AI Infrastructure and the Indispensability Trap | The Open Current Ep 5

Hey everyone, welcome to The Open Current with Alan Shimel and Margaret Dawson. Thank you for joining us today. Margaret, how are you?

I'm great. I was in San Francisco this week and meeting with a bunch of startups that are focusing on secure AI and role-based governance and keeping agents a little bit in control. So, it was a great week, and San Francisco's always beautiful.

It is, but for that kind of agenda, I'm surprised you didn't fly to Tel Aviv. They flew to us. Oh, they flew to you.

I knew they somewhere they had to be in Israel. Actually, you know what? I did have a takeaway from that week that I need to follow up on because they were all coming out of these amazing, you know how the Israeli military- 8200.

The 8200, and there was another one. There was like three divisions. Yeah.

One that was above the Mossad, above the, I'm going to say the name wrong, so I'm not going to even try to say it, but let's just say, and I've learned this back when I was doing network security. My whole engineering team was in Israel, in Tel Aviv and Haifa, and the training, both from a computer science and now increasingly security, but security even then is just, I don't think there's any other government institution that trains people and that has a startup ecosystem that when you leave the military, you immediately go into this amazing secondary training ground, and then it empowers people to do startups. It's actually kind of brilliant.

Well, it's a startup nation, right? I have a lot of experience there. I first became involved in the Israeli cyber startup, Yoav Landsdorf, who started something called YL Ventures.

Yoav pioneered just what you're talking about, but he added another twist to it, is after their seed round, when they got A round, he would bring the CEO, sales, and marketing teams from Israel to either Boston, New York or San Francisco, and then keep R&D and engineering in Israel. And very successful. Yeah, it's a smart model, and I think I've seen a lot of companies do that.

So they're all ready, I mean, a lot of these CEOs were already building out offices. New York is a common one. San Francisco is a common one.

Yeah. I don't think I've ever been in Boston. One was Florida.

I don't know why. I think it's just because they had people there. But- Well, the weather's not that- Might move to Florida.

Uh-huh. I don't know why either, but I get it. But let me just also say one other thing.

As someone who's been in cybersecurity for 25, 28 years, whatever. Yes, the Israeli cybersecurity training and state of expertise there is high, right up there, world leaders. Yep.

Don't underestimate for a second what the US has at the NSA. Oh, I would not. I think the difference is it's such a-- Because it's a smaller community, I think the way that you experience it is just different, right?

It's not- Yeah ... quite as spread out. Well, and they don't have the commercial follow-up.

Right. NSA, you're at the NSA forever. Though I have good friends, Bruce Schneier.

Bruce Schneier, who was NSA. Ron Gula from Tenable was NSA. I've met several other people during the course of my career that are ex-NSA folks who- Yeah ...

came into the commercial cyber world. Israel, though, Margaret, I think sometimes there are people parked in black Mercedes outside the door where they get their exit papers, and as they come out, they're writing checks or something. I mean, these guys all said, like, when they do their seed rounds, they get funded pretty fast.

So it's- Yeah, it's crazy ... crazy. It's a crazy thing.

No, it's definitely very supportive. Okay, we're going to switch gears, though. We can talk about cybersecurity all day.

All right, go ahead. What are we doing this week? So you always start with asking me a bunch of questions, and I'm going to flip the table on you today because you have- Okay ...

some exciting news, and I know- Yes ... you're incredibly humble and would never talk about this yourself. No.

Never. But you have a book that is, at the time this is released, it will-- Oh, gee, okay. Yes, you're so humble you actually got a copy of it and brought it to the table.

They call it a prop. Look, I happen to have one laying close by. I have nothing to show.

Okay. No, but I do. Yes.

I have a book coming out. " That's a very hard- Yeah ... I have to really think about that first word to say it correctly.

Yes. It took me a long time to spell right. No, I think it is one of those words that you can spell.

So let's just start. I mean, we'll talk about the name because I am curious about the name. But my first question is, it's not like you have a lot of time.

I feel like every time we talk, you're doing a million different things. Today, you were just talking about you've launched two things, and you were running over here, you travel all the time. So my first question is, when did you have time to write a book?

Because knowing you, I'm going to guess that you wrote it in AI didn't, but maybe that should be my first question. But mostly, why write a book? Why now?

And how- Yeah ... did you get it done? Because I think a lot of people talk about, I mean, I have a book I've been trying to write for five years.

I can't get it done. So I'm inspired. That's a great question.

Thank you. So I have threatened to write a book for 20 years. Wow, okay.

Actually, Mitch Ashley, you know Mitch. Mm-hmm. Mitch and I have been business partners and working together 25 years.

We've talked about writing a book together for 25 years. And there are some times where Mitch is pushing me, and some times where I'm pushing him, but we never got it together. Mm.

And people have approached us, and I just threatened and threatened. I never did it. I would never have done this book but for AI.

Right? Yeah. And I wrote this book, let me be clear.

I wrote the book. But I even put it in the acknowledgments. I thanked Claude and ChatGPT and Perplexity because the amount of research that went into, especially the historical perspective- Right ...

of the book, I would've been sitting in a library for months looking stuff up, right? And with AI, I was able to just pull-- Like, a lot of the historical perspective was stuff that I heard urban folklore about. Right.

Yep. But I had to get the facts. And by the way, everything in the book is fact locked, fact checked, fact sourced.

No, I find that the models now will be very good at giving you sources. You often have to ask for them to provide the sources. Yes.

But I do find that it's getting better and better. No, it is. com.

Yep. And if you go there, you could look up every claim in the book, find its source, its proof. I have that website open right now- Where I got it from ...

as a matter of fact. Yeah. No, and so AI incredibly helped that part of it.

AI did come up with the concept for the cover. So- I told it what I was thinking ... so AI was the inspiration, and that took you over the edge to actually get this book written.

And so, but what about AI kind of gave you that idea, that inspiration? Like, people can kind of see the premise overall and kind of looking at these, and you're almost looking at the Industrial Revolution or those different revolutions at different times, and AI being- Build ups ... one of these.

But what was it about those different kind of industrial innovations that really captured your imagination and got this started? Sure. Well, the big one for me personally was the dot com, the fiber optic boom.

I've got T-shirts and scars to prove. I was part of that. I started my first company during that, early on in the dot com, and I sold it to another company, and I helped them go public.

And we were an early ASP, application service provider. Yep. I did the whole thing.

We raised about $500 million. Wow. We spent about 600 million.

Right. That's pretty good, actually. Yeah.

That's lower than most. That was good for those days. I was going to say, yeah.

I remember sitting in the exec room, in the exec meetings where we had our monthly exec meeting, and the CFO, he'd get up there, and he'd say, "Great news. " And everybody would clap like we did a great job. Crazy.

After the dot-com bust, I realized that's not anything to clap about, losing 60 million a month. But anyway, I saw what happened in fiber optics, right? Right.

We had dark fiber. It took us 20 years. Yeah.

When I had known about how AT&T, and again, I lived, you did too, Margaret, through the AT&T breakup, right? Yeah. And what happened there.

That wasn't the first time the government- That's right ... decapped AT&T. " And Silicon Valley's here as a result of that government- So before you hide the lead there a little bit, so what you're talking about is not just, there's a million things you could be talking about with each of these kind of innovations or revolutions or moments in history that start to look very similar, but you're specifically looking at kind of layers of the innovation, the infrastructure versus things on top.

Right. Where does the money go? Talk about that a little bit, because that's fascinating.

So you're bringing together the data center build-out to the fiber optic, or just the pipe, the fiber. It wasn't even fiber at that point. I'm trying to remember what we called it then, like the initial pipe- It was copper ...

copper, that was what it was. It was pipes. To railroads, to things that we don't think about anymore.

So these are all grids, if you think about it, right? Mm-hmm. They're all grids.

So look, we're committing eight trillion, let me say that number again, $8 trillion to data center build-outs in the next three to four years. Eight trillion dollars. Let me just give you- Did you compare the valuation of these build-outs, like in that day's money?

Even apples to apples, like now to then, it dwarfs anything we've done before. Yeah. But just to put this in perspective, within 25 years of the railroad system being built in the US, 40% of the railroads went bankrupt because they built them just to invest the money in the railroad.

They didn't think about the business of transportation. Oh, interesting. The electric build-out, the grid, huge, very successful.

But the real money was not the Edison companies that were the utilities, because they were capped by the government. Right. It was GE who made all the appliances that everyone plugged into the wall.

GE was the company of the 1900s, right? Mm-hmm. Of that whole century.

Yeah, it was. Yeah. AT&T, I mentioned the transistor.

They had to give up because they wanted to keep the long distance business. The fiber optic line, we laid two trillion in 19, whatever it was, 2000, 1999 money. Two trillion dollars in fiber optics.

Some great names that you remember. MCI, WorldCom, Global Crossing, Level 3. Oh, Level 3.

Wow, I haven't heard that name in a long time. Remember them? Yeah.

They're all gone. Yeah. But what got built on each of those things that I mentioned, on the bones of them, were built the next revolution, upstack.

The value moved upstack. I think that's what's going to happen in AI. Interesting.

So, my first thought is what feels different is a lot of the companies doing that build-out also have the applications on top of that build-out. So do you think there's a learning there, or do you think we'll still see the same thing? So maybe we can start with the first example.

Before you answer that, let's go to where it starts, which is in beautiful Louisiana, Richland Parish, Louisiana, at Meta's Hyperion site. Which, just a side note- Yeah ... I'm on my third historical fiction book about Louisiana.

I don't know. I get on these kicks where I read everything about a topic, and currently it's Louisiana. There's fascinating history.

You guys should all, everyone should read it. There you go. The kingfish.

Huey Long and the kingfish. It's because it's the French connection. You know where this came from?

Yes. Yeah. A guy that was an Uber driver for me was from Louisiana, and he told me to go read some history because the French influence is totally different than the parts of America that had the British influence.

Anyway, we can talk about that another time. They don't have common law. They have civil law, which is French.

You're going to go into detail even though I said we shouldn't talk about that now. No, I'm not going into detail. Okay.

All right. You know I can't help myself, Margaret. I know.

I know. That's why I always should never do an aside about Louisiana, but I'll- No, you know what I have to ... put my recommendations later.

Okay. You open in Richland Parish, Louisiana. Yep.

Meta's massive data center build-out. Why did you start there? Yep.

I'm assuming it goes to what we just talked about, infrastructure. It's infrastructure, but it's also a snapshot of, this is weird. This is a weird infrastructure build-out.

Okay. So Richland Parish, one of the poorest parts of Louisiana, where cotton was king 100 years ago, and nothing's happened since. Okay.

Meta comes in, and they make the governor of Louisiana sign an NDA that he's not allowed to talk about their deal. And they make all the politicians, both at the state and local level, sign NDAs. They're not allowed to talk.

Was this the one that didn't get announced, and it just started building, and no one knew about it? Well, there's a lot of them in the process, unfortunately. Yeah, there are.

That's what I was going to... Amazon had one too, I think. Yeah.

But the thing about this one, so they got unbelievable tax breaks. They're not going to pay a lot of taxes for many years. They were given a bunch of land.

They bought the land. Let's be fair. They're investing about $200 billion in this project.

$200 billion in Richland Parish, Louisiana- That's a lot of money ... is a lot of frigging money. Yeah.

You know how many jobs? Got 200 permanent jobs after the thing is built. And this is literally GPU AI data center.

How many jobs? Yeah. 1,000.

1,000? I wish. Oh.

I think it's under 100. They guaranteed 35. 100?

They guaranteed 35, because these data centers are automated, right? There's not a lot- Sure. It's all robotics.

Yeah. And so- What about jobs to build it, though? What about construction jobs?

Well, there is that. Okay. You bring in your HVAC workers, your electricians- Right ...

your concretes, your steamfitters, and that will create some temporary jobs, and then they go on to the next thing. And then there's also, hey, we've got to bring power to this. So this Hyperion facility is going to consume as much electricity a day as Manhattan does.

What? Yeah. It's crazy.

So they- What's the eco promise? Was there some kind of deal? We don't know.

There was NDAs. I actually sent freedom of information requests to the state and county level. So it's not a county, it's a parish.

That's the Louisiana French thing. Okay. To the parish and state level, asking them, was there an environmental impact study- Yeah ...

or something like that? NDA. NDA came.

I didn't know Freedom of Information Act could be blocked by an NDA. That's interesting. Well, it was never- It must expire at some point.

I don't know, but when I was writing this book, I couldn't get any information on it. Okay, so why does this build-out kind of represent kind of a canonical version of- We've seen the same thing in state after state, county after county. But I've got to be fair and balanced.

Since I wrote the book, I continue to research and follow this story. The teachers in Richland Parish, Louisiana, over the summer, right after I'd finished writing, recently got bonuses up to $50,000 a teacher from the property tax that Meta did pay. Wow.

Now, 50 grand in Richland Parish, Louisiana, is a lot of money for a person. And so- I mean, that's a lot for a teacher almost anywhere. That's awesome.

Yeah. Well, that's a whole, we can talk about that another time. We could.

We don't pay our teachers enough. Yeah. So there is money flowing.

There is some money flowing in there. And let me bottom line the whole data center thing for you. I think it's a decision between the people who want to build it and the people who have to live with it.

Right. And if the people who have to live with it agree to do it, more power to them. Build the data center.

I'm not- I'm against data centers popping up overnight when the local population hasn't had a chance to weigh in and make their decision. And that's what originally happened in Richland. It's happened in Maryland, in Virginia.

We just, here in Palm Beach County, where I live, which happens to be the same county Mar-a-Lago is in, the county commissioners just passed a moratorium. No data centers- Wow ... even though there were plans to build one out in the Everglades.

So I'm not here to simplify it and say data centers are bad, this is good. I'm just saying people deserve to have input into what goes up in their neighborhood. This is the pragmatism.

This goes back to everything we've talked about with capitalism and technology and even AI. I think those of us in technology tend to be at the front end of the optimism of the, "Let's look at what we actually need," but there is also a pragmatism that has to be applied to it. So it's not pro or con, it's like, how do we make this work not just only for the sake of innovation, but not in spite of what- Right ...

the humans want, in spite of what the environmental impact is. There can be a balance to that. Yeah.

And you can say that about almost anything, I guess. So talk more about just the term indispensability trap. It doesn't immediately tell me what you're going to be talking about or what the dichotomy or issue is that you're trying to bring out.

So how would you define that for someone simply? Sure. So the indispensability trap thesis is the more indispensable something becomes...

Well, Jevons paradox says the more indispensable it becomes, the more it's used, right? It's like- It's like a utility. Right.

Right. And it gets cheaper and pushed out. Right.

The trap is the money that gets invested in that doesn't give you the return. Mm. The return gets trapped.

The return gets capped. That becomes the foundation in which value is built. But the value is built on top of it.

Right. So it's GE making the appliances that gets- Yep ... the value.

It's the computers built on top of the copper wires that has the value. It's AWS and Google and Microsoft that's built on the bones of Level 3 and- Right ... MCI's fiber that got the value.

Right. And we're seeing it here again. I'm not here to knock companies, but look at Oracle.

Larry Ellison has bet the farm on AI data centers. He's leveraged Oracle to, some would say, an unhealthy level. He's fired people, not replacing them with AI to pay for AI.

Mm-hmm. And that's a bet that according to what my book says in "The Indispensability Trap," he's in the trap. Hmm.

He's not going to get that kind of return. Although he is building his own models. He has cloud, he has applications.

Oracle has a portfolio of things that you could say are that layer on top of it. So is he just building out his own infrastructure to then take advantage of that for the things above, or you think it's still too much below? Oracle's AI factory build-out has one customer that's responsible for 60% or 65% of what's on the books, and that's OpenAI.

OpenAI- Mm ... has promised over a trillion dollars to various players that they're going to use their data center of output. Mm-hmm.

And if that don't work, something happens, whatever, we've seen this movie. We have. Right?

I think this is an interesting point, is there are still very few players that are accounting for the vast majority of what will leverage that build-out. I'm trying to think of how to say this very carefully. And if I even go back to companies that I've worked with, there is a huge amount of revenue coming from Anthropic, OpenAI, SpaceX.

You name where tens of millions of dollars for sometimes small companies where that just doubles, triples, quadruples their revenue base overnight. But you're then falling into a different trap, which I see from a lot of companies, where your revenue, the long tail versus the short tail of your revenue is dangerous. Right?

Yes, it is. And for a startup, you don't ever want to say no to that, but you very quickly have to diversify that revenue base, or overnight you could lose well over 50% or 80% of your revenue. And so it's the same mathematical problem that we're having there.

Yeah. And again, we saw this in the dot-com, right? Yeah.

The circular math, where I give you, you give me. Right. No cash really exchanges hands.

It's just on our balance sheets. It's the same thing we've seen here. Well, and so talk about that, because one of the things that really fundamentally drove me bonkers during the dot-com build-out was that valuations based on fundamentals.

And anyone that studies business, economics, finance, you're taught how a company's valuation is created. And it's based on, we don't need to go through the math of that, but it tends to be based on your revenue, your debt, your margins, foundational financial metrics. Sound economics.

Right, sound economics. And what we saw in dot-com is none of that. It was based on how much money they raised.

It was based on a lot of different false valuations, in my opinion. We are back in that trap, and I'm wondering, is there any of that economic trap that feeds into the indispensability trap that you start to look at? It is.

It's another aspect of the trap. OpenAI just reported, I don't know if they reported because they're private, but word was out this week on the street that OpenAI is now on a 70 billion, not million, $70 billion revenue run rate, and is still hemorrhaging money. One has to ask, how much do they have to make- Yeah ...

to be profitable? That's wild revenue, $70 billion in revenue. And not profit.

Not even. And I'm not saying break even. Not even break even.

I'm talking about hemorrhaging money. Yeah. So this is the trap right here.

We are pouring money into this grid, into this new frontier. We don't know what the killer app for AI is, Margaret. Yet we're in tech, so you probably know more about AI than your non-tech friends because you play with it- Yeah, sure ...

more than they do, perhaps. But can you tell me, Margaret, where's the money? What's going to be the killer app here?

I'll go back to where we started the conversation. Right now, where I'm seeing the money go is in security and governance because everybody's getting scared. That may not be the killer app, but it's definitely where some new value is going.

But I think that's more- Sure ... rational value. Again, these are companies that are building strong business cases, and they're doing pilots, and they're getting market product fit, and all those things that you're supposed to be doing to build that- Yeah ...

foundational organization and a strong business. And the valuations are, what I would say, feel more normal than the valuations that are in this other layer. So yeah, it'll be interesting to see.

In spite of everything- I mean, cybersecurity ... it's still early days, right? Yes and no.

It's early days, but this year alone... So really in the book, I track eight or nine companies. The usual suspects, you could probably name them without even looking.

They have invested a trillion, one trillion with a T now. Mm-hmm. A trillion dollars this year into this AI build-out.

And companies, the most successful companies in the history of the earth, Amazon, Google, Microsoft, Oracle- They actually are showing negative cash flow for the first time in I don't know how long. Wow. In terms of free cash flow.

They are investing more in this build-out than their free cash flow affords, and they have the greatest free cash flow we've ever seen. Yeah. So that should give you- I wasn't aware of that.

That's interesting. I'm going to go do some research on it. They all went negative this last quarter.

Interesting. How could Google and Microsoft not have cash cows? It's interesting because as a public company, that used to be just an absolute do not even attempt to go there.

That wouldn't have even been an option. So the fact that even shareholders are saying, "We're now willing to bet money on that," there's a huge downstream impact that you're talking, upstream, downstream impact, on this. So before we run out of time, I want to flip it a little bit.

We talk about open source all the time on this podcast, and in chapter 10, you start to talk about open source and you've seen this movie before. What did you learn about the role of open source or the open source wave of this in AI models? Sure.

So open source is another way that the trap gets sprung. Mm. Because open source is the equalizer.

What China has done here with OpenWeight- Mm-hmm ... which is similar to open source. We'll get to that question in a minute.

Yeah. Yeah. But what they've done is brilliant.

They couldn't compete with OpenAI, Anthropic, or Grok, or Google. Instead, and whether you buy into they distilled it from the American or not, they came out with their own models and used open source as distribution. Something you know all about at SUSE.

Mm-hmm. Something we're all familiar with. They used open source as distribution, and it served two purposes.

Number one, it got it into a lot of people's hands and their users. Right. And people are realizing, hey, this works pretty good.

It may not be quite as good as that one, but it's free. You got to remember, the one thing they have is a homegrown market of tens of millions of people. That's all the effect that they get to- Yes, they do ...

right. You know who the biggest users of the Chinese OpenWeight models are? Americans.

Silicon Valley, because they know how to save a dime when they can. And so they're happy. All of these Silicon Valley companies are running Qwen, or Qwen, whatever it's called, and the Moonshot, and all of these things.

Great. But then the second part of the open source one-two punch kicks in. Okay.

These people who are using it now begin to innovate with it and improve it. And so instead of these Chinese companies having to spend the kinds of money that Anthropic and OpenAI does to improve their model, to train the next model, they've got all these open source customers who are helping them with the next models. So the gap, and we had an AI gap in the world, right, that gap is closed now where these OpenWeight models are actually not very far behind the best American models.

Well, I also think it's become so easy for people to start building. You look at all the frontier models, and then you look at the small language models that are being created for very customized private use. And the way I think about that is it makes sense that maybe there'll still be a handful of LLMs that are kind of the big beasts that we're seeing today, but I think increasingly enterprises are going to decide to have more control over the models and create their own small language models that are more appropriate for a very specific situation, or vertical, or use case, or workflow, or whatever, fill in the gap.

So I don't think that area has fully transitioned, but I think this whole thing with OpenWeight is starting to move us towards that as people realize, oh, there's a different way to do it. Nvidia thought so. They paid, what was it, $13 billion for HuggingFace.

Yeah. And HuggingFace is the place where people go to get those smaller models. That's right.

And there's like a million of them, literally. There's so many different models out there. Yeah.

Well, and people are building new ones every day. All right. Yeah.

So let's kind of bring this back to the beginning. So first of all, this is available October 6th. Yes.

Amazon, or wherever you buy your books. Amazon, Barnes & Noble, wherever you buy books. I'm downloading it on Kindle, so yeah.

Mm-hmm. I get a Kindle version. How would you kind of wrap it up?

What will people walk away with? What's the takeaway? What's going to be the aha moment from this?

So the takeaway, it depends who you are. Okay. If you're an investor- Mm ...

I don't know if OpenAI or any of the companies that are pursuing the $8 trillion build-out, I don't know if that's a good investment right now. Wow. Short-term, long-term.

That's a lot. The big real estate guys are in here, BlackRock, and those folks. Do you need to do a disclaimer or something about all these companies?

Is there some deep takeaway from that? Yeah, I should mention it. So look, I am the CEO of Techstrong.

I cover all these companies. I write about them. I talk about them.

I didn't use any proprietary data that I wasn't allowed to have access to in the book. But I also, if you can't tell, I'm passionate about it. How did I write this book?

I'm passionate about this. I would stay up all night working on a chapter. I love this.

Everybody loves to talk about these periods in history that start to repeat themselves, but I think this is a very different take on it, using this kind of filter of the indispensability trap. So I think- Yeah ... it's great.

It's not just the technology. We tend to always look at it like, oh, technology and what it did for humanity at the time, and all these things. It's a very different perspective on it.

So this is aimed more if you're an investor, a business leader. You want to know, look... And we're hearing such stories, AI's going to make us extinct.

Right. Because now we're in the anti-AI kind of phase thing here. Right.

But again, I'm bullish on AI. I just think that the real wealth, the real value, isn't sitting in Richland Parish, Louisiana, unfortunately. My heart goes out to those people.

I don't have it with me, my phone, but I think AI's going to live on our phone. AI's going to live on our keychain, like this Muse thing that Meta just came out with. Yeah.

Or on our laptops. We're not going to need that data center because we're going to continue to shrink it down and make it more efficient. Interesting.

So it's going to follow what we've seen with other technologies where it goes to the edge, it becomes more embedded, which is not the way we've thought about it. So this is great. Well, I look forward to reading the full thing because I've read a bunch of excerpts and- Thank you ...

so I'm looking forward to going through that. And I feel like you're right- Margaret, I've got a confession to make. What?

This is the first time I've been interviewed about my book, and so you did it. Yay. Thank you.

I think I should interview you more often. Between the two of us, first of all, I was the former journalist, so I feel like we should be flipping the tables more often. But- No, you know what?

We'll put it into the cycle. We'll put it in rotation. Okay, great.

Perfect. Absolutely. This is wonderful.

Thank you so much. Thank you. You want to do your closing?

All right, hey. Yeah. Do your thing.

Yes. You've just listened to "The Open Current" with Margaret Dawson and Alan Alan Shimel. You can listen to this, I don't know where you're listening to it now, but you can get it on Apple Podcasts, on Spotify, on Techstrong TV, our OTT app, or our YouTube channel, Techstrong TV YouTube channel.

I usually post it on LinkedIn and other social medias as well. We do this every two weeks. We'll be back with more.

We didn't even mention, Margaret- What? you guys announced SUSECON, didn't you? We did, but it's okay.

We don't have to make it a commercial every time. All right. I wanted to say- I disclaimer that.

We're here to talk about technology trends, but- Absolutely, but I'm coming to Barcelona, and I'm really happy about it. All right. Next one, we'll talk about what we're doing around that.

No worries. Okay. Fair enough.

All right. Thank you. I will see you soon, my friend.

Thank you. We're out.