Apple Sued OpenAI for Trade Secrets. It Already Chose Google.

Apple filed a federal lawsuit against OpenAI on Friday, July 10, accusing the company of stealing trade secrets "at every level" of its organization, from junior technical staff up to OpenAI's Chief Hardware Officer. The complaint lands seven months after Apple quietly struck a very different kind of deal with a very different AI company: paying Google roughly $1 billion a year for a custom Gemini model to run the next generation of Siri, arriving this fall.

Neither story is really new on its own. The Gemini-Siri deal was reported back in January and detailed publicly at WWDC in June. The lawsuit is what's new this week. But put the two next to each other and they stop looking like unrelated headlines. They read like the closing chapters of a partnership that started with real promise in 2024, when Apple integrated ChatGPT directly into iOS, and ended with Apple suing its former partner while the assistant business went to Google instead.

What Apple is actually alleging

The complaint, filed in federal court in Northern California, names Tang Tan as a central figure. Tan is OpenAI's hardware chief and a former Apple vice president who left to help build the company's long-rumored consumer AI device. According to Apple's filing, Tan directed job candidates still employed at Apple to bring "actual parts" from Apple's supply chain to interviews, where he and his team allegedly used "show and tell" sessions to extract additional confidential information. It's a specific, almost mundane detail, and that's exactly the kind of thing that tends to stick with a jury more than abstract claims about misappropriated IP.

Apple also alleges that OpenAI coached departing employees on how to evade the company's internal security review before they left, and that one former employee, Chang Liu, took an Apple laptop with him when he joined OpenAI. Fortune reports that Jony Ive's hardware startup io Products, which OpenAI acquired last year, is named in the suit too, tying the alleged theft directly to OpenAI's push into consumer devices. None of this has been tested in court, and OpenAI had not filed a public response as of this writing.

A $1 billion bet that already looks shaky

The Siri side of this story has its own numbers worth knowing. CNBC reported in January that Apple agreed to pay Google around $1 billion annually for a custom Gemini model built at roughly 1.2 trillion parameters, far larger than anything Apple has trained in-house. Gene Munster, managing partner at Deepwater Asset Management, put the total value of the multi-year arrangement as high as $5 billion.

The reaction once Siri's rebuilt version actually showed up at WWDC wasn't universally positive. Ming-Chi Kuo of TF International Securities warned beforehand that Siri needed to "outclass Google Gemini, not just catch up after years of delay." By most accounts it didn't clear that bar. Citi's research team wrote that the deal simply "underscores Google's core AI advantages," and CNBC's follow-up coverage noted analysts pressing Google executives on the arrangement during earnings calls, treating it as a defensive win for Google rather than a Siri turnaround for Apple.

Two decisions, one direction

That's less coincidence than two symptoms of the same underlying shift. Apple leaned on OpenAI through 2024 because building a frontier model in-house wasn't moving fast enough. By 2026 the calculus changed twice over. Apple decided Gemini, not a model it built itself and not OpenAI's, was the right foundation for Siri. Separately, months later, it concluded OpenAI's hardware ambitions had crossed from competitive pressure into alleged theft. Different claims, same direction: Apple no longer treats OpenAI as a trusted partner in any part of its roadmap, assistant or hardware.

For OpenAI, the picture is messier than losing one integration. It's worth remembering what that integration was worth: a spot as the default assistant layer on a billion-plus active devices, the kind of distribution our own comparison of ChatGPT, Claude, and Gemini shows still shapes which chatbot people actually end up using day to day.

OpenAI loses a marquee partnership it once treated as proof that ChatGPT belonged on the world's most valuable hardware platform. It also now has to defend a lawsuit that, win or lose, puts its hardware ambitions under scrutiny they hadn't faced before. A company trying to convince the market it can build trustworthy consumer devices does not want "alleged systematic theft of trade secrets" as the framing story while that device is still unannounced.

What to watch next

Court filings move slowly, and trade secret cases in particular tend to settle once both sides see what discovery turns up. The more immediate signal is how OpenAI responds in public. Does it dispute the specific allegations, the Tan interviews, the Liu laptop, the io Products connection, or does it lean on a broader defense about employee mobility and non-compete law? How aggressively OpenAI pushes back will say more about its confidence in its own hardware timeline than anything written in Apple's complaint.