Apple Called OpenAI's Hardware Business 'Rotten to Its Core.' Here's What the Lawsuit Actually Says.

What Apple Is Alleging

Before you assume this lawsuit is about ChatGPT, AI models, or any of OpenAI's software products, stop. Apple's complaint filed July 10, 2026, in the Northern District of California (Case 5:26-cv-07078) is specifically about hardware — supplier relationships, manufacturing techniques, physical prototypes, and the confidential infrastructure Apple has spent decades building to produce devices at scale.

The allegations in the complaint are concrete. Apple describes employees emailing themselves supplier data before walking out the door. It describes job candidates showing up to OpenAI interviews carrying Apple parts and prototypes. It describes unauthorized network access used to download confidential hardware files. These are not vague IP claims — they are specific actions Apple says it can document.

The named defendants are Tang Tan, OpenAI's chief hardware officer and a former Apple executive, and Chang Liu, an engineer who joined OpenAI from Apple in January 2026. Apple's position is that these individuals didn't just bring general expertise with them — they brought proprietary information that OpenAI then used to approach Apple's own manufacturing partners, including Foxconn and Luxshare, to extract techniques tied to unreleased products.

The complaint describes this as a "pattern of theft." Apple's characterization of OpenAI's hardware business as "rotten to its core" is in the filing itself, not a press statement.

The io Products Connection

To understand why Apple named io Products as a defendant, you need the timeline. OpenAI acquired the Jony Ive-linked hardware startup in May 2025 for $6.5 billion. That acquisition was widely covered as OpenAI's statement of intent — a signal that the company was serious about building consumer hardware, not just software. io Products brought design credibility and, critically, a roster of former Apple talent already embedded in OpenAI's hardware ambitions.

Tang Tan sits at the center of that talent pipeline. As OpenAI's chief hardware officer, he connects the io Products acquisition directly to the conduct Apple is describing in its complaint. Chang Liu, who moved from Apple to OpenAI in January 2026, represents the more recent end of that pipeline. Apple's argument, taken together, is that the acquisition didn't just import Jony Ive's design sensibility — it imported a pattern of behavior that included walking out of Apple with information that didn't belong to them.

What makes io Products significant here isn't the $6.5 billion price tag. It's that the acquisition gave OpenAI an organizational home for exactly the kind of supplier relationship-building and manufacturing expertise Apple says was stolen. When Apple alleges that OpenAI used misappropriated information to approach Foxconn and Luxshare, io Products is the business context in which that outreach would have happened.

How Apple Says It Happened

The mechanics Apple describes in the complaint are worth reading carefully, because they are not sophisticated. They are mundane. An employee emails supplier information to a personal account before their last day. A job candidate walks into an interview carrying Apple parts — physical prototypes that have no business leaving an Apple facility. Someone with network credentials uses them to pull down confidential hardware files on the way out the door. These are the specific acts Apple says it can document across multiple individuals.

What makes this operationally significant is what allegedly happened next. Apple claims that information obtained through these methods was then used to approach Foxconn and Luxshare directly — two manufacturers that sit at the center of Apple's supply chain for products that have not yet shipped. The request, according to the complaint, was for manufacturing techniques tied to those unreleased products.

That is where the damage becomes concrete rather than theoretical. Supplier relationships and manufacturing process knowledge are not publicly available. Apple has spent years developing them. If a competitor can walk those relationships and techniques out the door and hand them to a manufacturer already under contract with Apple, the harm is not speculative — it is direct and immediate.

The complaint frames this not as isolated misconduct by one or two employees, but as a pattern. That framing matters for what comes next.

Where This Goes Next

As of July 11, 2026, OpenAI has not responded publicly. No court developments have followed the filing. What exists right now is Apple's complaint, its specific allegations, and language that is notably aggressive even by the standards of corporate litigation.

That language is worth taking seriously as a signal. "Rotten to its core" and "pattern of theft" are not the kind of phrases a legal team drops into a federal complaint without strategic intent. They are designed to be quoted, to frame the narrative before OpenAI has a chance to offer a counter-narrative, and to signal to the court — and to the public — that Apple is not interested in a quiet settlement. Companies that want to negotiate write measured complaints. Companies that want to litigate write complaints that read like this one.

The "pattern" framing also has a specific legal function. Establishing a pattern rather than isolated incidents is harder for a defendant to explain away. It implies institutional knowledge, not rogue employees acting alone. If Apple can demonstrate that multiple individuals engaged in similar conduct across different time periods, the argument that OpenAI simply didn't know becomes significantly harder to sustain.

What happens next depends almost entirely on how OpenAI responds — and when. Discovery, if this reaches that stage, would give Apple access to internal communications at exactly the company it is directly competing with in consumer hardware. That alone makes the stakes here considerably higher than a typical trade secret case.