Apple Sues OpenAI for Trade Secret Theft as Meta Muscles Up on AI

Hey, everyone. It's Monday. It's a little different Monday.

I'm not in Techstrong's studio Monday. I'm in my office here at Techstrong's headquarters. We had a little technical glitch in the studio.

And I have to tell you up front, I've got my bulldog in the office today with me. So if you hear a snoring noise, that would be her. But that's who's in the office.

Let's talk about who's on the gang today. We got a great gang and some great stuff to be talking about. It's Monday, and this is becoming the Monday gang.

I like it. I'm a creature of habit myself. I like it when you get a steady kind of setup.

We've got my friend Liz Saffron. Liz, looking great. Welcome.

Thank you. Also looking great, Yvette Schmidtter. Also from New York, because we got like a New York thing going on here.

Mm-hmm. We got the ting, ting, ting. That's right.

Mike Fleiss. And holding up Ohio, Hudson, Ohio, the one and only Steven Foskett. Welcome, gang.

How are you? So we got four New Yorkers here. We're going to have a party.

Yeah. All right. Guys, welcome.

Mike, it was a busy weekend. Yeah. As usual.

And a lot of it started last Friday with lots of chaos and craziness, but unless you were completely on vacation somewhere else, you might have noticed that Apple is suing OpenAI and accusing them of stealing their intellectual property. " Because this is the stuff you used to read about back in the day. And then I was like, "Really?

They actually brought physical IP into a job interview? They really asked them to do this? " I'm like, I haven't heard of this stuff in 20 years, but maybe it's happening all along anyway.

But it just seems like it's pretty cut and dry, at least. We have to wait for the actual suit. But Alan, you wrote a post and played the judge in this case.

What's your take, and what's going on here? Because it sure seems like Apple at least, shall we say, has the goods. Apple has no goods in this case.

So first of all, I did. I wrote a really tongue-in-cheek piece called Judge Shimmy on the Silicon Valley breakdown. Because reading everything over the weekend about what goes on here in this AI continuing soap opera, I started getting a little bit of a Judge Judy, Jerry Springer kind of thing going on.

And we're going to try to hit some of these in here. Let's first address the Apple lawsuit, though, because that's kind of the long pole in the tent. Look, Mike, did you think it was the '70s?

No, it was probably closer to the '90s, early 2000s, when Steve Jobs said he was going to go thermonuclear- Mm-hmm ... on Android because they stole iPhone because Eric Schmidt, former Google guy, was on the board at Apple. This is the same story here.

This is a rerun of that. Right? And they really do not have a strong legal case.

They don't have a strong legal case, but their game here is to throw sand in the works of OpenAI, and their IPO, and their march towards some piece of hardware. Now, interesting things. They named in the suit the OpenAI head of hardware, or whatever.

He was formerly an Apple head of hardware guy, and his assistant they didn't name in the suit, and I wonder if it was a tip of the hat to him- Mm-hmm ... or so they just didn't want to go there. Johnny Ive, that's the real player here.

And he left Apple, and OpenAI bought him, his company, for six and a half billion, or whatever it was, and they've hired 400 ex-Apple people. There is nothing illegal about that, and I would venture that most of those people were probably laid off anyway or went with Johnny when he left. So again, there's a big nothing burger there.

They probably had non-competes, and I'm sure the claim here was not against a broken non-compete. It was against theft of IP. And the theft was they said they brought pieces of hardware into their job interviews.

But that's a very common thing when you're talking to hardware designers. You bring in examples of what you've built in the past. And so this is what they did.

I don't think there was anything necessarily proprietary about what they brought in. Did one person perhaps log into an email after he left? Maybe.

That doesn't necessarily implicate OpenAI unless OpenAI told him, instructed him to go do that. Don't blame that team there. The next big nail in this coffin is, well, what product is OpenAI doing that actually Apple claims they stole?

Because there is no product right now, so the whole thing's premature. If you don't have a product to point to, how could you point to them stealing something to make your product? There is no product.

But it gets better because this is Jerry Springer's Silicon Valley. Mm-hmm. With the coming of this lawsuit, our friend Elon Musk can't miss a chance to go Sam Altman.

So he goes on X and starts a whole X series of insults calling Scam Altman, calling Sam Scam Altman, excuse me, and goes on how Scam steals everything, Apple IP and the nonprofit. Elon's still sore about losing that lawsuit a couple of months ago that he claims he's going to appeal. Sam Altman comes back, and Steven, I know you'll love this, and says: "You're calling me Scam.

" I'm not falling for it. " So there it is. Elon says they're going to have space-borne data centers within the year.

But it gets better. Then you got the wild-haired guy from Palantir jumping up and down, claiming that AI is stealing people's data, and that's really Palantir's thing to do. Right?

They use the data to make their products better. But it wasn't enough yet. Then Satya weighs in, as the kindly, rich, old uncle who says, "Wait a second.

Wait a second. What you're doing is not right. " Something Microsoft's been doing for a real long time, too.

So what this all comes down to is that, and this is in this forthcoming book I've got coming out, is that no one wants to be on the bottom layer of commoditization. No. And pure model intelligence is going to be commoditized.

Everyone wants to move up the stack and build up the stack. Microsoft has lived up that stack for a long time. So has Palantir.

Right? They want to build on that stack. And frankly, Apple is the king of up the stack.

Right? They like to own the whole stack. Mm-hmm.

And now, all of a sudden, these upstarts, OpenAI, Anthropic, they're looking to move up stack and Frankenstein's monster's escaped the lab. And that's what we're dealing with. So I think you're conflating a few things here and there along the way, but- Well, it is Judge Shimmy's show, and I've got freedom.

Yeah. But let's just take a couple of facts apart here. First of all, we don't really know what these job folks brought to the interview, whether it was something that was in from Apple's labs or a product that had already been shipped.

I agree with you that technically, OpenAI hasn't done the crime yet because they haven't actually shipped something. So you might be premature on that side. " So just saying.

But whose theft is that? Huh? Who committed that crime?

OpenAI? Well, did they ask him to bring that? I don't know.

Well, I think that the- Well, that may be the allegation. This is the same allegation they made against Android. But go ahead, Yvette.

I'm sorry. No, it is not. So Mike is right.

You brought a whole bunch to this party Judge Shimmy. But there's a couple of things. So the first one, you mentioned they didn't name Jony Ive, right?

So I think Apple's position is that the real acquisition target was everything Ive's people remembered. So that's one. And I also think that in California, trade secret law is the last non-compete standing in California, and I think Apple has found a way to weaponize it.

And the last point that I'm going to say is this. Look. What everyone here is fighting for, everyone is in this fight, is they're swearing that the data belongs to the customer.

And every one of them is building an ecosystem where the data never leaves. And that's not a coincidence. That's a business model.

Those are my two cents on that one. I agree with the last one. Okay.

Non-competes in Apple, yes, there is trade secrets. So out of the 400 ex-Apple employees, how many of them were Jony Ive's employees? How many in between and stuff like that?

Even with that on the books, courts are very reluctant to enforce that stuff. Very reluctant. Yeah.

It's such a fine line, right? What is a trade secret versus what is the expertise of the person? I'm sure there's a whole bunch of legalese behind that.

But practically speaking, did they bring a device that was patented, that the patent belonged to, and their name, was it on the patent? Was it not on the patent? You get so far down in the weeds that the legal decision is probably fairly abstracted from the actual issue.

Which won't be happening for another three years, maybe, at the rate the court systems go. So just to Alan's earlier point, maybe the whole point of this exercise is to gum up the works. Steven, I haven't heard from you on this.

Well, okay, so in this case, we actually do have facts. This is not something- Really? that people are making up.

There's a real lawsuit with actual allegations, and the allegations are super specific. So I know that you would love to have this be a discussion of trade secrets and Apple versus OpenAI. No.

No. No. This is a specific case, and this specific case involves a former Apple employee who walked out with his company-issued MacBook, and then sent a text message to his friend saying, "LOL, I can still access Apple network storage.

" And then colluded with that person to download stuff from the Apple internal file server over a period of a year, according to the allegations, and took that stuff- But why is OpenAI involved there, then? How is- Because they all worked for OpenAI. So if somebody works for company A- Yeah, but Steven, was that part of his job at Open-- Did OpenAI- Yes.

He's the head of hardware design. He downloaded hardware design documents. I mean, this is a specific allegation.

" No, they're saying this person- And- ... did a specific thing ... and given who these people are, they should easily know better because you're not supposed to do this stuff.

And it's one thing if you're a low-level engineer and you're clueless, but this is the head- But guys, legally speaking, that's not the issue. Legally speaking- Yes it is. That's exactly what they're charging them with ...

the only issue is did he do this as an OpenAI employee under the direction of OpenAI, or did he do this on his own? All right. So what we- Under the direction of OpenAI?

Who's going to direct him? He's the head of the group. He directs himself.

Well, now he's the head of the group. Now he's the head. " And then once you have that, you're pretty much receiving it.

And by the way, it's not the C-level guy from OpenAI who did this. It's a lower-level person who did this. Well, I think there's a bigger- It's the allegation, if he did.

I'm sorry ... there's a bigger miss here. So all you CISOs, all you VPs in IT, you all still have not understood the basics.

When someone leaves, turn off their access. You're still leaving people access to stuff. This is like you created your own monster on this one.

Agreed. You know what I mean? Agreed.

How is that not automated? If you can't control patch management, rotating, terminating people's access to credentials when they leave, this is a you problem. So you can- I mean, and this is Apple.

You bet. This is Apple. Yeah.

And if Apple can't do it- They have the security ... nobody else could. Like so like- Wait a minute, wait a minute.

You're both blaming the victim here. I'm not going down that route with you. Oh, God.

Hold on a minute. Just because I left my car keys in my car and you stole it doesn't mean it's my fault, right? Yeah.

Absolutely. Well, it would be considered a mitigating circumstance in a lawsuit. Well, it might be more like I left my keys in the car, I left the windows down, and the door open.

So what? Doesn't mean we can steal my car. You're right.

It doesn't mean you can steal the car, but here's the thing. It is like you are supposed to be one of the leaders, right? And if you are not even following fundamentals, you brought this upon yourself.

" Yes, it did happen to them, but they had the control to turn off his access. They failed to do so, and here's the ramifications. I say this to people all the time.

You can't pick your consequences. You can't. They can't pick it.

Look, this- You can't pick it ... this is Tim Cook ripping a page from Steve Jobs' book on his way out the door. That's all it is.

Well, like Mike was saying, we're going to see. This is going to be very interesting. But it's also going to be long.

This is not going to be quick. This is going to be in the courts for a bit, so- No, no. With Google, the Android Apple one took I think eight or 10 years.

Yeah. Yeah. Well, there's likely to be a device by then.

I mean, we know that they're planning to build something. Yeah. Like the next iPhone or whatever that's going to be.

Right? 5 billion for Johnny Ive's company. What did you think they paid that money for?

Of course, they're going to build something, and they're entitled to build something. And just because someone worked at Apple one time in their life doesn't mean they're precluded from building something. I just don't understand why you all are defending these people.

I know. They're all- I just don't get it. What's happening here?

This is Apple biting the square. Are we living in an upside-down universe where it's okay to steal stuff from work? No, it is not.

I'm with Steven on this. The complaint alleges specifically, thank you, Steven, for writing this down. They used their access to that file server to download presentations, hardware designs, manufacturing details, and testing procedures.

Those are all IP belonging to Apple, correct? Correct. And that's the thing.

But that doesn't mean Apple- Yes, Apple accidentally left some access- But that doesn't make OpenAI guilty ... but they apparently closed off other access. That shows intent.

I'm not going to litigate this here, but I don't understand how you can defend these people. No, no. Were they an OpenAI employee when that happened?

Yes. It says that they did it for a year. It doesn't say that.

It says they were coming in for interviews. No, it does say that. It says they've been downloading it for a year, and they've been an employee of OpenAI in that year, Steven.

And has OpenAI had access to that? " You think this doesn't happen all the time? Come on.

Oh my gosh. How could you be so jaded? How could you not be jaded?

So OpenAI, here's how I think this will break. If OpenAI had nothing to do with this, they'll fire those employees tomorrow and say, "Sorry, we didn't know anything about it," and we're back to square one. If they fight this, then they probably are using it.

They're going to fight it because look, whether you think he's Sam or Scam Altman, I don't really care. Right? But look at what he did with Elon.

Elon, his whole lawsuit that it was supposed to be a not-for-profit, and it was Elon's company, and blah, blah, blah. You know what? OpenAI took a scorched earth policy and said, "We're fighting this.

" And they won. And because of that, they're going to fight this, too. Well, of course they are, but it doesn't mean that they're going to win every fight- No ...

that comes out of it. No, it doesn't. What do they say on Wall Street about past performance is not indicative of future results?

You know what? So to prove these kinds of cases, Steven and Mike, with all due respect, you're focused on the actions of a single or two people. That doesn't necessarily imply liability to the company.

Not yet. Is this a civil trial? Will this be in front of a jury?

Of course it is. Okay. So, I mean- A civil trial would get brought by the state.

So- But the next phase of this will be a discovery process where we'll find out what OpenAI knew and didn't know. But we can't find that out until there's a lawsuit filed. So there you go.

Yeah. And it does bear a striking resemblance to let's play thermonuclear war, which Steve Jobs did with Android. I don't know, man.

Starting to smell like Richard Nixon and the plumbers, man. I didn't know nothing about it, man. What 18 minutes?

Well, it's like Zuck I see walking out now. You know what? No mention of Meta on this one.

But look, the bottom line is it's a tangled web. And the thing is, this whole group of these tech bro billionaires, all the people in my little cover graphic I did for that article, none of them are particularly likable. They're not heroes to the working class or anything like that here.

I don't disagree with you that the whole AI business model is a giant roach motel where data gets in, but it doesn't get out. Mm-hmm. But I am not sure that that connects to this case, but that's just me.

All right. You know what? We shall see, as they say.

Steven, not a word about the data centers in space? Not going to be baited. All right.

Within a year, he said, Steven. We'll see it if it happens. All right, Mike, what else do you got?

Before we go on, though, I should point out that this actually bears no resemblance to Apple versus Google because they never filed a case against Google. They never sued anyone over that. Oh, no, I believe there was a case about Android.

Okay, you can believe what you like, but it didn't happen. Actually, it might've been Samsung. Yeah, they sued Samsung.

Yeah, they sued Samsung. It was over Android. Yes.

I just got to say as I listen to this conversation, thank God that the legal process has a thing called discovery. Well, now you know why I do tech, right? I like tech.

All right. What else do we have for today? All right.

So Meta has been for years now, they talk about AI for as long as I can remember, but it was always in the context of how they were going to use it internally. Now things are changing. They're building models that they're going to share with other people, so you can use them to go build.

They had some open source stuff that they handed off, but you were pretty much on your own. But now they're saying, "Hey, we got a better model. " And they're going to become essentially a cloud service provider for AI models.

Liz, is Meta changing as an entity and a company now? They're basically going to be competing against OpenAI and everybody else head to head. Yeah, I think they're clearly trying, and it's not like there's nothing there, but I feel at heart they're a consumer company.

I am not a fan of Mark Zuckerberg, and I think that out of all the news from Meta, the most interesting one is the data privacy thing. Because I think at the end of the day, who is going to trust Meta with their data if you're an enterprise? Not that any of the other AI, like we're talking about the AI business models, is data comes in, but it never leaves.

But I think that Meta has a track record of misusing people's data and violating their privacy. So would you really want to move all your compute over to Meta? I wouldn't.

So you're saying they might take my data and use it for some purpose that I didn't understand or be aware of. Yeah, and I think they're taking big swings in a bunch of different directions at once, right? Okay, it's one thing to try and compete on the models themselves.

There's probably some there there, and that's not my subjective expertise, but they've been at it for a while. Competing with AWS, I don't know. But the one that resonated with me was the privacy issue, right?

Let's say they make some traction. Let's say they steal some customers. Let's say they get somewhere.

Then what? I don't know. Here's my problem with that line of reasoning, though.

It's like, all right, so if we're going to make likability a part of the purchasing decision pattern here, I'm not going to get very far with anybody. It's not likability, it's trust. Yeah?

It's trust. Trust. All right.

I get that. And you can't gauge future performance on past behavior, but look at the past behavior. It's like, fool me once, shame on you.

Fool me twice, shame on me. All right. Yvette, can we trust anybody in this AI space?

Is that the core problem? I am going to say this, and you all, I write about it all the time. So an AI swapped my face for Jensen Huang, so no, I don't need a white paper to explain why default on is a problem.

Consent that requires three menus to withdraw was never consent. It was a hostage negotiation. So no.

There's a trust and verify. No. I don't think anyone today, where your data is more valuable than a vote, can say that trust is something that you can think they're going to do the right thing.

It's not there. Trust no one. I don't know.

Mr. Schimmel, trust no one? Over 30?

Does that ring a bell? Yeah. What was the movie?

Logan's Run. Yeah. Yeah.

Awesome. Here's the thing. I wrote a bunch this weekend, another article that'll be published shortly Look, Meta's been playing catch up even though Zuck has put in literally billions and billions and billions, if not tens, if not hundreds of billions of dollars here.

5 generation of OpenAI. So it's one generation behind. 6 and Fable through stuff.

Huge improvement over previous versions. Gemini is around 54 as well, so it's on par with that. But here's the thing the world's going to be coming to, and I think this does go to what Karp and Palantir and a lot of the people are saying.

Number one, we've got to get straight the use my data thing. Right? If AI is going to do what it needs to do, we got to get the data part of it right.

And right now, here's the funny thing. Big companies, Fortune 500 companies, when they're doing their deals with Meta or OpenAI or Anthropic, they are getting guarantees against their data being used. It's the rest of us that are getting sucked dry.

Okay? That's number one. Number two, I think we've got to come to a world where you don't need a Cadillac for every job.

Sometimes the smart car gets you-- You just got to go pick up milk at the market. The smart car is the better car to take. And that, again, is going to be part of this maturation process that we're going to.

1 good enough? Yeah. Right?

So, I want to be clear here. You're saying that when I log into ChatGPT and there's this little thing at the bottom that says that my data will not be used to train AI models, that I can't trust that? Where is that line there?

Well, I guess the bigger question is did you go into your settings and actually opt out? Yeah. So- Right, but you're not an enterprise- But this is the guy who leaves his keys in the car with the windows open because it's not his fault.

So, here's the thing. I'm a New Yorker. Look, my spidey senses, you can't kid a kidder.

I wasn't born yesterday, but I just didn't fall off the turnip truck either. At the end of the day, right, we have the assumption that they're going to do the right thing, and that contract, that social contract that we have with getting your car insured and getting unexpected and having the Department of Insurance, making sure all these things, that does not apply when it comes to AI. None whatsoever.

So you thinking that this is commercially available and it's out there, they're not going to do bad with it. Well, that social contract doesn't exist when it comes to artificial intelligence, and Meta has shown us through what they did with Facebook to what's happening with Google, where you say you opt out, but they didn't really opt you out. We've seen it.

It's just not me being Debbie Downer, Negative Nancy or anything like that. This is stuff that's out there that you all seem to forget. Google has done this, has already been brought to bear on this.

Meta has shown us and shown and proved. 7 font at the bottom of the prompt saying that they're not going to use your stuff, then I got a bridge to sell you in Brooklyn. Yep.

Because they're using your stuff, will always use your stuff until you physically go in and turn things off. Now, if they get caught still using it, that's totally different. But you assuming because they said it, they're going to live to it?

Come on, man. Mike, come on, Mike. You know better than that.

He's been up in the Adirondacks too long. Yeah. Yes.

Amen to that. Amen to all that. I would say that there's a subtle difference here, too, that they'll probably use as a legal fig leaf that Alan likes to have.

But technically, there's a difference between using your data and using your metadata, right? So if I kind of have your data, but I extract some assumptions about you because I also have the metadata that you're wrapped around, so I kind of know what your IP address is, and I know what you kind of like. I don't really need your data to train the AI model.

I can use the metadata and use it that way. Correct? Very true, and with the amount of data breaches that have happened over the last several years, out on the dark web, you can buy one of the breached data and do a traceability matrix and back into who you are.

So yeah, the hardcore data, right? The PII or the metadata, you can have AI help you do it. You can back into whoever that person is with more detail and richer information with what's already out there.

So, I'm going to say this again. It's just like, look, you willingly going in and vomiting your whole life stories and your deepest fears into an AI platform and thinking good stuff is going to come out of that. Look, I got two bridges in Brooklyn that I can sell you.

Can I just mention, my favorite bridge in Brooklyn is the Kosciuszko Bridge going from Brooklyn to Queens. They rebuilt that bridge. It's beautiful.

It is. Well, that's not the one that we're going to sell you. No.

Maybe we will because it's really nice. Kosciuszko, they just rebuilt it. But that being said, look And Mike, by the way, I just had a look at both my apps for chat and Claude.

I don't see anything in there that says we won't use your data. Well, I guess the one I look at is the corporate- Well, there's an opt-out ... one that we have, and that's it.

It's on the bottom of the thing, so- Yeah, there are opt-outs. But I have one thing to say about the whole Muse thing, because the image stuff made me think about yet another lawsuit that we haven't talked about, which is the one that The New York Times and Sarah Silverman and everybody has against OpenAI and Anthropic about whether they could just take their copyrighted stuff. And the argument that the AI companies are using is that they're transforming all that stuff into something new, so it's no longer the same thing.

So if Meta can take all this image data and say that they're turning it into something else, then the end user is losing control of their data. And if they win that lawsuit, I think that that's bad news, really bad news for consumers. And businesses, too.

So I think that maybe now we can talk about the other issue that came up that's not really part of this, what Alex Karp is suggesting, that companies are basically giving up the golden goose to Anthropic and OpenAI. Now, he's not arguing that Anthropic and OpenAI are stealing companies' data necessarily. What he's arguing is that it's stealing companies' business models and companies' alpha, in his tech bro speak.

But essentially, that these companies are, by hosting the processing of the data, essentially learning how you do business, how you do what you do, and then putting themselves in a position to usurp you in that. And that is exactly what you all are upset about from a consumer standpoint. I don't think anybody thinks- But how do they do that without actually using the data?

Well, forward-deployed engineers. Ha ha. Yeah.

I was just about to say that. Well, yeah, there is a kettle- I was told- ... calling the pot black over there, right?

Yeah. Well, I'd love to talk about that if I can. Go ahead.

All right. So how they do it is actually quite simple, and it relates to what Yvette and Liz were just talking about, which is essentially by watching what you're doing, they can learn how you do what you do, how you make the decisions, what you do. They don't actually have to look at the data at all.

What they have to do is trace the logs and see, let's say Alan uploaded a list of companies that are attending Black Hat. Then Alan asked which of these companies have a new product coming out. Then Alan asked which of these companies- Is that not my data?

might be interested in sponsoring something. Aha, Alan is prospecting for companies at Black Hat, so therefore we, evil AI, can do the same thing in a parallel process that's unrelated to Alan's data and do that for every tech company. And then we can say, "Oh, well, Alan also processes all of the podcast descriptions and show notes through ours, and let's take a look at how he does that.

Let's take a look at the structure. Let's see how he does it. Oh, you know what?

" And none of that involves your data. What it does is, you were calling it metadata, it involves your processes, it involves your business processes. Now write that larger and imagine you're Citibank or Archer-Daniels-Midland or Ford Motor Company or anybody.

And if these AI companies can see essentially the kind of questions you're asking, the thought process behind those questions, then they can build a model that doesn't necessarily infringe upon your intellectual property, but really does. So based on what you're saying, we should lie to our AI models about 20% of the time just to keep them confused and guessing? What do you say?

Well, my suggestion would be to use many, many different providers and rapidly juggle between them so that nobody can see what you're doing because everybody sees just one of the queries. Nobody can see all that you're doing. But Stephen- But wouldn't you need another layer to manage that?

Wouldn't you just be relying on a different kind of Palantir? Yeah. That would- And that's exactly what Alex Karp is trying to sell you at Palantir.

Exactly that. Imagine that. But here's the thing, Stephen, you just described what many in the webinar business call intent data, right?

Yeah, but I was just using that as an example. Okay. Just wanted to make sure, because I was about to shut it down.

Are you suggesting that we're collecting data for purposes? We don't collect anything. All right.

Let's go to topic three before I'm the next person in this Apple lawsuit. There you go. All right.

So more legal issues out there, but EU has basically cut a deal with SAP finally, where it says that SAP cannot force people into support contracts and make these kinds of broad edicts that they have been making. And SAP's not the only company guilty of this. Let's be clear about this.

But I wonder if we finally reached a time now where people are going to say, "Hey, no more of this. " And this is just going to become the standard way we do business across all these applications. What do you think, Yvette?

Am I too hopeful, or is this just one-off little thing, and then we'll forget about it next week? I think you have a little bit of a bear there, but out of all the stories, I think this is the sleeper story that most people are just scrolling past. You know- They agreed to overhaul its maintenance and licensing practices globally for 10 years.

That makes the EU antitrust investigation go away. They didn't admit to anything, no fine. Right?

And that's exactly why this really matters, because SAP looked at its own contracts and looked at what discovery would reveal, I love the legal aspect. You dig into stuff, discovery, and chose to surrender, not fight it. Right?

So for decades, we all know this, enterprise software's dirty little secret has been maintenance fees. Right? The most profitable, least competitive revenue in the industry, in my opinion, were protected not by quality, but by contractual handcuffs.

Right? Mm-hmm. So the EU just declared the lock-in itself an antitrust problem.

So, but here's the fine print that nobody is reading. The deal covers on-prem only. So the S/4HANA, Cloud, and RISE are untouched, explicitly.

So SAP conceded the shrinking estate and kept the walled garden exactly where it's hurting everyone next. So the regulators freed the prisoners from the old jail while the new one was still under construction. Hmm.

So if you're an SAP customer and you're not renegotiating your contract like right now, this quarter, you are voluntarily, voluntarily donating your leverage back to the vendor. So- Yeah ... here's one thing.

And this is what's not being said in between the lines, so we're going to read the tea leaves a little bit. So the cloud carve-out is the whole game. So SAP spent years using support terms and deadline pressure to push on-prem customers toward what?

RISE and SAP S/4HANA. Right? Cloud.

Mm-hmm. So this settlement makes that old estate far more flexible, divisible installations, per environment, support choices, third-party maintenance, single, all that good stuff. Right?

But their strategic weight shifts from to the cloud, where none of that stuff applies. So the lock-on didn't die, it just migrated. That's my take.

So will people read this and then decide that, "Well, you know what? " Yeah. So maybe the time is- I do think SAP and Oracle, they've been the main providers in the market for years.

Right now, the EU Commission just blessed that whole model. But I expect the sector to grow, and I expect Oracle's lawyers to be reading the settlement guidelines very, very clearly, very, very closely, because Brussels rarely stops with one vendor. Right?

So now we have the Brussels effect with SAP, an EU investigation that produced global policy changes. So this is a tell. It happened with GDPR, it's going to happen with this, and it's going to actually happen with the EU AI Act as well.

I think the EU enforcement wave is not going to stay in Europe either. That's my honest opinion on that one. Well, but it's not just those.

They also came down on Google. Right? On search, and made Google change their search.

They also came down on Apple and the App Store. Yeah. Right?

So clearly, they've got an agenda. Right. But this is probably the story I know the least about in terms of all the details.

But to Mike's point about will more people end up moving their data on-prem, sort of as a paranoid backlash to everything that's happening with the AI models, and what for the customers that do have on-prem data centers, what data is that? Because aren't they highly regulated companies that might be loath to actually move to the cloud? Is there an assumption that once they're free to do so, they will?

Because that's sort of what I'm hearing. Is that what you're saying? Right.

And there's a subtle difference there. I can bring my own cloud, so I can put my data in a cloud that I manage. It's just that I'm not going to use the SAP cloud that they manage.

So that's kind of where that little subtle line can be. So when we say on premise, we're kind of saying anything that doesn't really manage by SAP per se, but it doesn't have to be in a local data center. But the other way to think about this too is to say, hey, if this is how this is going to play out, does the EU have a better understanding of the business models of these companies?

Or are we in the United States just ignoring all of this stuff because, well, it's maybe inconvenient to trouble some of these companies? SAP's a German company, but Oracle's a US company, and so is Microsoft. I think it's a question of political will- Yeah ...

not necessarily a better understanding. They have a political will to do so. And the other thing the EU does that I admire is they don't just phase in their regulations or these things in a day.

They actually kind of put it out there, and they give it a two-year thing, and they let it cook, right? Before it's done. Well, here's another thing that is another tell.

So customer collectives work, and it wasn't the Fortune 500 CIOs that got this done. It was a German-speaking SAP user group that had the seat at the negotiation table. That's who did it.

Right? So organized customers extracted concessions that no customers, got its concessions that no single Fortune 500 CIO could That's governance lessons that most enterprises haven't even internalized. Your user group is a negotiation weapon, not a conference.

Mm. So the AI industry should kind of lead, kind of read this as a prophecy. Today it's ERP maintenance lock-in, tomorrow it's AI model lock-in, fine-tuning weight portability, data egress from AI platforms to agent framework switching costs.

Every AI vendor currently designing these sticky contracts just watched a preview of their own settlement. So wait a minute, why do you like this phase-in approach? Because what I heard was, it's okay to continue to steal my car for two more years while we figure this out.

Look, no, I'm not blaming stealer, I'm not blaming vict- ... Look, it is what it is, right? If you want to keep going, okay, there's an area.

There's literally a town in New Jersey where the local police department has to send out a reminder for people to lock their cars. All I'm saying is that there's people who come to these towns to steal cars. Why?

Because the doors are unlocked, and the keys are in the ignition. And they come back multiple times to the same house to re-steal the car. So there's a point that I'm making here, right?

You have a responsibility to take the keys out of your car if you don't want to lock it, right? If you don't want it get stolen, right? Mm-hmm.

Got it. So you can either do that on repeat, or you can lock the car. And the argument that I'm making is just even if it takes two years, right, they're having these discussions, the onus is still on you as the individual to do what?

Take your keys out of your ignition- Take responsibility ... and lock the car. Right.

Let me also make a- Let me clear this point up here. I may be stupid for leaving my car windows open, but I have a God-given right to leave my car out there and have it not stolen. Yes, you do.

Right. And if you get stolen, it gets stolen. And maybe you don't have the right to steal it, but they're going to steal it, right?

But they don't... That's okay. You can claim your God-given right as you're walking.

Right. But the point is from a practical point of view, SAP had to get something to settle this because, quite frankly, if they didn't settle, it was going to take at least another two years for it to wind its way through the court anyway. So what did they give up?

Not much, right? The two years, it was going to take that long for this case to go through. So that's the art of the deal.

I think someone wrote a book about that once. I guess. Stephen, I still- Yeah ...

interest you to say anything? I would love to. Well, first off, kudos to Yvette for actually...

I have very little to add to what Yvette said about her analysis of this. It was spot on. One of the things that she said I want to highlight here, and it kind of relates to what you just said as well, which is essentially what is being challenged and what SAP gave up here is the old way of doing business.

It's the old world of enterprise. This is like the equivalent of Microsoft agreeing, I don't know, not to ship Windows on floppy disks anymore or something. This is how enterprise software used to work.

It's why these companies became huge, because they bundled all these things together. They forced you to buy the super double platinum ultra diamond license for every install, even the development server, if you had it anywhere on your site. They forced you to buy everything they possibly could.

They forced it down the customers' throats. Customers hated it. EU hated it because the customers complained to the EU, and now finally they're giving it up, but they're only giving it up now because that's not how things are going to be done anymore.

Right. Because in the future, we live in a completely different world. We live in a cloud world, we live in a SaaS world, and now we live in an AI world.

And it's not like the AI companies have a business model, but if they did, I imagine they wouldn't force everybody to use the ultra-secret double platinum model. They would probably have something a little more flexible. And so basically, SAP took a look at this and said, "You know what?

" And that's kind of my perspective. " I don't think it's just Accenture. I think there's a whole ecosystem of- Yeah ...

of providers. I'm just picking them because they're the biggest. Well, so is SAP.

And good for them. They can do it now. Yeah.

Guys, we're over time, though. I've got to end this. What a dynamic, exciting Monday here on techstrong® gang.

Stephen, Yvette, Liz, Mike, thank you so much. Thank you all for watching. Just a reminder, we do this every weekday at noon.

Mm-hmm. tv, our techstrong OTT app that plays on just about any screen you want to watch this kind of stuff on. Our techstrong TV YouTube channel.

We also are on demand on all of those places as well. We will be back tomorrow at noon with more, but until then, thanks for watching, everyone.