New vs. Old: The Love-Hate Relationship Between OpenAI and Tech Giants

Turn the clock back two years—just when Apple had, at WWDC (Apple’s Worldwide Developers Conference), ushered ChatGPT into the iPhone in grand style, marrying it with all the pomp and ceremony.

Back then, it looked like a match made in heaven: Apple sat atop the world’s largest and most far-reaching device empire, while OpenAI stepped in to tutor Siri, which had been slow to learn how to talk like humans. Apple lacked the model; OpenAI lacked the entry point. Each got what it needed, and the partnership felt perfectly complementary.

But today, Apple has turned around and dragged OpenAI into court.

On July 10, 2026, Apple sued OpenAI, io Products, the hardware company under OpenAI, and two former Apple employees, Chang Liu and Tang Yew Tan, in the U.S. District Court for the Northern District of California. Apple alleges they systematically stole Apple’s confidential documents, hardware components, manufacturing processes, and unpublished product information—through methods including recruiting, interviews, guidance after resignations, and supply-chain cooperation—to accelerate OpenAI’s consumer hardware business.

That’s business warfare—simple and unvarnished.

What’s often called “strategic cooperation” is frequently just a polite way for both sides to avoid crashing head-on when their business lines haven’t collided directly yet.

Once they begin competing for the same users, talent, supply chains, and next-generation hardware entry points, even the most devoted allies can turn on each other instantly. Especially when OpenAI’s expansion path over the past few years has been, for almost every step forward, pushing right into the core territory of a tech giant.

But if you look closely—why does OpenAI seem to have beef with just about everyone?

This company has an almost magical ability: it can always first shake hands and collaborate with someone, then bring its business right up to that partner’s doorstep, and finally succeed in turning a collaborator into a competitor—while outlasting investors until they become the “family members” who are fireproof, theft-proof, and OpenAI-proof.

Here’s a quick rundown of OpenAI’s love-hate entanglements with major tech companies.

Apple: Sweet back then, now facing each other in court

Apple and OpenAI’s relationship is a classic case of “love at first sight, divorce at first sight” in the AI circle.

In 2024, Apple was clearly a half step behind on generative AI. ChatGPT had already strongly taken over global users’ mindshare. Google pushed Gemini hard, Microsoft shoved Copilot into Office for office assistance, while Apple—once the “AI big brother” that had been the chatgot-in-motion—was still struggling on stage at a developer conference, painfully proving that Siri can really understand human language.

So there was only one option: bring in outside help.

Apple integrated ChatGPT into Apple Intelligence. When Siri encountered questions it couldn’t answer, it could politely ask, “Want me to ask ChatGPT for you?”

Through this, OpenAI got one of the most valuable consumer electronics entry points worldwide. Apple, meanwhile, temporarily covered up its own model capability shortcomings.

But in reality, from the beginning the two companies didn’t want the same thing.

Apple wanted ChatGPT to be a plugin—ideally quietly living deep inside the system, only called when needed, and never stealing the spotlight when it wasn’t.

Apple controlled users, hardware, the system, and distribution. OpenAI provided part of the capability.

But what OpenAI wanted was clearly more than just a premium outsourced solution living on an iPhone. It aimed to become a new consumer entry point for users. If, in the future, users no longer opened apps but instead just said to the AI, “Help me get this done,” then whoever controls the AI assistant could potentially bypass the traditional operating system entirely.

Later, OpenAI realized Apple was “all in.” It pulled in Jony Ive’s team—one of the key figures behind products like iPhone, iPad, and MacBook. With great fanfare, the team moved into consumer hardware. So Apple naturally couldn’t sit still. It stopped mentioning its past goodwill with Siri and ChatGPT altogether. Now it was not only bringing Gemini into the core capabilities of the updated Siri, but also suing OpenAI.

According to Apple’s lawsuit narrative, this wasn’t just “normal employee poaching.”

Apple says that after a former employee, Chang Liu, left Apple to join OpenAI, he still accessed Apple’s internal servers through vulnerabilities and downloaded a large amount of confidential engineering materials. He was also accused of helping other Apple employees copy files and evade security reviews to prepare for OpenAI interviews. Apple further claims that OpenAI’s recruitment system would instruct Apple employees on how to handle resignation and departure screening—reminding them not to reveal their destination too early (especially not to mention OpenAI), not to sign documents casually, and to extend system access permissions as much as possible.

Even in interviews, OpenAI’s former Apple employees allegedly used Apple’s internal project code names and required candidates to prepare a “technical deep-dive analysis” related to their current work. Some candidates were even asked to come to interviews carrying Apple batteries, circuit boards, logic boards, and prototype components. Of course, as of now these are still Apple’s one-sided allegations; whether they hold up ultimately depends on evidence disclosure and the court’s review. But regardless of how the case turns out, Apple and OpenAI’s relationship is already very hard to return to what it once was.

Microsoft: As the money source, it has to guard against it flying solo

Microsoft and OpenAI’s relationship is simple: Microsoft fears OpenAI running off; OpenAI fears Microsoft managing it. And now that Microsoft has poured in so much money, what it’s most worried about is that OpenAI might become too successful.

Microsoft’s early support for OpenAI was extremely tangible—money, compute power, cloud services, enterprise customers, Office entry points, Windows entry points, and GitHub entry points—basically the ultimate “big backer.”

OpenAI’s ability to train and deploy massive models early on absolutely depended on support from Microsoft Azure. After ChatGPT went viral, Microsoft rapidly inserted OpenAI models into Copilot, Office, Bing, and enterprise services.

Microsoft looked like it had hit the era’s jackpot. Years ago it had even been paying slightly to catch up after missing the mobile internet wave. In recent years, with OpenAI, it flipped overnight and returned to the very center of the tech narrative.

But the more successful this relationship becomes, the more awkward it gets for both sides.

Microsoft wants OpenAI to sell Azure, sell Office, sell Copilot. OpenAI wants to build its own consumer entry points, enterprise platform, developer ecosystem, search products, operating system agent, and even consumer hardware.

So Microsoft also started keeping backup plans. On one hand, it continues to provide compute and channels to OpenAI. On the other, it trains its own models, brings in competitors like Anthropic, and reduces reliance on OpenAI as a single supplier. OpenAI, for its part, also isn’t willing to hand over its lifeline entirely to Microsoft.

It starts looking for more cloud providers—partnering with Oracle, CoreWeave, and even Google Cloud—constantly weakening Azure’s exclusive position. Both sides keep emphasizing strategic cooperation on the surface, while quietly preparing contingency plans for life “even without the other.”

Anthropic: Like junior brothers leaving home—who’s the orthodox?

OpenAI and Anthropic’s grudge is the textbook “same school, family fight.”

In Anthropic’s core founding team, many people came from OpenAI, including Anthropic co-founder and CEO Dario Amodei.

Their reasons for splitting back then mainly centered on AI safety, speed of commercialization, and corporate governance. In simple terms: Anthropic believed they should be more cautious, while OpenAI believed products needed to ship first and make money.

So after the Anthropic team left, they built a very clear brand narrative for themselves: We care more about safety, we care more about interpretability, we care more about long-term risk—and we won’t floor the accelerator for growth.

Of course, the subtext of this story was obvious too—who’s the one stomping on the accelerator? Not us, for sure.

OpenAI naturally wouldn’t admit that it was the reckless driver. So the dispute quickly escalated from ideological arguments to product competition. ChatGPT vs Claude; OpenAI API vs Anthropic API; Codex vs Claude Code; enterprise customers vs enterprise customers; researchers vs researchers; safety narratives vs safety narratives. Once both sides truly entered the same commercial track, the former debate over ideals quickly turned into a very real fight over revenue. Later, even financial reporting standards and revenue recognition methods could become weapons against each other.

It’s like two junior brothers separating in a fit of righteousness, each insisting the “Way” is different and they can’t work together. After years of fighting, they only then discovered what they were truly competing over: whose mountain gate it would be, whose incense would carry on, and who would hold the title of martial arts alliance leader.

Musk and xAI: I hate that I created you

If OpenAI’s conflicts with other companies have some degree of business interest, then its relationship with Musk has a much heavier flavor of personal grudges. As everyone knows, Musk is one of OpenAI’s co-founders.

Back then, OpenAI still wore the halo of being “open,” “non-profit,” and “for all of humanity,” hoping to prevent advanced AI from being monopolized by a small number of big tech firms. Later, when Musk left, OpenAI stepped by step moved toward commercialization, became deeply intertwined with Microsoft, models became increasingly closed, and valuations kept rising.

From Musk’s perspective, it’s basically like: those hope schools we helped fund back then—after many years, you come back and find the public school has grown and become private, with a Microsoft sign still hanging at the entrance.

So Musk began to loudly emphasize attacking OpenAI for betraying the original mission. Attacks in words weren’t enough. He also founded xAI, launched Grok, and directly jumped in to spar with OpenAI.

From then on, both sides entered an extremely stable rhythm:

• OpenAI releases products, Musk mocks them;

• OpenAI raises funding, Musk questions it;

• OpenAI adjusts its architecture, Musk sues;

• xAI releases models, Musk executives reply with a few pointed remarks;

Both sides’ employees then continue lobbing verbal attacks from afar.

The legal disputes between OpenAI and Musk have piled up so much that they could almost support a standalone legal column.

Coincidentally, xAI had previously sued OpenAI over non-compete agreements and commercial secret issues—alleging that former employees brought Grok-related information to a competitor.

That makes Apple’s new lawsuit even more tangible.

Not long ago, OpenAI emphasized in court that employee job-hopping and introducing past work experience don’t mean a new employer is stealing commercial secrets. Now Apple has turned around with more detailed documents, chat logs, and hardware-related allegations, and even OpenAI’s lawyers barely have time to switch to a new set of talking points.

Finally: Why does OpenAI always turn partners into competitors?

Under the sun, there’s nothing new.

More than 2,000 years ago, ancient Greece had two of its strongest city-states: Sparta and Athens. They even once fought side by side against a common enemy and maintained a period of superficial peace. But as Athens kept growing stronger, the old hegemon Sparta became increasingly uneasy; and the more Sparta tried to guard itself, the more Athens felt itself being suppressed.

Ultimately, fear, suspicion, and conflicts of interest piled up layer after layer, evolving into the long-running Peloponnesian War. Later, the historian Thucydides wrote that the true cause of the war was “the growth of Athenian power, and the fear that this growth caused in Sparta.”

Later generations summarized this situation as the “Thucydides Trap”: a rising power rapidly grows and begins to encroach into the sphere of influence of an old hegemon. Because the old hegemon tightens its space out of fear and prevention, even if neither side intended to start a war at first, they will still gradually slide into conflict.

And OpenAI is Athens in the AI era’s rapid expansion. At the beginning, it was just a big-model lab. It needed Microsoft’s funding and compute power, Apple’s device entry point, cloud providers’ infrastructure, and the entire Silicon Valley ecosystem to keep feeding it.

Back then, everyone was willing to collaborate with it, because it had enough potential but wasn’t yet strong enough to threaten anyone’s roots. But once ChatGPT burst onto the scene, OpenAI stopped being satisfied with only building models. Step by step, it moved into search, office software, programming tools, browsers, agents, operating system entry points, and consumer hardware—it was redefining how people interact with computers.

Every step it took forward would land it in the core territory of some tech giant. So the companies that once welcomed it, invested in it, and supported it began to feel uneasy—more and more like Sparta. Perhaps there’s still an ultimate war to be fought between the AI newcomers represented by OpenAI and the traditional internet tech giants.