Apple's lawsuit against OpenAI over hardware secrets support Palantir CEO's warnings about AI
The artificial intelligence boom has been nothing but a headache for Apple (AAPL) since it started. The stock quickly gained a reputation as an AI laggard.
The company recently had to raise prices for some of its products due to the skyrocketing memory prices, caused by the AI buildout.
The difficult situation prompted it to start lobbying the Trump administration for clearance to buy memory chips from ChangXin Memory Technologies (CXMT), which is on the Pentagon's blacklist.
Apple is now going straight for the heart of the AI bubble by suing OpenAI.
Apple alleges trade secret theft by former employees and OpenAI
On July 10, Apple filed a lawsuit in a federal court in Northern California, alleging that its former employees Chang Liu and Tang Yew Tan, currently working for OpenAI, stole Apple's intellectual property in order to develop OpenAI's own consumer hardware.
The filing states some very damning things:
"This much is clear, however: at every level, from members of its Technical Staff to its Chief Hardware Officer, and in coordination with business partners, OpenAI has been stealing Apple's trade secrets and confidential information."
Drew Pusateri, spokesperson for OpenAI, posted a response on X:
"Our statement in response to this suit: We have no interest in other companies' trade secrets. We remain focused on building innovative technology that empowers people everywhere."
If the allegations are proven true, they will also prove that Palantir CEO Alex Karp's recent statements about the AI industry must be taken seriously.
Alex Karp blasts the AI industry on CNBC Squawk Box
Palantir CEO Alex Karp's interview on CNBC's Squawk Box quickly turned from what was supposed to be a discussion of Palantir's new partnership with Nvidia into a heavy critique of frontier AI model companies.
Karp talked about the issues very passionately, using too much expert-level terminology, to the point that what was being said left many people scratching their heads.
However, for those proficient in the inside baseball, he was absolutely making sense, though he could have done it a bit more coherently, as his thoughts were jumping too much from one topic to another.
Related: Apple's iPhone cost problem reveals AI's hidden bill
Here is the key part of the interview where Karp laid it all out:
"In this country, at every single enterprise I deal with, these people are livid. They're like, I am paying for tokens that create no value. These people are stealing the weights and alpha of my business, and they're creating a wealth tax that does not help the poor."
He also said:
"[If] I can make you $1 billion tomorrow, wouldn't I say I'll make you $1 billion, and I want 30%. Why are they charging for tokens if it's so valuable?"
If we unpack what he said, he raised three key issues within the industry:
- AI customers feel that Large language models (LLMs) create no value
- Charging for tokens is impractical
- Frontier model makers can steal their customers' IP if they wish
Pointing out the issues was certainly motivated by a promotion of Palantir's product Ontology. However, he didn't really try to hide that.
Here is what he said:
"We have this thing called ontology that now everyone's copying, but de facto, it takes a large language model, it makes it safe and useful and precise. So safe because it doesn't touch your underlying data, safe because it prevents the large language model from caching your data and replicating your business."
Karp continued: "Safe because it doesn't transfer your IP, or how to fight secret data, top secret data, or in a clinical context."
Former AI czar speaks up in support of Karp's statements
Former White House AI czar David Sacks posted a long post on X about why Karp is right.
Sacks points to Figma being blindsided by Anthropic's launch of Claude Design, and adds that it isn't the only example.
He wrote:
"Anthropic has launched Claude Science, Claude Security, Claude Legal, and of course, Claude Code - each expanding into categories previously served by companies building on top of their models. The pattern is consistent: watch where value is being created, then move in directly. Dominate the model layer, then use that position to capture the most lucrative verticals."
Apple's lawsuit now alleges that OpenAI was ready to do much more than that, and it should wake up many companies to rethink their approach to AI use.
But if the allegations are proven to be correct, they will pretty much prove that Karp's other key points make sense, too.
Technical limitations of LLMs make charging per token impractical
OpenAI has gone from promising to cure cancer, replacing workers, and enabling people to prompt into existence billion-dollar businesses, to allegedly stealing trade secrets.
The lawsuit presents another serious bump in the road for the company's IPO, on top of the previous one.
By the previous one, I am referring to tech writer and prominent AI skeptic Ed Zitron publishing OpenAI's leaked audited financial statements. The data was verified by the Financial Times. This leak revealed an increase in OpenAI's net loss, from $5.09 billion in 2024 to $38.53 billion in 2025.
Someone might defend this by saying, "What is wrong with charging for tokens? Everyone pays for electricity exactly how much they use it. No one is paying for the value of what you do with that power."
However, that would be making a wrong comparison.
When you get electricity, you use it to power things you get value from. Let's take, as an example, the simplest thing: a light bulb. It is absolutely easy to calculate how much that light bulb will cost you at the end of the month. It will also work predictably.
More tech stocks:
- Cathie Wood buys $22.8 million of surging tech stock
- 5-star analyst sends AMD stock investors a warning
- 5-star analyst sets bold SpaceX stock price target
The problem is that when you get "hallucinations" from AI, you still have to pay for them. You can never know how many tokens a prompt will end up burning, so you can never tell how much it will cost you. If electricity worked like LLM tokens, it would look very different.
You would never know how much electricity you'll spend when you turn the light on. It would be totally random. Sometimes, keeping it on for an hour would cost you as much as keeping it on for a year.
The light would sometimes be normal, sometimes it would start to work like a disco ball, sometimes it would produce just red light.
Calling LLMs "artificial intelligence" is stretching the truth. The term hallucinations is also misleading; it "humanizes" LLMs and makes it sound like they are a bug that can be fixed.
In a simplified view, LLMs are randomized algorithms, and what is termed a "hallucination" is perfectly normal functioning of said algorithm.
The reason AI companies charge per token is that AI inference (processing tokens) is extremely expensive.
These inherent issues have led to the end of the era of token-maxxing, with Tesla (TSLA) among the most recent companies to fold. According to The Telegraph, Elon Musk has set a $200-a-week limit on Tesla employees' AI spending.
Both Karp's interview and the Apple lawsuit show that the industry is in crisis.
If LLMs were really as valuable as their use costs, OpenAI and Anthropic would not be selling access to them but would be using them to solve real problems and make money from the solutions.
Related: Bank of America sets alarming SpaceX stock price target
The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc.
This story was originally published July 13, 2026 at 7:33 PM.