Episode 4: New Hope, Trillion-Dollar Funny Money, and Why Data Centers Do Not Belong in Space

Zane: We can say five if you want. I'll skip a four. I only got three on the site. This is a good intro. I think it's episode four. We've done, yeah, yeah. What what could fool you is if you go to our YouTube page, for some reason episode one's on there twice. I don't know how I did that, but I left it. And they have different titles. Different titles.

Greg: Shoot, that that math adds up. No, you're a hundred percent right. It has been three. That's what got me. I was fooled by the man. They do. They do. Yeah. And they had different podcast names because I'm a dope and don't do due diligence before going live. Because that's how I roll, baby.

Zane: So I'm really hoping we get double hit on that. Yeah. And what's hilarious, it's when we had our old name. So the one with two names.

Greg: It actually was already somebody else's name. But you know what? That's how the cookie crumbles. What are you gonna do? And in the fascinating world of AI, you gotta change this.

Zane: Yeah. I think when it was, when I found out Joni Mitchell had that podcast, I was like, maybe that's just not our demo.

Greg: Probably could have took her to court, let's be honest. We probably could have won that.

Zane: Yeah, she didn't have any more fight in her. She done. She done it all. Hot take. Hot take.

Greg: Exactly. Hot take Joni Mitchell, thumbs down. OpenAI, thumbs up. Hot take of the week. There's your YouTube short right there.

Zane: You had your time, man. Find out Joni Mitchell's had an AI podcast this whole time.

Greg: Yeah, she actually backed deep seek the whole time. So, episode four. We earned it. New Hope.

Zane: As I named it here, New Hope. I mean, that is episode four in Star Wars. Star Wars. I did do it real. That's insane. Which by the way, I like to remind my son, it's just episode one. It was only New Hope because of the other three episodes. But you know, whatever.

Greg: True. Different time, different era, different generation, as it turns out. But we got a lot to cover, man. Here's the thing about doing AI topics, and yet we don't do weekly takes. You get a lot of collateral you can cover between episodes, right?

Zane: That's the thing. 'Cause every week I have some new stuff we can cover and then we don't cover it and now we have to combine it all.

Greg: And life happens, you know. So where do we start? I think we start with a lot of hot IPOs coming out here recently. And even more pending IPOs dropping, like valuations of AI companies. If you look at the scale of some AI companies, we'll just call the name is eight hundred and fifty two billion dollar valuation. Really a company that nobody voted for. Is it company or infrastructure at this point?

Zane: I'm pretty voted for it. Do we vote on companies now? I guess.

Greg: I'm just saying it's a product that would just become a brand, right? So we talked about this earlier offline, it becomes almost a cult of personality sort of thing where your company or product name becomes something. Hey, if I needed a bandage, you'd probably say, man, I cut myself, let me go get a band-aid. You'd say a band-aid, right? So when your brand gets synonymized with things. You remember when search engines started becoming a thing? Like, hey, I'm gonna go Yahoo this thing, or I'm gonna go ask Jeeves this. Like, no, everyone says hey I Googled that.

Zane: I think Ask Jeeves happened for a bit. I remember like Ask Jeeves, but it never caught on. How much of it was the butler? How much of it was the branding of the butler versus the actual service?

Greg: So cut that to the last two years as chat based AI has become a phenomena in our culture. Globally, you say, hey, I ChatGPT'd this. That's a status of you having impact on culture. Does that mean all of a sudden I have a company? Or am I just a product? What else does OpenAI do?

Zane: Or my process or a verb really.

Greg: Right. If I look at SpaceX, they have Starlink, they have other sources of revenue. Anthropic has... does OpenAI have other sources? These billion dollar valuations, is this evidence of a bubble, or is this just the evidence of phenomena and aspirations versus an actual company? We're not financial advisors. Please do not take financial advice from us.

Zane: They do have a different take, right? It's all gonna blow up. Stop investing in this. But I think where you're going down is OpenAI, ChatGPT, started as an OpenAI company, and now here we are with this large valuation. Is it a tool? Is it a paradigm shift? Whatever the internet did to us. Who are the ones that are gonna hold the keys? Who's gonna make the money?

Greg: So it sounds like you're morally conflicted, and I am an open and honest hypocrite across every single aspect of this. But you did mention GPS. Shout out to TomTom. How's that like a relic of the past now? Is that where we're going? Guys, you remember when we used to do AI on our phones? Is that gonna be a conversation that our children and grandchildren have?

Zane: I just looked at dispatch today with cowork and cowork dispatch was like you can use your phone to connect to your home PC. Is that the future or is it my home PC is my data center?

Greg: So you're building your own DCs. For my projects. So I have an app that I'm writing that's gonna manage my projects for me. It's gonna be called Could This Meeting Have Been An Email? As we all know, some people in project management love to just have meetings for the sake of having meetings. But is it a bubble? Is it a bust? Time will tell. There's a lot of money being moved around.

Zane: So I tend to think the B2B avenues are the ones that'll be the future. We're coming off this social media, this whatever economy where it was all about consumer, but at the end of the day, it was really about information and selling information and selling ads. So it really was B2B behind the scenes. So I think any tool that isn't business to business aren't going to be the big moneymakers.

Greg: The stuff that's impacting our industry right now with Ramageddon that's going on, folks that have a larger data estate have had to slow down. They've gotten the multi million dollar bills from OpenAI of tokenization. And there's this fascinating new metric that every CFO is gonna be talking about, which is instead of replacing humans, you're replacing tasks. How many tokens it costs per task. Tokens per task.

Zane: Where's the compute? Who owns the infrastructure underneath? This much work goes in, what did it cost to get that work out? And does it make sense?

Greg: So this comes into task analysis. If I have a certain task that is very complex for an AI to do and it costs thousands in tokens if I keep doing it every single day, but it saves me five minutes, is that really valuable to the business? Probably not. In the words of Bill Belichick, do your damn job.

Zane: Or maybe look at the task and see where you can optimize it.

Greg: One of the interesting facts I found out recently, Google pays Apple roughly twenty billion dollars annually to remain the default search engine. Twenty with a B. So when you hear that Apple is paying Google a billion dollars to get Gemini by way of Siri, doesn't really sound that impactful in the larger scheme of things.

Zane: Okay, so that's the part I missed. They've already been promised 20 billion.

Greg: It's like I pay fifty thousand dollars in taxes every year and guess what? You're getting a twenty dollar tax rebate, my friend. Pretty sweet, huh?

Zane: If you're myopic and you focus on the one thing, you're like, I'm ahead.

Greg: I think it's gonna empower their B2B business. What is the number one competitor in handheld devices to the iPhone? It's Android. So what are we doing?

Zane: Are they admitting defeat? Apple knows how to do hardware. It's a hardware guy. Cook's out, it's the hardware guy.

Greg: I don't think they've lost the hardware. Globally, Apple is still the vast minority in the hardware market. So maybe we'll just take a page of the Chinese here, go to the old thousand year plan. We'll get their LMs, power their LMs, and you just kind of bleed them dry.

Zane: Are they just feeding each other in Silicon Valley? Money's going back and forth and it keeps the machine running.

Greg: This weird triangle trade of NVIDIA to Oracle to OpenAI. It's just funny money. But speaking of those folks, this is a hot topic, especially in my own backyard. This episode brought to you by Bang, American Berry. This is our fourth of July episode. AI data centers. Ever since the announcement of Stargate and the rush to the trillion plus dollars being fed into these AI data centers, not all the math is adding up. Stargate, the big investment down in Texas, now the largest data center in the US.

Zane: There's that proposed one in Utah.

Greg: Is this like they're doing the thing that Walmart used to do where they'd open another store that was one square foot larger just to one up each other like a bunch of pedantic children?

Zane: I'm a big baseball fan and the Padres just sold, it was the biggest deal ever in MLB. Every time there's a sale, it's going to be the largest because the years move forward. That's how time works.

Greg: There are some being proposed, some already approved in a very stealthy way to basically do nothing but eat up our natural resources and push up my power bill. The offset is well, what about the stimulus to the economy? We're giving you a bunch of tax breaks. Could you not pay your fair share if you guys are just handing off 20 billion dollars off the hip to be the default search engine?

Zane: They're enticed to come to an area. And there are, because of the way they can vote in a region based on hectares of land that I own, I have more voting power. So now an out of state entity can have more of a say of what we do with this land and suck the resources out. That's easy for me to say 'cause they're not building a data center here because there's too many trees. I'm in Seattle now.

Greg: I'll give the counterpoint. I saw an interesting turn of events in Nevada that Google actually has a zero carbon footprint data center. Actual, not buying carbon credits. They're using geothermal, pulling power from the earth, going 8,000 feet down. Is that sustainable across 500 data centers? Probably not. I think it's gonna be more of a con than a pro to the consumer. And my son does not like AI. The generation after us is not a huge fan of it because of not availability of jobs, threat of economic pressure.

Zane: Same. My daughter's artistically bent, she's like, it steals this, it steals that. Does the future care about AI?

Greg: And it's not just folks in the US. The EU also has a different stance, more along the lines of governance and protecting people's data and privacy. Is this overregulation holding back progress, or is this the pragmatic approach that is needed?

Zane: Technology moves at such a rapid pace now that it is okay to slow it down to be like, how does it impact and how do we work it into our lives? They're all sovereign countries and they should have a say in how data moves outside of their environment. If we just let it go at the speed of progress, what is it, four or five people that get to dictate how AI is gonna be used in the world?

Greg: The EU approach is looking at it from the human perspective out. That being said, why Mythos was banned was not out of a moral obligation. It was out of, we turned Skynet loose and my god, it came in with a flamethrower because our most classified applications were not just saying, hey, you have vulnerabilities, by the way, I'm also gonna exploit them too. You can't put the genie back in the bottle. Mythos has been enabled for certain mission critical systems, if you're protecting national security or core infrastructure. But that's not stopping the overarching problem, which is we think we know exactly what's going on, applying things that have not been tested, for the sake of innovation, or is it through fear? Well, if we don't have this, somebody else is. We're still in Cold War era thinking.

Zane: It's an arms race. How do I get to the moon first just to show you I can?

Greg: Look at China's power infrastructure. Look at what China's doing with compute.

Zane: They're embracing alternate energy sources. They pioneer, they industrialize technologies, EVs and all those sort of things.

Greg: There was a massive electronic vehicle expo in Shanghai and Tesla didn't even show up because their technology is embarrassing compared to what was available at the Chinese companies.

Zane: They're cheaper, they're more efficient, they're self-driving. I'm afraid they're gonna do that on quantum.

Greg: Let's talk about the AI and quantum regulation that was passed recently. We've established a standard for post-quantum cryptography. Agencies have to have a plan in place because PQC is mandatory in like three and a half years. Quantum cryptography is a very serious thing. If AES-256 breaks, it's gonna be a bad day for everybody.

Zane: A AES two fifty six with quantum turns into one twenty eight. That's today. So you still got pretty okay protection. But compare it to the speed of innovation right now, how long is one twenty eight gonna stand?

Greg: Quantum is only going to get better. We have to get very serious about quantum computing because it is not a matter of if, it's when. One of the largest supercomputers on the planet, Frontier, at Oak Ridge National Labs, had a task where it took approximately one million years to complete. Quantum was like, I can do it after this coffee break. It's a different game. When cryptography breaks, Bitcoin breaks, anything in flight breaks. Financial institutions are worried about data being harvested now, decrypted later.

Zane: If I steal your data now and just sit on it, I'll have the tools later to break that thing. You have no control over how I have this data.

Greg: It's like stealing a safe. I don't got the combination yet, but I'll get it eventually. But what's really gonna save us from quantum computing, and I really believe this, is data centers in space. If you wanna keep the algorithms honest, do pigs in space and then quantum computing.

Zane: Data centers in space. Interesting concept. Papa Elon. Are they the leaders?

Greg: They wrote the white paper. It is definitely a white paper. A lot of assumptions. Fifteen gigawatt capacity. I believe the radiator alone required to cool that amount of energy would be about the size of Munich, Germany.

Zane: So the radiator is Munich. Got it.

Greg: You're a physics major, right? What are the three ways you can transfer energy? Convection, conduction, and radiation. Radiation is the only way and it's the worst way to do it. Because there's nothing in space. It's a vacuum.

Zane: If you just float in space, you actually wouldn't freeze because there is no transfer. This is all movie magic stuff.

Greg: And what are we gonna do with all that heat? So if I'm facing the sun at all times, which way are my radiators facing? The Earth and the ionosphere. What could go wrong? To quote Mark Cuban, is OpenAI a company or a product? They get over a trillion dollars for AI data centers, and they're gonna burn that. There's no way to make that money back. So let's apply that to data centers in space. It's gonna be the government.

Zane: You gotta launch it up there, build it, support it. Let's go back to Elon Musk's early companies. The government propped them up. Tesla's not Tesla without government money. SpaceX is not SpaceX without government money.

Greg: Do you realize the complexity just to repair the International Space Station, which is a lot smaller than the city of Munich? Billions of dollars to maintain. GPU's gone bad, power supply's gone bad, somebody has to go to the data center and replace it.

Zane: You get a four hour response on those?

Greg: Is someone just gonna live in space? We've got a parts depot on Jupiter. Don't worry about it.

Zane: That's why we're going to the moon, Greg. There's a lot of rare earth minerals on the moon. We're building the components there. That's probably not true at all.

Greg: So we're gonna go a week further than the data centers to get the parts. I don't think we're gonna meet our SLA. What's really gonna save us? It doesn't make sense, man. It doesn't bring a lot of strength and comfort of fear of, is AI a bubble or not.

Zane: So the real answer is like how do you make AI more efficient, so it uses less resources?

Greg: What's really weird is we've talked about all these topics and I've not got a single conspiracy theory. Is it the moon the conspiracy theory? What is the Zane conspiracy theory of the week?

Zane: So that's it. We're gonna put the Parts Depot in a studio in Hollywood that's landing on the moon to make you think that we're delivering it to the space data center.

Greg: Solving the world's problems one episode at a time. Where do we go from here? From a utopian perspective, we want to think it's gonna modify and enhance the human experience. I could use these superclusters to understand the human genome to prevent disease. Or I can have some really cool cat videos where they come in playing a ukulele.

Zane: We wanna see evidence of the AI immediately. We don't wanna see it as a benefit over time.

Greg: And maybe just one day they can prove that the earth is not flat for some people out there.

Zane: I am fascinated by the flat earth conspiracy theory. I like conspiracy theory because I like to see logic break down. It's just my jam. I like to see what's the seed, how did this come to it? The flat Earth amazes me because it's so bad. It's like watching bad TV.

Greg: The conviction and passion is of a level of zealotry that is unparalleled. Have you ever sat on a beach? How can I still see the ship when it's all the way out there? But can you see it forever?

Zane: And then the explanations for it are like crazy. That's my favorite conspiracy theory. Take that energy and put it into everything we just talked about and maybe you could solve some of this stuff.

Greg: So let's land this plane. Number one, there's a lot of possible good coming down the pike from AI innovations. There's a lot of perceived bad for our economic impacts. There's some hype on the valuation of it. And most importantly, don't let anything we say here make you think these guys need innovation, they're curmudgeony, shaking fists at cloud. No, we're actually being pragmatists because we're looking at it from the human approach and not just from the cool sizzle things. Do your own research, and actually do your own research, don't just get into someone's ChatGPT and say, well, ChatGPT agreed with me. If you talk long enough, you can convince ChatGPT that the earth is flat. LLMs aren't supposed to understand knowledge.

Zane: I just read an article about, so Anthropic uses some constitutional LM. In the middle of long chats, it'll tell people to go to bed. It has this sense of well being for a human. I did some website stuff the other day and it said, hey, is this a natural place to pause, Zane? It's telling me to go touch grass. So I felt a little positivity there. That's that human element trying to get through.

Greg: Maybe it's not about the technology. Maybe it's about the friends you made along the way.

Zane: Definitely is. It's the journey.

Greg: Well, on that bombshell, two things we can promise you next time. Number one, episode five, and number two, the results of a man in his forties drinking a Bang energy drink at nine forty-five at night. Cause I am wired for sound and we're just getting going. All right, people. We'll see you next time.