Apple Record High|iPhone Secrets Inside OpenAIs Device Lab

The Device in Two Places at Once

Apple Inc. posted its first record closing high in over a month last Thursday, pulling ahead of every other member of the Magnificent Seven at a moment when Amazon, Alphabet, Microsoft, Nvidia, Meta, and Tesla all sit below their own all-time highs. The thesis driving that premium is straightforward: Apple's hardware design is a moat so deep that no competitor can replicate it. What the market has not priced is the possibility that it already has been.

On July 10, Apple filed a 41-page federal complaint in the Northern District of California alleging that OpenAI's hardware subsidiary io Products did not merely hire former Apple engineers — it directed them to bring actual physical hardware out the door. The complaint names Tang Tan, OpenAI's chief hardware officer and a 24-year Apple veteran who served as vice president of product design for the iPhone and Apple Watch, as the architect of what Apple calls a systematic extraction operation. Candidates still employed at Apple were allegedly instructed to carry batteries, logic boards, and system-in-package components to OpenAI for show-and-tell sessions. These are not documents. They are the physical objects the viewer holds in their hand.

Apple's roughly 16.5% year-to-date gain is the strongest in the Magnificent Seven — a group in which Microsoft and Tesla are actually negative for the year. Wall Street is rewarding Apple's low-capital-expenditure approach to AI precisely because it assumes the hardware advantage is already built and protected. OpenAI's director of strategic communications issued a flat denial: 'We have no interest in other companies' trade secrets.' The two claims cannot both be true, and the gap between them is now a federal case.

The Network That Stayed Open

The complaint's most legally specific allegation does not involve Tang Tan. Chang Liu spent eight years at Apple as a senior systems electrical engineer before leaving for OpenAI in January 2026. He did not return his company-issued MacBook. He did not schedule Apple's standard exit confidentiality session. Then, on or around February 9, 2026 — weeks after his departure and while already employed at OpenAI — Liu discovered that an authentication vulnerability in Apple's internal network still granted him remote access to shared file servers. His message to a former Apple colleague: 'LOL, I found out I can access the network storage, so funny.' The colleague allegedly replied: 'I'm ready.'

What followed, according to Apple's complaint, was a download of dozens of confidential hardware files — technical specifications, engineering presentations, and proprietary project data — with one collection exceeding one thousand pages. Apple further alleges that OpenAI approached a shared contract manufacturer and arranged a demonstration of a proprietary Apple metal-finishing technique — the kind of precision surface treatment that took Apple decades to develop — while misleading the supplier into believing Apple had consented. Liu had also told a colleague on his last day at Apple: 'I still have another computer,' referring to a second Apple-issued device he planned to keep for continued access.

Here is the pivot most coverage has missed. California law does not permit Apple to win by arguing that hundreds of former employees at OpenAI will inevitably leak what they know — the state's courts have explicitly and completely rejected the inevitable disclosure doctrine, and California's prohibition on non-competes reinforces the same policy. But Apple's complaint is carefully structured to avoid that trap. It does not rely on background knowledge traveling with engineers; it relies on alleged specific documented acts: a retained device, a post-resignation network intrusion, physical components removed, and a supplier demonstration conducted under false consent. Whether those specific acts are proven is a different question from whether the legal theory is viable. The theory is viable.

What the Moat Is Actually Worth

The market's current read on Apple is that the hardware moat is not just intact but compounding. This week, Broadcom confirmed a new agreement to manufacture custom chips for Apple extending through 2031 — a deal analysts called Apple's newly de-risked cash machine. Separately, Apple is closing in on Nvidia's position as the largest U.S. company by market capitalization, with investors treating its low artificial-intelligence capital expenditure as an advantage. The logic is that Apple does not need to spend $200 billion a year on data centers because its installed base of billions of devices already delivers the inference edge, and because the physical design of those devices is a secret no competitor can replicate.

That assumption is what the lawsuit puts under pressure. OpenAI acquired io Products — the hardware startup co-founded by Jony Ive, Tang Tan, and other former Apple design executives — for approximately $6.5 billion in 2025. The device io Products is building has not been described in public, but its design leadership spent decades at Apple working specifically on the iPhone and Apple Watch. Apple's complaint draws a sharp line: the knowledge those engineers carry in their heads is legal and inalienable under California law, but the specific files, physical parts, and manufacturing process demonstrations they allegedly transferred are a categorically different matter. The $6.5 billion acquisition price already implies the market assigned significant value to that team's design capability; the open question is which portion of that capability came from their skill and which came from what they allegedly brought with them.

The Magnificent Seven divergence makes the stakes concrete. Amazon guided full-year 2026 capital expenditure toward $200 billion; Alphabet set its figure at $175 to $185 billion. Apple is outperforming every peer year to date by spending a fraction of that, because investors believe its moat is built into the physical object rather than into server counts. If Apple's complaint allegations are proven — or if discovery reveals that the technical specifications of its system-in-package architecture are already in OpenAI's design files — the premium Apple commands over its peers rests on a foundation that was compromised before the market knew to ask the question.

What Decides the Trade

The lawsuit's most consequential near-term checkpoint is not the eventual trial outcome but whether Apple seeks and obtains a preliminary injunction restricting io Products' use of the allegedly misappropriated specifications. A preliminary injunction requires Apple to show a likelihood of success on the merits and irreparable harm — a standard that California's rejection of inevitable disclosure makes harder, but that the specific documented acts in the complaint make more plausible than typical trade-secret cases. If a court grants even a partial injunction restricting io Products' access to specific hardware specifications, it confirms that the alleged extraction was both real and legally cognizable. If the court denies it, the presumption shifts: either the alleged acts were not as Apple described them, or the specifications did not cross the trade-secret threshold.

For holders of Apple at a record close, the decision variable is not the existence of the lawsuit but what the preliminary injunction motion reveals about the specificity of Apple's evidence. A denial on procedural grounds — California courts are skeptical of trade-secret injunctions absent clear irreparable harm — does not vindicate OpenAI; it signals that the market premium survives because the court found the harm too speculative to restrain a competitor. A grant is the first confirmation that Apple's most protected manufacturing knowledge is in active use at a direct competitor, and the record close begins to look like the high before that disclosure. For watchers considering entry, the metric to monitor is not the next earnings date but the first substantive motion filing — the moment at which Apple's evidence must be stated with particularity rather than alleged — because that filing either validates the premium or begins to unwind it.