Google’s Final Antitrust Defeat and the Rise of Forward-Deployed AI

Hey everybody, welcome to Techstrong gang. Happy Monday, and as usual, we got all kinds of mayhem to talk about, but let me introduce our guests to start with. Stephen Foskett, how are you, sir?

Good to see you as always. Good to be here. Happy 4th of July.

I don't know, happy new quarter millennium. Happy something. We'll see how it all goes.

But at least we survived 250, and hoping for the next 250, right? All right. I'm hoping for the next one.

I'm a little more optimistic, I guess. Yvette, welcome to the show. How you doing?

Always good to see you, Mike. Good to be here. All right, good to hear you.

And Liz, can you hear me or what's going on? No. Liz has got some issues with the mic.

She may need to reboot, we'll see. And then Alan's coming late, so we got all kinds of joy. But, happy Monday.

I think they call it stormy Monday and Tuesday's just as bad, and if you don't know the rest of the song, the rest of the week kind of sucks till Friday anyway. So there you go. All right.

Well, I'm going to jump right into this thing and just get us started here, but Google's been fined again by the European Union. This time for, I forgot what the number is off the top of my head. 7 billion.

Here we go. So, this is a regular occurrence now it seems like. The EU is just fining Google and all these other companies, and they all seem to nod their head and go, "Well, we disagree with that," but it doesn't seem to me that they're actually changing any of their behavior as a result.

So, the issue is, I guess I'll throw this out to Yvette first then is, are these companies just too big now and they're too big to be fined, and they're too big to care about it? Or it's an annoyance, but they're not going to change their business models because they got a fine from the EU? Well, we'll talk about a term.

I love doing this, Mike. So, I think the auditors call it the materiality threshold. So, which is, how much money does a company lose before it actually notices?

So below that number, it's not going to change a single thing, and that's exactly what we're seeing. So this is just the cost of doing business. It's a rounding error.

And so, I don't know if you've done this, Mike, but I went back for the past 15 years and counted up all the fines. It's close to 11 billion euros that Google has been fined, right? 7, it doesn't even crack 3% of Alphabet's annual profits.

So, there's nowhere in there that this materiality threshold... This is just cost of doing business for them. And I'm going to say this early.

This is just a continuation of what's been a long-going chapter, and what they've just done is like, "Hey, we are not going to change what we're going to do. " That's it. That's the only thing.

All right. Liz, you want to jump in here? But it seems like the question is, it's not just Google, but are there maybe seven or eight companies that are just too big to fine?

Yeah, well, it seems that way to me. Yvette, you have a great point, but I'm glad they got fined because I feel these companies are... I actually think that the nature of the fine seems a little bit like, didn't we do this 20 years ago with Microsoft, right?

Bundling Explorer. Doesn't seem like it's anything new, but it seems like a way for them to be a gadfly, a pain, and show them that they can't just run roughshod. But, like you said, it's kind of a rounding error, so are they just buying off the regulators, or is there any kind of meaningful checks that the EU can put on Google?

I don't know. Ooh. Well, I don't know if you all peeped, a lawyer close to the case said this ruling just closes the first stage of Brussels fighting with big tech.

Stage two has already started, and it's called the Digital Markets Act, and it doesn't wait eight years for a verdict, right? It watches in real time. I think this is where the rubber hits the road.

So this is how we see real time, and this is how we're going to figure out how the fine never touched behavior. Because the original case about Google, them favoring their own shopping service over everyone else, search results, classic, self-preferencing, right. So right now, in February of this year, the European Publishers Council filed a fresh antitrust complaint against Google.

Same accusation, same legal article, Article 102, different product. This is about its AI overviews, right? So it's, again, AI summaries sitting at the top of search results- ...

allegedly answering the questions and keeping the click instead of sending you to the publisher who's actually reporting it. So, it's the same thing, different product. Same thing- Mm-hmm ...

different product. So are they going to learn? I don't think so, because right now, they're continuing to do what they've been doing for the past 15 years.

The Digital Markets Act is still being violated. Again, so if we follow the timeline, fine for self-preferencing in shopping, fine for self-preferencing in Android. Now under investigation for what looks like the same exact thing, wearing now an AI costume.

So, it's not that the company got caught once. It's a company whose product roadmap just keeps on arriving at the same violation It's just a new interface. Well, you could also say that maybe the European Union is wrong here, because they just keep attacking all these big tech companies, basically for a litany of different things.

Think about the Google cases versus Apple, versus Meta, versus everybody. There's another aspect to this where you could say, wait. As you said, I don't think that this is a rounding error.

I think it's actually a pretty big number, and considering that they fought it for eight years, I would say they probably think it's a big number and an important thing too. But that being said, ultimately, it's not going to really change their behavior. Right.

On the flip side, look what's going on with Apple. My European friends are a little bit vexed that Apple came right out and announced at Worldwide Developer Conference that they won't bring a lot of features to Europe because of the threat, just like this. And I would be not at all surprised if Google did the same with Android.

If they just said, "Well, you know what? " And if that's the case, well, then I guess you could say that these actions are successful and that they're protecting European consumers from activities that they don't like. But I guess you could also say that they're unsuccessful and that European consumers don't get to buy iPhones and Android phones.

Right. Or at least not fully featured ones. And not to mention the lost tax revenue that the EU may see when suddenly less phones are being sold.

But that brings me to the next question. I can't help but wonder, and I'm occasionally cynical, but it seems like these fines seem to land at just the sweet spot, right? It's just enough for the company to kind of pay and be reasonably annoying, but billions of dollars for the EU, which goes to help fund various things that they're doing.

So I have to scratch my head and say, if you're serious about it, at what point are you going to come up with a fine that is something that is material, or are we just going to treat this as a new source of tax revenue and just keep fining these companies for time and time again? Yvette, what do you think? I think that at the end of the day, the fines are going to continue to mount.

They're going to continue to have this little dance. But at the end of the day, this is what the European Union is showing its citizenry. We are trying to protect you.

And here's the one thing that I think that, of this whole story, that should actually worry Google's investors more than any fine. And a fine is a number that the company can absorb. Sure, they got, what, 15 billion up to now over the time?

But trust is not. Once you lose trust, it's gone. So I do believe that the fines are going to come to a point where the investors at Google and the company itself are going to come to a point where we got to do something, because we can't lose people's trust.

All right. Let me take it in another direction, though, and I'll throw this to Liz. We saw during the World Cup that the president is willing to intervene in even the smallest matters.

And so there was a red card, and suddenly that red card was pulled, and we have a player again for United States. What is the probability that the president's going to call up Europe and say, "Hey, this is yet another example of you guys trying to do stuff that is anti-competitive to US companies," and then respond in kind, and then try to force the EU to back off on these fines? What do you say?

Well, I'd say the chances of him doing that are pretty high. The chances of them just bending over and complying, I don't think that that's as likely as it would've been just about a year ago. But just to your point about these fines just being a slap on the wrist and not really...

And Mike, you were saying that, well, then Google and Apple, they'll just leave, and they just won't compete in the European markets. So you hear that all the time, right? " But if they're not happy with their practices, maybe instead of fines, some sort of other consequence, like market limitations, might not be such a bad thing, right?

And so if the European Union is willing to have Google or Apple or whoever back off, then isn't that the point of all this? If they're really trying to protect their citizens, then maybe that's not such a bad thing. And maybe that might even be more effective than fines that they could just pay, like they're leaving a tip in a restaurant.

Well, all right. So Steven, will this just create a black market for phones? People will run to the United States and pick up a phone and bring it back with them and it'll be like smuggling- Yeah ...

fruits and vegetables or something on the plane. Yeah, it's a funny thing, isn't it? Like I said, already Apple has aggressively moved to restrict access to some of their signature features, including the new enhanced Siri, and for a year now, Apple Intelligence, though it seems like they're enabling some of that now.

And so clearly, I think that the Digital Markets Act can have an effect. And also consider the impact that it has had on Apple's App Store, which has absolutely been opened up. The whole conflict there about the App Store being closed, and now Apple has a version that can run third-party apps from a third-party app store for Europeans.

" But I think you might also say, "No, it's not working," especially when the cool new features don't come to Europe, ever. And I could totally see somebody. Already people come to the US, buy MacBooks and iPhones here because they're cheaper here than in Europe.

I could see somebody activating it here with an American account because then they get all the features that they're paying for. So I do think that this is having a material effect on the companies, but I also think it's having a material effect on European citizens, and I wonder if they're going to be all that happy with the effect that it's having. Mm-hmm.

My cousins won't even come here. " I guess so. I guess so.

Yeah. Yeah. All right.

So if you look at this for a minute here, Liz, I'll put this to you. Does Google care about its image as it relates to this? It used to be, it was Google for good back in the day.

Yeah. But they just pretty much given that up at this point, and they're just another one of these- Well, you know how long ago did they change their tagline, right? Gone are the days of do no evil, right?

Or don't be evil or whatever it was. So I don't know. I don't think they're too concerned, but I don't know.

But no, I think that they're p****d about losing. I don't really think they see it as a breach of trust. Do you?

About the fact that they lost, or whether or not they see it as- Well, I think they're p****d that they lost, but I don't think that they think that they're losing more, but than money. Right. Well, here's the part I don't get.

So, and Steven, I'll throw this your way, but couldn't we just do split the middle here and let people who live in Europe opt in, therefore, they know that they are getting involved in whatever it is, and they have a choice? And then there's free will involved in this, and nobody needs to be having a hard rule one way or the other for everything. " It seems reasonable to me, Steven.

Yeah, I'd like to see some flexibility here. I for one love the fact that on American phones, whether it's an Android or an Apple phone, you can change the search engine provider. I love the idea that perhaps with Apple Intelligence, you might be able to opt in to a different AI provider.

Again, that's a rumored future feature for it. But, I think it would be awfully nice if we could have the option to do things like that. And I would love to see a global initiative to encourage platforms to have options.

But at the same time, I also appreciate the fact that iOS being locked out has allowed it to be a very safe computing platform. Probably the safest- Mm-hmm ... computing platform in the world, simply because you can't do some of these things.

You can't have malware run on it, and Apple has moved aggressively to protect users for that. Would I like to see it opened up? Yes.

Would I also like to see it locked up? Yes. So I guess I'm in a conundrum here.

Personally, if I had the option to use a third-party app store and to install random software on my phone, I would not do it, and I would tell everybody I know not to do it. Don't do it. But it does kind of chafe to know that I don't have that option, and I'd like to have that option.

So I'm with you. I'd like to see that. I think European customers would love to see the option of opting in or opting out to things that they want or don't want.

But at the same time, well, let's make sure it works. Mm-hmm. All right, well, I'm going to leave this here since Alan just showed up, and we'll go to the B block, and we'll move on from there.

But Alan, welcome to the show. How you doing? Thank you.

I'm doing great. I apologize. I had a doctor's appointment, ran a little late, and just got back to Techstrong Studios.

Good to see you, Alan. Yes. Come on in.

Good to see you guys and ladies and a lot of friends on today's show. Yeah. There we go.

All right, well, I'm going to shift to B block, and what this is about is that both Microsoft and Amazon Web Services of late, but they are not the only ones, are tossing around this term forward deployed engineer, FDE. And the idea here is that somehow or other, if I have an AI project, this will go better if they send a couple engineers that come and embed themselves into my environment. Much like, I guess they took this term from Palantir, and Palantir took the term from the military because they've always had this sense of forward deployment.

You had a forward deployed base, and now they're just co-opting this whole term. Liz, I'm going to start with you on this one, but at first glance, it seems to me this is something of an admission that says this AI stuff, it just ain't turnkey. Yeah, I'd say so.

I also think that the fact that, I don't know if they were coordinated, but I think that this is a preview of a shift in work. Right? There's like a consulting model.

What's going to happen when they're done? How is this going to change the way AI is run? I know they say that everything stays with the customer, right, after they leave.

But then there's stuff like how about cybersecurity? Like five or six engineers come in and train their whole staff and then leave, and then AI is sort of evolving in real time. I just feel like are they going to have the operational knowledge to be successful, or is this just a foot in the door to a new business model to have more consultants, since AI is automating other parts of their business?

I'm pretty cynical about it overall, and I think it's really interesting that the big AI companies are launching this pretty much simultaneously. Well, Yvette, to her point, you and I have been around this block a couple of times, but I seem to remember that back in the day, if I bought software from Siebel, Oracle, or Microsoft, or any of these folks, they would drop off people who were there to help me implement it. They were engineers.

They were sometimes called sales engineers, and basically, I always looked at them as hostages. I'm keeping these people until something actually works. So is this concept of forward deployed engineer anything new in your mind, or is it a new label for something we've been doing all along anyway?

You know what? It's a tell. I think it's when the four Largest competitors all reach for the same thing at once.

That's not a strategy. It's a memo that nobody said out loud because everybody got away, got it anyway. So here's the thing, when the model is now the cheapest part of AI, the most expensive part is everything that happens after you buy it.

So what did they actually build? And not a better model, they built an army of embedded engineers now who sit inside your company and build the thing for you. Which sounds like customer success, but is it?

Is it actually the exact same move that private equity made against founders and department stores made against local shops? Get inside the walls, learn their operation, and then like you said, Mike, never ever leave. So, they call it forward deployed engineering.

I call it what it is. Twenty years ago, Deloitte, PwC, and Accenture got paid to walk into your company and translate your company's operations into technology and vice versa. Now, the lab that sold you the model is sending its own people to do that job cheaper, faster, as the hook.

And using AWS's own language, they admit that that's what's really happening. " That's not a deliverable. That's a vendor writing down everything they learned about how you actually do your business and how your business breathes.

So I got some thoughts on this. So first of all, you know what they say about the NFL. The NFL is a copycat league.

Mm. Once someone wins a championship with a certain style defense or a certain style offense, by the next season, every team in the league is sporting that same style offense or defense. Tech has always been a copycat league, too.

Right? Yep. Someone succeeds, the next subject of a Harvard Business Review, everybody's doing it.

Palantir is a lot of companies' darlings. Right? And their whole forward deployed engineer strategy, which is kind of a takeoff of the military, and that's one of Palantir's biggest customers, so it's a mission that they understand in military terms.

But I give you, for instance, we announced it here at Futurum last week, right, Steven? The forward deployed analyst. The forward analyst, based a little bit on that, right?

Daniel Newman has written a pretty big manifesto on this. You could go check it out. " You know.

And they are sales engineers, and they are much like that Accenture, Deloitte kind of thing. But here's a lesson I learned also about 25 years ago when IBM bought ISS. Right?

At the time, that was the largest security... Liz, you remember this, I'm sure. I do.

This was the largest security buy ever. I think at the time it was maybe $600 million, unheard of amount of money for a security company. And when they looked at the books of ISS, and ISS kind of invented a lot of the security categories we play in today, I think it was 40%, maybe more, of their revenue was consulting, was services, was sales engineers being left behind, taken hostage.

And they didn't like to keep that on the books. They were private, they didn't have to, because that wasn't sort of the good kind of revenue that you get your multiple on. It wasn't reoccurring revenue, though we didn't talk reoccurring then.

It wasn't software revenue. It was body shop revenue. And no one paid a lot of money for body shops.

And nothing has changed. " And it'll be interesting to see what revenue multiples get associated with that kind of revenue versus something else, because it's not that reoccurring revenue model. Steve, you're shaking your head, Steven.

What do you think? Yeah. Well, I think that what you're hitting on there is the fact that whenever there's these trends, especially in tech, but in everywhere.

As Yvette was saying, certainly this trend came to accounting and consulting decades ago. And whenever there's these trends, there are people who do it right and there's people who do it wrong. And the difference often is the extent to which they're able to embrace and adapt to the new way of working.

And so I think that you're absolutely right that some of these forward companies, and in fact, the stories that we're linking here, I don't want to throw stones at any of these organizations, but I guess you got to ask yourself, are any of these organizations truly going to embrace the forward deployed software engineer model that has made Palantir so successful? Or are they really just, as you point out, body shops who are sending people to their clients at perhaps, as I heard there, too, reduced rates compared to an internal staff? And is that really going to achieve the goal here, or do we really need to do something different?

I think that if this is to be successful, it needs to be something different. But that being said, there's another question here, too, which is companies wouldn't be doing this if they didn't see a huge revenue opportunity. Clients then are going to be asked to pay, and will they pay?

Will they want to pay? Is this something that the customer wants, or is this something that the vendor wants? I can totally see why the vendor would want to have a forward deployed architect out there helping the customer to decide to build their entire system around my stuff.

That sounds great, but who's paying for that? Ultimately, the client is going to have to pay for that because they're the ones who are going to be trapped into that cycle, or who are going to be maybe deploying something that they wouldn't have deployed, or maybe just paying for it. Maybe that's the answer.

Maybe they just- Right ... pay for these people's time. But ultimately, are they getting value for their money?

And that comes back to the core question that we've been trying to answer in all of these consulting-type situations for decades, right? I think somebody finally noticed that, for every $1 of product, there's $7 of services, probably more so. Yeah.

It looked like there's more money over on that side of it, but Yvette, there's a thing called channel conflict. I'm sure you're familiar with the term. And, if AWS and Microsoft are essentially going direct now to the customer with their own army of consultants, there's a bunch of folks over at places like Accenture and some of your former companies that you worked for that are going to be going, "WTF?

No, you're not. " So how's this going to play? This has always been happening behind the scenes.

All these partnerships between Big Four consulting and the cloud platforms that are out there, they still have their professional services side in AWS that's competing with the Big Four's consulting. So this is not new. But what I am going to take this just to a little bit shift a little bit, because here's the one thing that nobody's talking about.

So if you build all your stuff on my platform, my model, right, and then you hire my forward deployed engineers to build it, who's going to audit it? You can't grade your own freaking homework. You know what I mean?

" You going to tell me the stuff that you built on your stuff is... How are you going to do that? So now this is the conundrum.

You can't grade your own homework. You can't do that. So when, okay, if someone says, "Yes, they have trust and safety teams," that Dario, his people already admitted that they have a hard time balancing making money and doing the right thing.

Mm-hmm. So how is that going to help you? So Google can't audit Google's cloud.

No. So why would any of these four labs be assumed to audit their own embedded deployment? How can that be any different?

So I want people to see that where's the free here? You're going to be paying for it twice. Workday is living it today.

They use an AI, right? " So how are you going to disassociate yourself from the frontier model, their forward deployed engineers, your thing going sideways, and who says who's going to pay the price? The client, the employer, whoever hired them is going to be paying the price.

The customer. The customer. The customer.

I do think- So here- I think that through before they think that this is a really good idea. So I call this the inevitability trap. Inevitability trap, right?

And you hit on it, Yvette. The problem is the models are being condensed into-- They're just flat, right? They're commodities.

Once the models are commoditized, and quite frankly, you don't care whether you're running on ChatGPT or Claude or Gemini or Meta or the DeepSeek. It's just a model. You've got to find where the value is.

You got to go upstack. And now we haven't quite figured out what upstack means here in AI. Does it mean purse a forward deployed engineer as a way of going upstack?

Does it mean that a forward deploy engineer is going to come in with a set of best practices and apps that sit on top of the model, right? Continuing to lock you into that stack, into that provider. Right.

But they're groping for it, and there's seven or eight companies, right? Because be clear, Google, AWS, Microsoft, they're no different than the model providers because you know what got put down before the model? That infrastructure of the cloud, which is sitting on yet another commoditized fiber network.

Right? So as these things get commoditized, everyone's always trying to go upstack and find what sits on top of the commodity. But the thing about AI is the commoditization is happening so quickly.

And I'm going to tell you right now, it's putting people services on top of the commodity has never been where the big money is. It's a solid business, but it's never where the big money is. And I think the big money hasn't quite-- Like what's going to be the killer app of AI, we haven't really seen yet.

Here's the one thing I want-- I wonder where- Liz, I think, Liz, you'll pick this up with me. It's like, so you have Big Four, they're fighting over the enterprise logos, right? Now you have Anthropic that now has deployed a Claude Core for the not-for-profit sector, for these forward deployed engineers.

So they're locking up everybody from enterprise, for-profit, to the not-for-profit. They spent like $150 million. They have 1,000 fellows in over 400 organizations that funded, trained, and placed by Anthropic themselves.

Like on paper, oh my gosh- bleeding hearts You're so selfless, right? Yeah, right. But maybe some of it is, but maybe it isn't.

But if you look how Wall Street gets paid tier of embedded engineers, and not-for-profits get the subsidized tier, different price, same architecture. A lab's own trained people sitting in your organization decide how their agents get built on your daily operations. One researcher actually said, they put it better than I could.

" And they can't be responsible for their own regulations or their own definition of what an altruistic mandate is. You can't do that. So I do think that while you have the forward-deployed engineers, are we really missing the dangers from enterprise paid tier to the subsidized for the not-for-profit?

Right. Liz, you go. Yeah, but you know what?

I feel like- From management anyway. Yeah, let Liz go. All right.

So Alan, you were saying that the consulting isn't the big money business here. But if it's not the big money business, and these companies are extremely wealthy and well-funded, and they're also launching the same services at the same time, then what is the win for them? And to me, I think it's the institutional knowledge.

Maybe they want to take over. Who really wins, right? So the customer will have their forward deployed team for however long.

Maybe they keep them longer, maybe they make them jump through a couple of hoops, they learn some stuff. But who leaves with that IP? It is AWS and Microsoft.

And so as technology becomes more commoditized- No, it's not that it's not money though, Liz, it's not multiple. Right? When you go to borrow or raise money or sell stock around it, the Street doesn't value that kind of revenue as much as other revenue.

I would- That's fair, but I think what my point is that I think that if the AI companies are gaining all this institutional knowledge, that's going to give them even more of a lock than they already do, right? I think that it's their engineers. It's their engineers, right?

It's their- Right ... forward deployed team who will be leaving at some point. It's- So who's winning?

They are It's more than just the lock. If I have that access and I can see how your ops work, what's to prevent me from launching a company or a service that now competes- Exactly ... with you eventually.

Exactly. So it's not just getting locked into a stack, it's ultimately who do you trust? Yeah, and that's really an extension to the current concern I think a lot of businesses have about trusting foundational AI models with their business.

It's funny, one of the people who most eloquently, and I guess vociferously exclaimed that people can't trust third parties with foundational models with their businesses because those model providers are going to steal their business, is none other than Palantir on CNBC just the other day- Mm ... " And at the same time, he is literally the leader in embedding his engineers into his clients' businesses. That looks like the same thing to me.

There's an old saying about how do you say trust you in Yiddish? There's a joke. Mm-hmm.

But I can't say it here, we're not- Yeah ... it's family friendly. But we've got to jump, Mike, to our third thing.

We do. We ran this one a little over, I'm sorry. Which is we're coming into one of Steven's favorite topics, which is how we're going to process computing in space.

And now there's a report talking about how a startup plans to embed GPUs in a telescope so that we can process images at the point of the capture and then ship them back down to, I guess I'm assuming us here, or maybe some base on the Moon someday. I don't know where this is going to wind up. But the point is that the processing is occurring somehow in space.

And Steven, we've talked about this in the past, but it's challenging to put together anything that looks like processing in space. So how do you think this is going to work, and will it work? This is going to work because they're just talking about putting a little thingy in space.

That's it. So let me set some context here. So this story is a very small component of a very big thing.

The very, very big thing is the Artemis mission to the Moon, which is NASA's mission to return humans to the Moon, building out all sorts of capabilities that we don't have. It's very, very aggressive. It's actually very exciting to nerds like me who love space and want to see us go back to the Moon.

So it's pretty cool. Artemis involves the SLS rocket, it involves the traditional legacy rocket and spacecraft makers. It involves the new ones like SpaceX, but it also involves startups.

Now, you may not have heard of Firefly Aerospace, but they're another one of these startups like SpaceX that's doing things. But of course, with SpaceX, and their Falcon, taking the bulk of the revenue out of space, Firefly had to rapidly pivot and failed, and pivoted and failed. And it's a really exciting story, but now they actually have a pretty cool business here, which is what they're calling this Elytra concept, which is essentially that they're going to have a spacecraft, not a rocket, but a spacecraft that gets launched by somebody else, somebody that rhymes with, I don't know, Place X.

And that will get up there into space, and then they can use that spacecraft to do stuff, useful stuff like moving payloads from low Earth to high Earth orbit, or maybe moving payloads to the Moon, or maybe from the Moon's orbit to the lunar surface, et cetera. So this Elytra, which is part of their Blue Ghost mission, I know I'm just dropping names here. It's all part of this whole Artemis process.

Now, what we're talking about here ... " And as you can imagine, it's pretty important that it knows where to drop it and that it drops it in the right place. It's also pretty important that we have really good maps of the Moon, because we're going to be doing a lot more stuff there.

So the idea is, instead of having all the data from that telescope go back to Earth, we're going to have a small NVIDIA-based, Jetson-based processing system that can do some image processing in space. So in answer to your question, is this going to work? Yeah.

This is a little component of a big mission. It's not even an important one, because if it doesn't work, they'll just send all the data back to Earth and process it here. But it's probably going to work because it's a small scale.

The Jetson is literally designed for portable and compact use cases. It's not going to generate a lot of heat. It's not going to need a lot of power or cooling.

And you can put it in a lead box to shield it. So I don't see any reason why this won't work, and it's going to be cool because it's going to be the first AI thingy in space. But it ain't an AI data center, that's for sure.

No. So clarify that point a little bit. Do you think that we will see data centers in space, or is that a goal too far and a long way from sticking an embedded processor in a telescope?

Hell no, never work. It is not possible, not practical, not logical to put a data center in space. Space is not cold.

Space is a vacuum. And trying to get rid of all that heat is really immensely impractical. Even the things that SpaceX is talking about are going to require heat radiation devices like the size of multiple football fields for just a single rack unit-sized system.

That just doesn't make any sense. Yes- Steven, yet if you look at the SpaceX financials, short of them building data centers in space, I don't know how they make that business worth the $3 trillion or whatever they were talking about. Well, they make it worth $3 trillion by convincing a lot of suckers that they know what they're doing.

Yep. There's a sucker born every minute. This ain't- No, I agree with you ...

going to work. It's going to cost more to dissipate the heat than it does to do the computation. And they can- And we still haven't discussed latency.

Yeah, exactly. Then there's communications latency. There's bandwidth problems.

There's shielding and radiation. Yeah. You cannot send modern systems up there with all of the radiation and expect them to continue to work for very long.

You have to shield them. That adds more weight, more mass, makes it even harder to cool it, makes it even harder to launch it. I just feel like this is simply just not going to work.

But sending an NVIDIA Jetson board on a satellite, yeah, that's going to work. No problem. Yvonne, I think he's saying I shouldn't become a forward-deployed engineer for AI data centers in space.

Yeah. No. Keep your day job.

" It's a great movie. It is. That is a good movie.

But the whole point of "Hail Mary" is that the shielding goes away, and when the shielding goes away, they all die. Yep. Okay.

Yes. Sorry, spoilers. Steven, question for you, question for Steven.

So wait. So there's one thing that you said is going to work, but there's one little thing I want to poke at. So the one thing that we didn't mention is that it's the first fully autonomous AI decision-maker on record, operating 240,000 miles away from a human.

Right? There's no human in the loop in this. Well, it's not actually doing anything all that interesting, to be honest.

All it's doing is image processing, and we're pretty good at image processing. We've figured out how to do that a lot. And it's not like it's actually going to, on its own, decide where to drop the payload.

No, it's just going to do some image processing, and I'm okay with that. Okay. So there's no chance of hallucinations or anything like that?

Oh, I suppose there is. There's always a chance. But this isn't running LLMs either.

This is actually running basically just massive-scale neural network image processors. And so I suppose it could hallucinate seeing aliens, but the worst thing that could happen is it comes back and says, "Yo, there's an alien there," and we go, "Nope," and that's it. It's not like it's going to crash the spaceship into the planet.

No, but it might mismatch something. Like say there's a crater here when there's not, or... Yeah, but I bet we're going to check that before we land anything there.

Mm-hmm. I think- Yeah, this is like level one autonomous driving, not level four. Yeah.

Okay. So I was in London while Waymo was doing its training. " No, I'm with Liz on this one.

For me, this is very problematic. We have something that's making a decision 240,000 miles away from a human, because it's going to take a few... There has to always be a human in the loop.

Mike knows this about me. I say this all the time. I don't care if they're processing widgets or pictures, somebody's got to look at it.

Well, hopefully there's an astronaut on the Moon that's checking this out somewhere, or that somebody's- Yeah ... remotely managing all this stuff, and that's where maybe your human in the loop is going to be. And otherwise, we just send up a bunch of robots to the Moon and call it even.

We don't need a person to go. But people want to go to the moon. Oh, want to go to the moon.

Not because it's easy, because it's hard. Uh-huh. As someone said.

Anyway, I guess we'll leave this one here. Steven, I love to see how excited you get about space, though. Yeah, you completely geeked it on out with this.

I get more excited about stupidity. Well, no, but you know what? He's from Ohio.

John Glenn country, right? Yep. Good for you.

Yeah, we got the John Glenn Research Center for NASA right here, and once a year, they invite people to come visit it, and I actually got to go, and it was so cool. I got to stand inside their wind tunnel. It was awesome.

I think I've told you this. I've taken the kids here to Cape Canaveral, the Kennedy Space Center. I'm going to be excited about remote control moon buggy racing in the moon.

Yeah. Maybe Aaron will just kind of have a little contest there. Not as exciting as you on the moon doing remote control moon buggy racing.

But hey, data center's on the moon. Anyway, guys, we're out of time. Mike, thanks for taking the helm today.

I appreciate it, as I was late. No worries. Steven, Yvette, Liz, thank you so much.

You know what? We were struggling here a little bit- Of course ... but I think we got a solid Monday.

Solid Monday crew on the gang. I hope to see this continue. Thank you out there for watching.

tv, on the Techstrong TV YouTube channel, or the Techstrong TV OTT app, which is available on any kind of screen you like to watch this kind of stuff on. We also have Techstrong TV content playing throughout the day. Hey, Steven, when's the next Tech Field Day?

Not for a while. Summer off. Yeah, we got the summer off.

I figured I would let them go. It looks like we're going to be at Black Hat. It looks like we're going to be doing Cloud Field Day out in September.

Black Hat's in August, so not for a little while. Excellent. I will be at Black Hat as well.

Yeah. And do, if you're interested in that Forward stuff, go check out Daniel Newman's LinkedIn post on Futurum Forward. It's very interesting.

Yeah, Futurum's got a press release about it, too. Yep. Go check that out.

But for now, on behalf of the gang, thanks for watching. We're out.