting people from other firms — from Sony, from HP, from Motorola, from suppliers whose relationships Apple later ate. A company with that hiring history framing a competitor's recruiting as espionage is making an argument the courts, and the press, will examine with some care.

The contrarian read is that this is a nuisance filing dressed as trade-secret litigation: Apple has the balance sheet to litigate indefinitely, the motive to slow OpenAI's hardware launch, and the confidence that even a losing case buys time. That read is credible. It is also compatible with the hardware-moat thesis: Apple would not spend the reputational cost of a nuisance filing on a threat it considered soft.

What this tells us about the next two years. The story most people are reading today is Apple versus OpenAI. That is the surface. The story underneath is that hardware engineers, industrial designers, supply-chain leads, and manufacturing partners are now being treated by both sides as strategic AI assets — assets worth suing over. The software-only frame for AI competition is not wrong so much as incomplete. The winners of the next cycle will be the ones who can integrate models with devices, and the moat around that integration is being built now, in hiring, in supplier contracts, and, as of Friday, in court filings.

Whether Apple wins the case matters less than the fact that it filed. The filing itself is the signal. Apple has told the market where it thinks the fight is. The rest of us should update accordingly.

Footnotes