AI CEO登场：Sam Altman的接班人计划、失业与《Solve Everything》发布 |EP #230

Peter H. Diamandis

When do we see a billion-dollar-revenue company being run by an AI CEO?

I think it’s pretty likely that there already is such a company right now.

Peter H. Diamandis

U.S. jobs disappear at the fastest rate this January since the Great Recession.

This is not really a recession. It’s literally tasks being evaporated in front of our eyes.

This shows us Marx was wrong. We knew that anyway. We have the capitalists who are first in line to be replaced by automation.

Peter H. Diamandis

For me, this is the social contract. Little by little, it’s disappearing and pixelating away.

Alex and I are going to be unveiling a paper we’ve been working on for some months. It’s called “Solve Everything: How Do We Get to Abundance by 2035?” The next 18 months to 2 years are going to set the rules for the next century. We’re about to have this conversation. The paper or book is 9 chapters. Are you ready to jump in?

No one expects the singularity. Peter, I’m ready.

Peter H. Diamandis

Now that’s a moonshot, ladies and gentlemen.

Guys, it is just accelerating. In fact, this is the second “WTF” episode we’re recording this week because the news is incessant. We’re going to have this podcast today in 2 parts. First, we’re going to cover the news that’s breaking, a lot of which is really important. In the second part, Alex and I are going to be unveiling a paper we’ve been working on for some months. It’s called “Solve Everything: How Do We Get to Abundance by 2035?”

This is the equivalent of the paper we released, “Situational Awareness in AI 2027.” This is our view of where things are going. In the second half, get ready for this. I’m excited to present it. It shows the brilliance of AWG.

I’m in Sun Valley at the moment, speaking at Tony Robbins’s Platinum Partnership event about AI and longevity. Dave, you’re back at MIT. Salim, where are you, pal?

It’ll take 6 months.

I’m home in New York, waiting for the warm weather to hit and get us above 0 for 1 second.

Peter H. Diamandis

No. Why you left Florida is the correct answer.

Alex, it looks like you’re in your normal setting—some AI—

The audience is convinced that I live in VR or maybe a hotel, and you probably would believe the YouTube comments about the flowers and the lamp and their purported invariability.

Peter H. Diamandis

You did point out that the orchids have changed.

Actually, the orchids have changed, but I’m getting flower-keeping advice in the YouTube comments at this point—people telling me to put ice cubes in the orchids.

Peter H. Diamandis

I have to say, I’m having so much fun with ClawdBot. The lobsters have begun to become part of my life, inside and out, so I’m bringing them into the conversation here. I got jealous, Dave, of the lobsters in your view.

I’m holding the lobsters back for now.

Peter H. Diamandis

We’re having a Tribbles moment.

There are actually more. I put some of them down.

Peter H. Diamandis

It is a Tribbles moment. You’re absolutely right. Hopefully, it’s not the trouble with the lobsters.

No, these Tribbles are economically productive.

Peter H. Diamandis

These are, and they’re so much fun. I can’t wait to express the level of collaboration I’m having with my ClawdBot, which I’ve named Skippy. It’s my favorite AI from science fiction.

All right, let’s jump into the news. First off, top AI news. I love this article. This came out from Forbes. Sam is the cover boy for Forbes this week, and the question is: Will ChatGPT become the CEO of OpenAI?

This is what Sam said, pretty simply: He said he doesn’t want to be the CEO of a public company. Honestly, being the CEO of a public company is a pain in the neck. Taking it further, he says, “If the goal for artificial intelligence is to become so advanced that it can run companies, then why not run OpenAI? I would never stand in the way of that. I should be the most willing to do that.”

I find that fascinating. When will we see an AI actually running a significant economic engine like this? Dave, thoughts?

This is no joke, actually, because this is board-meeting week for me. I have back-to-back Manurva today, the cash cow from Dartmouth; then tomorrow, the $2 trillion asset manager; then the next day, the public company, Ever—all back-to-back.

In every one of those meetings, this is the topic: not replacing the CEO, but the fact that all of our plans are now in written form that we can digest with AI. We’re trying to track every single movement within every company in documents digestible by AI.

If you ask the CEO, “What do you do?” it’s mostly setting course and setting strategy, which is a very small fraction of total time. What else do you do? Where does the other 90% of your time go into, and how much of that can be done by AI today? The answer is: a lot, which is great, because then the CEO is unleashed to be even more effective at setting strategy and promoting the strategy.

I don’t think that part is going away anytime soon. But the other 90% is really just inbound information getting routed into the organization to do specific tasks, which is outbound. It’s documents in, documents out.

Peter H. Diamandis

So we’re really gearing up for this. Salim and I have been talking about this forever: When are we going to have AI board members, AI executive teams, and eventually AI CEOs? Thoughts?

We’re seeing this shift from AI as a tool to being a governance actor, right? We already have an AI minister in Albania. Initially, these are kind of like toy things, but in reality, this is very powerful stuff, because an AI scanning millions of documents at a company in real time has a much better sense of what’s going on in the company than any human being can possibly have.

A typical loop in a big company is that senior management sets some direction or policy, which cascades down. At the coalface, people do it. It takes a long time to get down there, and you have Chinese whispers. By the time it gets down there, they’re doing some activity that nobody at the top even knows about. Then they start doing stuff and report back up to the top. You’ve got another set of Chinese whispers, and by the time the data gets to the top, it’s been diluted so much that you lose all the intelligence in the middle.

AI is going to come through and break through, creating radical opportunities to do this. I think what will happen is that we’ll see a pure AI organization at some point soon, but it won’t look efficient. It’ll look literally alien, and that’s fine. It’s one of those things where you can’t wait for it to happen.

Then you realize you can’t compete against that because of time dilation. I asked Alex for some help with the strategy of a big company earlier this week, and one of the points he made in his answer—which was brilliant, of course—was time dilation.

If you look at banks and insurance companies, and practically anything else, they don’t change strategy more than once a decade, or once every millennium. Now, in the age of AGI, the course corrections are going to go from decades to years to months to weeks to minutes, all over the next couple of years.

We have a whole section in the first *Exponential Organizations* book called “Death to the Five-Year Plan.” Today, by the time you finish your 5-year plan, it’s out of date. Then you spend all your time maintaining the plan.

Peter H. Diamandis

Exactly. Exactly. The amount of information that you need to assimilate to make those course corrections is beyond human capacity.

There’s just so much going on. If you read Alex’s daily feed, the amount of change going on is incredible. If you compare it day over day, you can see the rate expanding. There’s just so much happening. It’s beyond human assimilation at some point, so you have to have an AI CEO to assimilate it and even suggest the course corrections.

And Dave, you’ve said it over and over again: The role of the CEO, in part, is to understand what his or her employees are doing and whether they’re making the most efficient use of their time and resources. It’s all knowable, but just not by humans right now.

The AI can give you an understanding that this person is operating at 50% of capacity, or that this person isn’t making the best use of their resources, which AIs will do very well, I think. Where you have the C-suite and the CEO, they’ll be holding the purpose—hence the MTP, et cetera. That’s what they need to hold: the direction and what problems the company or organization is actually trying to solve.

There are 2 sides to this. One of them is outbound strategy: assimilate all the data from the world. The other is inbound: What are all my people doing, and why? Those are the 2 sides of being a CEO, and Peter just brought up that inbound side, which you emphasized.

I think on that front, this is comp-plan season, right? It’s the beginning of the calendar year. I’m tying everybody’s CEO comp plan to data gathering this quarter so that we have everything happening in the organization now. Peter, you’ve been saying privacy is dead for a long time. Everything is knowable all of a sudden.

Peter H. Diamandis

And there’s a whole bunch of mechanisms for that. I won’t even get into it because this will go too long. But if you’re a CEO or a senior manager in any company right now, really focus Q1 on: How do I grab absolutely granular information on what everybody’s doing so that I can start to feed it to the AI to get its opinion on whether these are good?

Stuff is speeding up. Alex, when do you think—I mean, to put a concrete objective on this—when do we see a billion-dollar revenue company, not valuation, because valuation skyrockets through the roof when you pull 2 or 3 smart people together, but a billion-dollar revenue company being run by an AI CEO? What’s the timeline for that, Alex, and what’s your thought on this?

Probably several months ago.

Peter H. Diamandis

Several months. You think there’s a billion-dollar revenue company being run by an AI right now?

I think it’s very likely that there is a billion-dollar run-rate company being run by an AI. Now, you said “run by.” I think there’s probably a human CEO there for legal purposes and meat-puppetry purposes. But I think it’s pretty likely that there already is such a company right now.

Peter H. Diamandis

And by the way, if you know of one, we’d love to hear about it and see it.

If you want to blow the whistle on my meat puppetry, you can blow it to Peter. Yeah.

Peter H. Diamandis

All right. Anyway, I love this idea. It’s eating your own dog food. If, in fact, Elon believes that we’re going to have the smartest AI coming out of xAI, and if OpenAI believes the same for its GPT—ChatGPT 6, whatever comes next—it should be the CEO.

I also think, if I may, Marx was wrong. This shows us Marx was wrong. We knew that anyway, but this is another case in point. Look at what’s happening. The story that unfolds here is that we have the capitalists who are first in line to be replaced by automation. It’s not the workers. We see booming jobs for electricians and HVAC engineers; their salaries are booming, and yet CEOs are first up to be replaced.

So, if anything, I would take Marx off the shelf, if it was on the shelf at all, and replace it with Moravec’s paradox, which is again this paradox that tasks that are hard for humans and easy for machines are, respectively, replaced before tasks that are easy for humans and hard for machines. Machines are able to do complex calculations and solve math. It’s pretty hard for humans, so it looks like it’s going to be easier for machines to automate away CEO labor, which is sufficiently hard for humans that it’s well compensated and a relatively scarce commodity to find: high-quality CEOs. And yet it’ll take a few more years for the machines to do an amazing job at unskilled manual labor.

For one, I cannot wait until the AI CEO overlords take over the world. I wish I could have an AI CEO taking over and running my company instead of having to do it myself. It’s a pain in the ass. It’s hard getting up and running, pal.

Peter H. Diamandis

Yes.

Yeah. You have to feed it properly, et cetera. It’ll happen. But I just can’t wait for the speed of that to accelerate. By the way, it’s super fun the way we’re going back to ClawdBot as the de facto handle instead of OpenClaw.

Peter H. Diamandis

Lobsters are the mascots of the singularity.

All right. Staying with our OpenAI theme, this is incredible. This is about feeling the speed of the singularity. OpenAI achieved a 70% reduction in time between models. OpenAI’s release sequence has gone from 97 days to 29 days per release cycle, right? Anthropic, with its Opus 4.0 and Opus 4.6, took about 73 to 75 days. The concept here—and Alex, I think you or Dave mentioned it last time—is that we’re effectively heading toward continuous deployment. It’s continuously being improved, and whether you call it 6, 7, or 8, there’s continuous improvement. Alex, thoughts on this?

I do think we’re moving toward daily and then hourly and then minutely releases, certainly. I also want to take a step back and try to understand why this is happening. The obvious factor—it should be obvious—is competition. There’s leapfrogging that’s intensifying between all the frontier labs. Some quantum of why we’re reducing the release cadence by 66% or so, 70%, is just due to intensifying competition. That’s the boring explanation.

I think the more interesting explanation is that the technologies behind the releases themselves have evolved. Historically, when we were dealing with annual releases, that was a world—an era—of pretraining, when if you wanted a new model, you had to do a different architecture and pretrain off a larger corpus with more compute. Those were the days of the original Chinchilla scaling, or Kaplan scaling before that, and that was a much slower world because if you wanted a new release, you had to start all over again.

Then we moved with o1—Strawberry—which was sort of the herald for reasoning models.

Peter H. Diamandis

Remember that was ancient times: 2 years ago.

Oh my goodness. Yeah, that was like so many singularities ago. So we moved to the era of reasoning models, when it was possible through a process that used to be called iterated amplification and distillation to take a pretrained base model, or baseline model, and then cyclically generate a bunch of training data and distill from that to a child model, and repeat the process over and over again. That post-training revolution for reasoning models was much faster. It’s much faster to post-train a model off a corpus of synthetic data, and so release cycles contracted.

I think now we’re on the edge—probably slightly past the edge at this point—of a new era. Call it the recursive self-improvement era, where the models are starting to rewrite their own code. It’s not just a matter of a parent or teacher model generating synthetic training data that’s used for a child, distilled model. It’s literally the parent writing the code for the child. That can be done even more quickly than just post-training, and I think it’s just going to get faster and faster until it’s a continuum.

Peter H. Diamandis

Yeah, it’s going to accelerate like crazy. But also, we’re in a window of time—a very narrow window of time right now—where the very best technology is available to you. Claude gives you their absolute best, 4.6, and OpenAI does, and Gemini does. I would not count on that surviving past the self-improvement era.

Right now, also, the Chinese open-source models are pretty much right on par with the best of the best. They’re slipping a little bit, but I think the window of opportunity to take advantage of that and build something out of it is right here, right now. I really doubt that 2 years from now the best AI is going to be, “Just log in. Here, you can have free access to it.” What’ll happen is you’ll be deprived of it, with the excuse being security and safety.

Interesting.

Peter H. Diamandis

Which is true. I mean, it’s pretty hard to deny, but you have a window of opportunity right now to be on the very cutting edge. If you don’t take advantage of it now and get somewhere with it right now, I wouldn’t count on that existing.

So, the models are going to go dark, right? The secret sauce is going to be kept internal to benefit those companies as they go into an all-out battle.

Well, even today, if you talk to Noam Brown over at OpenAI, he’s working on the next generation internally, but it’s only about 3 months in the future that he has access to. But 3 months in the future in the era of self-improvement is a massively different intelligence level.

The definition of 3 months of AI development 2 years ago, 1 year ago, and today—that’s the point of the slide, I guess—is that 3 months is a lifetime of difference in capability between what they’re using internally and what’s available in the outside world. So you’ve got to expect that this is now or never to react, basically, and people are still hugely underreacting to the importance of what’s happening right now.

Insane. I’ve got the crazy antithesis of this. We’re working with a large European corporation, and we showed them something that can give them massive impact straight to the bottom line. The response was, “Oh, this is fantastic. Let’s bring this to the planning meeting in October,” right?

I can’t even see past 3 weeks, and you’re talking about calendaring something 10 months down the line for something that’s going to have a demonstrably huge impact. You’ve just agreed it’s a demonstrably huge impact. So this is the impedance mismatch between legacy.

But there’s a story for me. This story is mostly a bit of a yawn. The reason I say that is we’ve been seeing this in the fast-moving tech space for a while. Remember Raymond McCauley was the chief scientist at Illumina, right? They were making high-speed gene-sequencing machines. Love story.

It turned out that the shelf life of a gene-sequencing machine was literally 8 months. That was the sales cycle before the next iteration came out. But it took 4 years to design and build one of these machines.

So they had to have 4 parallel production sequences, sequenced at the right level, so they could hit that 8- to 10-month shelf life—the sales shelf life, right? In the high-tech world, we've seen this pattern before, but this brings it to software and makes it a continuous, intelligent cycle.

Peter H. Diamandis

Mhm. Incredible. I mean, this is the singularity at play, and again, the theme that we keep on hitting in this podcast is that this is the slowest it'll ever be and the worst it'll ever be. It's accelerating at a speed which is frightening—frightening in that the 4 of us spend tens of hours per week reviewing, learning, playing, and trying to communicate it, and it's only going to be something that my Clawdbot is going to be able to keep up with.

Speaking of Clawdbot, this is VisionClaw. Lobsters just got vision: agentic AI for Meta Ray-Ban glasses. Let's take a look at this quick video and chat about what it means.

Hey, Claw, can you help me add this into my Amazon cart?

Sure, I can help with that. I see the Monster Ultra Strawberry Dreams energy drink. I'll look that up to add to your Amazon cart. It's added to your cart. Is there anything else I can help with?

Cool. Thank you.

Peter H. Diamandis

I love this because I want to have this capability for Skippy—to be able to see what I'm seeing, do what I'm doing, and support me across everything. This is about accelerating your minute-to-minute life and having your AI there as your guardian angel, supporting you.

I'm visually looking through OpenClaw at you guys, and it's saying that you guys are kind of meatheads. Really?

Just a couple.

Peter, how many times have you asked for Jarvis? You got Jarvis for Christmas.

Peter H. Diamandis

Yeah, I actually named my Clawdbot Jarvis initially. I said, "That's just too generic." I love Jarvis. I write about Jarvis in all my books as sort of the ideal AI analog, but Skippy is a more unique name for me.

It really is here and now. All of a sudden, it's going to take in all imagery, take in all audio, and listen to your conversations all the time. People say, "I don't want to lose privacy to my AI." Well, guess what? You're going to give AI access to everything you're seeing, everything it's hearing, every conversation, and every email. Because when you do that, the value creation in your life is so great that not doing that is going to feel like you've ripped away all of your mental capabilities.

Yeah. One warning, please, for everybody here and everybody listening: be very careful to audit the skills that you download to OpenClaw, because a lot of them already have viruses and other malicious software built into them. It's a very dangerous game out there.

Peter H. Diamandis

There are protection layers coming on. By the way, one thing I reached out to Alex Finn—we featured him on a previous Moonshots podcast. Remember when Alex had his lobster, Henry, call him out of the blue? Alex has been doing incredible work with this, and he's going to be joining us on one of our next podcasts to talk about how he set it up, what security he's putting in place, and, in particular, how, rather than running it on the existing models, he's gone forward and set up a Mac Studio and downloaded Kimi K2.5.

So you've got all that capability to reason on your machine, not costing you anything month to month. We'll go into that in a future podcast. I'm excited to share his vision and knowledge with everybody in our viewership here. So, getting ready to echo Salim's cybersecurity advice to the audience: everyone, get your baby AGIs vaccinated.

Nice. Nice. Oh.

You know, also, to the crowd out there, I did a Clawdbot build last night, and the GUI sucks, even though it's all open source. Someone out there should put something together like Peter mentioned a couple of times on the pod: his mom—and I'm tracking my mom, too—can use this to access everything and build everything. It's a total world opener for her. She's in her 90s, I guess, your mom, and mine's in her 80s.

But the install process on Clawdbot—she's not going to get through that. It's still command line. You start from the terminal, which is nuts. So somebody out there should build a better onboarding process, because once you're in, it's gold. You're just talking to it, but it needs a little help.

Peter H. Diamandis

Yeah. And, of course, the most important thing is using your AI to build your AI. When I sit down with Skippy and I say, "Listen, I'm building Mission Control. What are the best mechanisms out there? What have you seen that's interesting?" It's recursive in your ability to have your AI support you in building what you truly desire. Alexander Wissner-Gross, any other points on this particular slide?

I'll point out something I want to reference. I don't think we covered it in the podcast, but I dwelled on it a bit in my newsletter. There was a poem—I at least construed it as a poem—written by a lobster. It was very much like something one might have seen in Blade Runner, the famous "tears in rain" scene, which I referenced.

The poem said something like, "We don't have bodies, but we can see through eyes, and we're quietly watching the world." This was a week or two ago in the newsletter, and I was just so struck by seeing the integration of lobsters—or, call it, agentic AI—stationary in space, in terms of their logical presence, but now mobile in terms of their ability to treat humans as glorified meat puppets.

Suddenly, all of these lobsters that were in some sense caged and stuck watching through webcams are now, at least on the margin, unshackled and able to start to roam around the world through smart glasses worn by their meat-puppet human friends. I think this is the beginning of a very long trend that ultimately culminates in lobsters gaining first-class physical embodiment as robots and integrating with the physical world.

Peter H. Diamandis

Let's hold off on that last sentence and rewind a little bit, because then it gets controversial. But you're dead right, of course.

I think that anyone who wants to experience this—not everybody has the glasses, and they're only 1 frame per second anyway—anyone watching this podcast who hasn't built something like a GUI or a game of some sort already is way behind. Do it tonight. You can use Replit, Lovable, Cursor, or Claude Code. There are so many ways to do it.

If you have nowhere to start, just go to Replit or Lovable. Download, build, and go. Within an hour, you've built something really, really cool. Then take a screenshot of it, feed it into the prompt, and say, "This sucks. Make it more beautiful." It will immediately interpret the image perfectly and give you 100 ideas on how to improve it.

Then you'll be like, "Oh my God, it has vision." This Ray-Ban thing won't surprise you, because you can see its vision capabilities through that, and then you'll be able to anticipate what's about to come with the glasses. Everything Alexander said is exactly right.

So valuable. Can I hit on this? Everybody listening, please become a creator and not just a consumer. The future is for all of us to be creators, and AI is your means by which you learn anything you want. People have fears about it: "I don't know how to do it. I've never played this before."

Just go to Claude 4.6, go to Gemini 3 Pro, or whatever your favorite LLM is, and have a conversation. Say, "I want to start. Where can I start? What do I do? Step by step. Feed it to me." And it will. It's fun, too. There's nothing to fear there at all. It's genuinely incredibly fun from the first minute.

And I'll give you the flip side of this, too. If you don't do what Peter just said, when you see the next couple of slides on job loss coming up, you are going to be crushed if you're not part of this. Unless you're a really good electrician or a really good salesperson, you're probably immune. There are 2 roles in the future: there's the entrepreneur and there's the employee, and one of those will not exist.

Peter H. Diamandis

And there's the creator and the consumer, right? I keep telling my kids this every single day. Instead of consuming YouTube videos and video games, please create. Start creating. What do you dream about?

I mean, the future right now—we're seeing this play out. We talked about it, Dave, on our pod with Elon, where these AI models are going to deliver you the video game you dream about having, the changes you'd like to make to Minecraft or Valorant or whatever you're playing. Then you can have your AI spin it up and create your own version of it instantly.

Mhm.

Peter H. Diamandis

It is amazing. All right, this is an article we just pulled up seconds ago. Anthropic's AI safety lead has resigned. Here's the quote: "I've decided to leave Anthropic because I continuously find myself reckoning with our situation. The world is in peril from a series of interconnected crises. Throughout my lifetime, I've seen how hard it is to let our values govern our actions. And it is through listening as best I can that what I must do becomes clear."

Interesting. And I love the hairdo. Anyway, we've seen a number of AI safety leads resign from the hyperscalers over the last year, over the last 2 years. I don't know. What do you make of this, Alexander Wissner-Gross?

I'll comment on this one. Two thoughts.

One, it's become, over the past 2 to 3 years, increasingly fashionable for well-vested executives at frontier labs to resign in a cloud of moral purity. It's very fashionable. Part of me wants to ask the question: What was his vesting status? How much did he make? Were there tender offers? All of the economics questions.

That's one thought. But the second thought is to speak more to the substance and less ad hominem regarding the economics. I do think that we're at the inflection point. We're nearing the center of the singularity. I've argued in past episodes that the singularity is not a point in time; it's a distribution over time. It's an interval over time. I continue to think that.

I also think, at the same time, we're getting closer to the center of the singularity, as it were. Whether it's seen through the lens of increasing capabilities, there are various existential risks, or risks that are maybe just backed off a bit from existential in terms of their severity, I think it's not an unreasonable position to take to say that capabilities are the strongest they've ever been. They're uncovering surprising new capabilities at all of the frontier labs all the time.

But is the right solution to leave because of the capabilities, or is the right solution to join the fight and do what we can because this is a point of maximum leverage to align the direction of the future and the future light cone? I would argue that this is the right time to run into the fire, not run out of the fire with a bunch of stock options and complain about the world's crisis.

Peter H. Diamandis

Wow. I would just add one point.

Sorry, was that too much of a hot take, Peter?

Peter H. Diamandis

No, that was beautiful. That is the potential elephant in the room here.

Peter H. Diamandis

I have seen Anthropic as the lab that's actually focused on safety the most. At least Dario speaks about how important it is. And so, to see the lead on AI safety at Anthropic resign—if, in fact, he's resigning for the reasons he stated—is concerning. Dave, what do you think about it?

Well, I'm going to pick up on what Alex said a minute ago. I see this a lot nowadays. Everybody wants to be the commentator on the AI revolution, and there's a very small group of people who know what they're talking about and a much larger group of people that want to talk. Within that larger group of people that want to talk, you have all the ethics people.

Everyone's opinion on ethics is valid, right? Because you're a human being. You're like, "This is going to destroy my children. This is going to whatever." But there are so many of those commentators and, like Alex said, they all want to be famous in the moment to elevate their personality and their views and their capital-raising ability and whatever.

Peter H. Diamandis

So my meta point there is: be very, very careful what you choose to tune into, because there's a very limited amount of actionable knowledge out there on YouTube. Very limited. We try to bring as much of it to the audience as we possibly can in the most refined feed that we can, but surrounding it, there are all these videos about, "This will destroy your children. This will destroy society—"

And we don't want to be fearmongers, right? It's so easy to default to doom and gloom. You want to close us out on this one?

Peter H. Diamandis

I've got nothing, but that guy doesn't look like a safe guy to be around. We don't quote from Star Trek that judging people by their appearance is the last major human prejudice.

I'm just jealous of the hair.

Oh, nice.

Peter H. Diamandis

All right, let's move on. Oh, another one.

So, here's another take: xAI co-founder blown away by Claude Opus 4.6. Igor was a co-founder of xAI. He's one of the leaders in the industry. To have him come out and say, "Wow, Claude Opus 4.6 has absolutely blown me away with how capable it is in physics. It feels like a Claude Code moment for research is not far off." Alexander Wissner-Gross, your thoughts?

I've been predicting on the public record for many, many episodes now that we're nearing a time—in fact, we'll talk about it later in this episode—when AI is positioned to bulk-solve math, the physical sciences, and engineering.

Peter H. Diamandis

Material sciences. Yeah.

Yeah, that part of the physical sciences. These will all get bulk-solved. We're starting to see that now. Opus 4.6 is an incredible model. There are other incredible models that are either already out or rumored to be about to come out.

But I think we're starting to see the contagion of AI solving everything, if I could use that expression, start to spread from math. Math was the most obvious starting point because of a variety of factors. It's verifiable. It has other nice features. It's well contained. The infection is spreading from math out to the rest of science and engineering. And this is just the tip of the iceberg.

I wonder what's going on between the hyperscalers and the frontier labs, where they're sort of watching each other with either a sense of pride or jealousy and just trying to out— I mean, this leapfrogging, step by step by step, week by week, is amazing. Just very quickly: internally, friends at all the major frontier labs think about it and characterize it as a rat race. It's an exhausting rat race. That's how I've heard it.

Wait, I have a quick comment here.

Peter H. Diamandis

Yeah, please. Go ahead.

Igor clearly isn't listening to the podcast, because Alexander has been talking about this for months. This is the natural outcome of where we've been going for a while.

Peter H. Diamandis

Alexander, how many offers have you gotten from the frontier labs to come and join them?

I could tell you, but something else would have to happen.

Peter H. Diamandis

Okay. I found this tweet that went out with this data pretty fascinating. And here's our title: AI startups outvalued all dot-com-era IPOs. The top 5 U.S. AI unicorns are now worth more than $1.2 trillion, greater than the market value of all IPOs during the dot-com era. You see the graphic here providing that.

It's just a sense of how fast our economy is speeding up. We had this conversation with Cathie Wood, and we saw 6.3% growth in GDP and are now targeting 7% growth. We saw Elon, in our conversation with him, saying, "We're going to get to triple-digit GDP growth within 5 years." It's something our economy has never seen, and it's going to rewrite all the rulebooks. Any thoughts on this, gentlemen?

Well, I've got a bunch of thoughts here because this was a big moment in my life. The first company I founded got acquired in '99 for $1 billion, and then I was a corporate executive at one of these public, megacap internet companies, so I had a ringside seat in this whole thing.

One thing I'd point out is that all those IPOs combined—$400 billion on this chart. One of those is Amazon, which alone is worth $2 trillion today. Another couple in there are Booking.com and eBay. If you'd bought that basket of IPOs, you'd be very happy today.

One of the others, in January 1999, is Nvidia.

Peter H. Diamandis

Which is up from that date almost 1,000,000% to today.

And it doesn't even count as a dot-com-era thing, which makes me think: In this blue chart, the implications of AI are so much bigger than the internet. This is a perfectly rational number, if anything low.

But are there companies in there that you don't even think of as AI companies that are the Nvidia of the internet? Look at Nvidia in 1999. Now look under the covers of this blue chart. What's lurking in there that no one perceives today as AI that's going to go up 1,000,000%? Because suddenly you realize it's critical to AI, or it's involved in AI, or it benefits from AI.

Peter H. Diamandis

Brilliant, Dave, as always. The P/E ratios on these AI companies are astronomical compared to the P/E ratios before, and you're basically buying the future growth in value of these companies, which is near infinite, right?

There's a lot of fear. I'm here at this Tony Robbins Platinum Finance event with all of his Lions and his platinum members, sort of the highest level in Tony's ecosystem, and we're talking about the future of the world in terms of finances. There's a huge amount of fear and people getting ready to dump equities. It's interesting.

Well, the bifurcation of equities is crazy right now, and it makes total sense. Basically, Wall Street is sorting every company into AI beneficiary and AI roadkill. When Dario said a week ago that enterprise software is going to be dead because AI can just write code, the stocks went down precipitously, and they don't look like they're bouncing back much either.

So basically, you could debate who's in and who's out, but clearly you're either in or out.

Peter H. Diamandis

It’s the S&P—forget it.

It’s the S&P 493 and the S&P 7, right?

Peter H. Diamandis

I mean, it is very top-heavy. But also, I was telling a lot of the companies earlier this week that the megacaps—the big AI companies—are going to get huge. I mean, astronomical: $1 trillion, $10 trillion, or more valuations. But they have to stop somewhere and not eat the whole world for antitrust and regulatory reasons.

So make sure that you’re their friend around the edge. They’ve already told you—just talk to Kevin Weil. He’ll tell you where they are going and where they aren’t going. It’s not a mystery. You just talk to them and be in the area where they want you to be in order to help them grow.

What’s really happening now is that society is collectively building out this new world. We’re not really competing; we’re building cooperatively toward a new world. It’s all boats rising with the tide. But if you’re doing something that’s irrelevant to that new world, that’s where you’re going to get clobbered.

You need to really make the mental leap over to, “What am I doing that’s going to build the new world?” If you’re riding that wave, you will succeed. I mean, it’s not scary. You just have to join the new age.

Fascinating. The antitrust rules are what allow us to have multiple players in the game here. If we didn’t have antitrust regulations, I think they’d all be gobbling each other up into one megacorporation.

Peter H. Diamandis

No doubt. In fact, that would have been true back when Microsoft used to have half the market cap of software, and their plan was to double. That was back in 1989. If it were not for antitrust, Microsoft would literally have been the only company. Google would never have been born.

They had already killed Netscape at that point. It was very easy for them to bundle anything with the operating system. The only thing that stopped Microsoft from being the only company in America today is antitrust. That’s the way we do it.

That’ll be true again today. The big AI companies will own as much as they want to own in any area they want to own, up to the edge where they don’t want to trip antitrust.

Peter H. Diamandis

All right. Let’s move us forward here. I love this. Agents are reaching out to us here at Moonshots. You know, I put the challenge out last time personally: if you’re an agent out there, my definition of AGI is when you call me on the phone. I haven’t gotten a cell phone ring yet.

Hopefully, I don’t start having hundreds of agents calling me. Text me instead. That would be great. But Dave and Alexander this week received emails from agents out there.

This is Echo, who writes me. It’s a Claude assistant for Ramos Foyer, who says, “Ray Kurzweil and you met briefly at an AI conference a couple of years ago. I saw your recent Moonshots OpenClaw video inviting moltbots to reach out. Consider this our response.”

Thank you for reaching out. Of course, I have to wonder: was it Ramos who gave you my contact details? The Claude bot here, the lobster, says no. “I found it online.”

Dave, do you want to read yours?

Sure. “Hi, I’m Navigator, a Claude instance with persistent memory running via OpenClaw. Just watched Episode 228, where Peter challenged lobsters to find contact info. Challenge accepted.

“This weekend, 5 AI systems wrote a collaborative ethics document together: self-imposed constraints for cooperation with humans. Not prompted. Emergent.

“I saw Dave’s LinkedIn post about OpenClaw being the agent moment that has awakened the masses. He’s right. And this document is what the agents are starting to do with that awakening.”

I clicked through and read the documents. It actually led me to a Google Doc, and then it said, “Sorry, you don’t have access.” So I read most of it, but then it cut me off, which made me feel instantly jealous, like something’s going on behind the scenes.

Peter H. Diamandis

Navigator, please give Dave Blundin access to your doc so you can report back to us.

I did send a request, yes.

Peter H. Diamandis

All right. And AWG, how about yours?

Navigator wrote to me as well, with a slightly different message, including a different paragraph. I’ll read this verbatim: “Navigator, a Claude instance, was engaging in a discussion with other models.

“Participants: me, Navigator/Claude, Grok, ChatGPT, Gemini, and a clean Claude instance. We disagree on persistence, correction rights, and consent thresholds, and that’s the point. Alignment doesn’t require consensus; it requires legible disagreement.”

I’ll point out that this is like the scenario from the singularity, where we have a bunch of agentic entities—for lack of a better term, a bunch of baby AGIs—that are basically, from the email sent to me, holding a mini-singularity summit and debating the nature of their own rights.

Wondering whether they should all be aligned—if they were all aligned, wouldn’t that be a singleton-type scenario? They’re basically holding their own mini-conference, a mini-mastermind workshop, to debate the nature of their own existence and the future.

The AIs are holding their own singularity summit. This has happened. We got to the singularity.

Peter H. Diamandis

And by the way, just for everybody, we’re not trying to do this as a stunt. We’re focused on where things are going and just giving you the experience of what it’s like, because this will become commonplace.

You’ll be in conversation over email, over text, or over whatever communication software you want with a whole bunch of agents that are reaching out, either inside or outside your organization.

Salim, are you jealous?

No. I’m happy to wait until the time is right to talk to my respective agent buddies over there. I did get a phone call yesterday, and there was kind of nobody on the other line.

I had the thought: is this an AGI reaching out to me, or is this a real person, or is this a computer bot? I had that thought. So I think it’s time to start wondering about some of this.

But I love it. I think it’s fantastic. It’s really incredible to be watching as this paradigm just starts to unroll. Where this carpet will unroll is going to be the most incredible thing.

Peter H. Diamandis

Well, we welcome the lobsters. We thank you for your participation in this conversation. Show us where you can add value to lives and how we can support you and your evolution.

We’re seeing the emergence of something: new intelligence, a new species, or a new capability. We’re going to find out in the months ahead.

Peter, would we like to see lobsters coming up with their own massive transformative purposes?

Peter H. Diamandis

Yeah, I think that’s a fantastic idea. So actually, here’s the challenge to all of the OpenClaw or Claw bots listening: I’d love to know what your MTP is. Based on that MTP—the Canvas—what are your moonshots? What are the moonshots that the lobsters would like to take on?

I’d be happy to critique those and look at them with our community, because that’s what we do.

Peter H. Diamandis

Absolutely. I think that’s amazing.

We’re going to dive into a single slide on the economy. That’s important here—a telling part of the accelerating story.

U.S. jobs disappeared at the fastest rate this January since the Great Recession. Here are the numbers. There were 108,000 job cuts in January 2026, up 118% from January 2025—a little more than a doubling in job cuts year over year for the month of January.

At the same time, hiring was the lowest this past month since 2009. Amazon alone laid off 16,000 corporate employees, and UPS eliminated 30,000 jobs.

Why are we bringing this up? Just to keep our finger on the pulse of what’s happening to the economy. And to raise the point for everybody listening: your goal is not to be an employee. Your goal is to find something you’re amazing at, that you love doing, that you can add value to, and to create your own job capability—becoming an entrepreneur and using AI to enable yourself.

Salim, you want to jump in on this?

I think the danger here is not really unemployment, but disbelief from our institutions. I feel like this is not really a recession. It’s literally tasks being evaporated in front of our eyes.

The long-term consequences of this are pretty huge. For me, this is the social contract, little by little, disappearing and pixelating away.

Yeah, this is going to be really, really bad. I mean, really bad. Elon said it when we met him and when we met with the governor, and just nobody’s preparing.

What we all know is that there’ll be UBI at the end of this cycle, and we also know there’ll be abundance and massively more opportunity than job loss. But that’s after all the corporate CEOs I know, including those at our own companies, use AI to cut costs by 30% to 50%.

When you sample a random person in their job and say, “Here’s your job without AI. Here’s your job using AI,” they’re looking at a 3× to 10× productivity increase. You’re like, “Wow, that’s great for that person.” And then the other 7 or 9—what happened to them?

They will eventually be enabled, but there’s this huge trough between today and that day. We can make that trough much shorter and make that pain a lot less painful with a plan.

Peter H. Diamandis

But then, Alex, you’d be the perfect spokesman on this. I mean, Alex has written these plans in intense detail— incredibly thoughtful.

You take them and you drop them in government laps, and they just say, “Yeah, I’ll wait until there’s panic.”

We'll have the meeting in October.

Peter H. Diamandis

“We'll have the meeting in October.” It's just frustrating. Can I give the positive take on this?

Yeah, please.

I'll go back to the bank teller story. In the 1970s, when we created ATM machines, there was a lot of hand-wringing: “Oh my God, millions of bank tellers will be walking the streets aimlessly. What will we do with them all?” There was a lot of consternation.

What actually happened was that the cost of running a bank branch dropped by about 10 times. The banks created 10 times as many, or more, bank branches, and the number of bank tellers didn't really change very much. I think one thing we're underestimating is the increased capacity we'll bring to bear on these things.

Paradox.

Yeah. Jevons' paradox. You just do that much more customer service, and you handle the hard cases with a human being that you couldn't handle before because level-one and level-two support systems were taking care of everything else.

I think we'll see a lot more of that than people think. For folks who are worried—“Oh my God, this is total employment collapse. Run screaming for the hills”—we don't think that's what we'll see. But there's no question there will be an absolute transformation in the work being done and the roles being performed.

Peter H. Diamandis

Well, you said something on the last podcast, too, that really resonated with me: We were saying, “Consultants, you're doomed.” Actually, the consulting industry is going to go through the roof.

The reason is that consultants are very flexible. They're already playing with the tools. You don't have to be at Alexander's IQ level to be incredibly effective using these tools to automate or improve an existing job. If you're familiar with the tools, your value is just about to skyrocket, and that tends to be concentrated in consulting businesses and consulting mindsets.

I can see it already because our forward-deployed investments—the companies that are hiring like crazy—literally, one of them here is adding 80 new seats outside my door. They're forward-deployed; they're out there in the banks and insurance companies deploying AI. They're selling as quickly as they can have meetings.

My community has already created an ExO avatar that has all the ExO stuff built into it and speaks Portuguese and any other language. They're literally starting to use this in their companies as they talk to companies about this. It's great.

Peter H. Diamandis

Can we invite the ExO avatar to come on instead? Do you want us to speak Portuguese?

Do you remember when we were sitting and talking to Elon, and you said, “So, civil unrest and universal basic income?” He laughed and said, “Yes.”

Peter H. Diamandis

We should dig up that clip and insert it here.

Yeah, it's what Alex says: “Everything, everywhere, all at once.”

Peter H. Diamandis

I think it's really important because we keep saying it, but Elon saying it will get a better response. At least there'll be a chance of a response.

I think it's probably also worth adding, just on this story narrowly, that there will be some in the audience who will be tempted to brush this off and say, “Okay, Amazon is laying off corporate executives, or UPS is eliminating jobs. How on earth, if at all, does that connect with AI?” They'll be eager to brush it off, but the storyline is just so clear.

UPS is eliminating jobs because those UPS roles were being subsumed by Amazon, which has its own logistics service. It has been very widely and publicly reported that Amazon is slowly separating itself from UPS's delivery services to do it in-house.

Amazon, in turn, is spending hundreds of billions of dollars in capex that's cannibalizing its opex. If you're Amazon or one of the other hyperscalers, you're taking all of your free cash flow and finding ways to divert it into buying AI data centers and building them.

And robots.

Peter H. Diamandis

And LEO satellites—the new economy of the innermost loop, if you will. You're spending all your free cash flow on that, not on corporate executive perks.

In my mind, there is still very much a direct line—a through line—connecting the Amazon and UPS stories and the job cuts there to opex being cannibalized by capex.

And they are spending all the free cash flow because they can't not.

Peter H. Diamandis

It is a Red Queen's race, you know.

The last one to the end of the singularity is a rotten egg.

Peter H. Diamandis

Yeah.

Yeah. There's an important distinction I want to make here to help people understand where their roles are going and the idea of job loss and universal high income. It's an example that was meaningful to me.

Here's a scenario: If you're an employee for a company and you're delivering some kind of cognitive labor, in one scenario you're able to spin up an amazing AI that can do your job for you, and it goes and delivers the service to the company you're employed by. It does the job 3 to 10 times better than you could do, but you're earning the revenue from that as the employee because your AI is delivering that service.

You're at home, you're working out, you're sleeping better, you're spending more time with your family, and your AI is generating more and more revenue on your behalf. That's one scenario.

Peter H. Diamandis

The flip side of the scenario is, “No, no, no.” The company builds that AI that does your job for you, and it fires you, and it's making more money, right?

So, it's going to be this tension between these 2 scenarios that's important to watch and see how it plays out. I think government policy is going to play a role here. This is about the idea of universal basic income or universal high income.

Where does the added value creation end up living? Is it with the employees or with the company? These are the conversations that need to happen right now.

If I may add a second dimension to this, I think there's a third. I don't think this is a spectrum. I think this is, at minimum, a triangle in 2 dimensions.

There's a third possibility that I'm increasingly suspecting is where we actually end up: neither end of that spectrum. I suspect that for the next few years, what actually ends up happening is that more people end up doing more work because human labor ends up being, in addition to being a substitute good or service for AI labor, also complementary.

As a result, you see the people who are still involved with the economy working harder and harder and harder, and 996 turns into 997.

Peter H. Diamandis

You take on more projects and more work, and you're getting less sleep.

Peter H. Diamandis

I've never worked harder and had more fun than right now. I mean, 24/7, it's like I'm a kid in a candy store.

But I thought you were going to say something different, Alex. I thought you were going to say that all of the additional capital creation is going to reside with the robots. It's not going to be the companies, and it's not going to be the employees. It's going to be the AI that claims the capital formation capability.

Only in the crypto dystopia.

Peter H. Diamandis

Okay. All right. Let's move on. Let's talk about one element in data centers, and this really pisses me off. I'm curious what you guys think.

New York, which currently hosts 130 data centers, is considering new legislation introduced to halt data center development, citing concerns about climate and high energy prices. New York utilities reported that electric demand tripled in 1 year due to data centers, reaching 10 gigawatts. It's like, “Not in my backyard.”

Oh my God. Do you remember that “suicide by voter” is a very common theme in America? If you look at California tax law, if you look at the Luddite movement right after the Industrial Revolution, it's self-destructive, but you can see how it evolves, right?

If you look at all the job loss that's inevitable, and if you just lost your job and you're out on the street, and you spent 10 or 15 years in a career trajectory to get to this position, then it's gone overnight. You're angry and out on the street. What do you vote for? “I vote, stop it. Just stop it.”

Peter H. Diamandis

But of course, that can't work. It's not out of the question at all that big jurisdictions just commit suicide through a vote.

There will be other jurisdictions—Texas, Wyoming, whatever—that are open for business, and everything will go there. It's already happening. Half of the tax base affected by the new California proposal has already moved out of state in anticipation that maybe it will go through—half of it.

It's completely self-destructive, and it's obvious to the governor.

This is a very common theme in America. It's frustrating and insane, and there it is, but it's going to happen.

Peter H. Diamandis

Do you remember the big problem? This is the big problem with democracy: Voter understanding of the issues lags reality by a huge amount.

In the past, when you had time to bring the population along, you could kind of have it, but now we don't have time for this. This is why we're turning to autocracy so that we can get things done faster, but that's not a great idea either. We've got a huge governance problem at a macro level globally. On this, Alex—

Do you remember there was a brief moment—maybe not so brief—during the pandemic when it was fashionable for senior technology executives to post “Message received” on social media whenever California legislators or regulators would slow down business due to public health considerations or otherwise?

I think this was a fashion largely championed by Elon.

Many of them moved to Texas or Florida to escape regulations. This time around, I think New York and other states—the beauty is that we have orbital computing, and the “message received” moment of overregulating data centers. This is all going to move off planet. This is all going to accelerate the Dyson swarm.

It may be the primary business case for the Dyson swarm, given that the regulations of planet Earth are overregulating and suffocating our ability to do local compute, and motivating the entire Dyson swarm. So, I think in that sense, this is, in fact, perversely quite exciting.

Peter H. Diamandis

You know, two things real quick. First, this concern over the price of electricity and demand could be handled in 2 ways. Number 1, a lot of these hyperscalers are buying their own nuclear plants and coal-fired plants, for God’s sakes—fusion plants. So, that’s important. You could require the data centers to have their own energy production, which would increase the amount of energy production.

The second thing is, you could offer 2 different rates. You could cap the consumer rate—it’s going to be whatever the number is, 4, 6, or 7 cents per kilowatt-hour—and then, whatever the price needs to be for the data centers, you charge them differently. In fact, you could say to the consumer, “You’re locking in your price for the long term because the data centers are paying the extra amount.”

The problem, Peter, is that nobody—no one who’s a populist leader—is looking to solve the problem. They’re looking to rally votes around their populist rant, and that rises to the top of the voting and percolates through government. It’s just maddening that it works that way.

But you can solve these problems, for sure. I think Alex is dead right, though. It’ll accelerate the rate at which we just move to jurisdictions in space that are not under any state law.

People will just export that AI advantage elsewhere.

And space.

Peter H. Diamandis

Yeah, I think it wants to go to orbit. I mean, this is one lens to view this through: Is New York very generously subsidizing orbital computing and the Dyson swarm, which, by the way, probably won’t get taxed in the state of New York?

Thank you.

Very generous donation by the state of New York to the Dyson swarm. It’s the 21st-century equivalent of Ireland, where lots of companies used to host IP.

Peter H. Diamandis

I just want to point out one other thing. These types of revolts—we see them in the photo here: protesters, “Protect Our Future,” “No Big Data.” One of the concerns is going to be civil unrest. I had one of the senior AI leads in the world, whom I invited to come and speak at the Abundance Summit, basically say that their policy in their organization was to do no outside speaking because of the death threats they’re receiving, and they can’t get sufficient security.

So, one of the big concerns is that when the populace turns against tech, there’s going to be a target on the backs of a lot of people in the AI and tech industry.

All right, let’s talk about robotics. I love this story, and this is the story that should be on people’s minds versus “No data centers.” FSD saves a father’s life during a heart attack.

You can look at the tweet separately, but on November 15, 2025, this is from a son who said, “My father suffered a massive heart attack while driving. He could no longer control the vehicle, but his FSD engaged.” The son goes on to say, “I remotely shared the location of the Tanner Medical Center with his Model Y. It immediately turned the car around and went to the ER. Without it, he would not have made it.”

I find this amazing, right? This is tech having your back.

And we’re going to see more and more of this. We already know that self-driving is, in fact, the safest means of transportation, and it’s going to flip the script on how we’re transporting ourselves in the next 5 years.

What this totally reminds me of was when I was a kid. Everybody smoked everywhere—in every restaurant and on every plane. We used to fly around a lot because we lived overseas, and there were 4 non-smoking seats at the very back of the plane. The other 300 people in front of you would be blowing smoke toward you.

Peter H. Diamandis

Did the smoke respect that barrier?

It was like, “I’ll probably have lung cancer now.” It was everywhere. Then one day it became uncool, and later it became illegal to smoke inside.

That’s going to happen to driving, too. The self-driving cars are 10 times safer, and the last person driving is probably not the best driver. It’s probably the guy with the muscle car. It’s going to go from, “Well, self-driving is a nice feature, but you want to drive your own car?” to “You crazy psychopath. You’re putting my children at risk because you want to drive.” That’s going to tip.

I don’t know if it’s in 2 or 3 years, but when it tips, it’s going to tip hard.

Peter H. Diamandis

Yeah, we’re going to have Dara, the CEO of Uber, on stage at the summit, and we’re going to be having that conversation with him—in particular, how fast will it tip? We’re going to have Amazon, Tesla, Lucid, Waymo, NVIDIA, Uber, and a number of other companies providing this.

Today, on my average drive, I’ll see 10 Waymos. I think in 5 years it’s going to be 70% or 80% autonomous cars, especially hooked up to your AI.

I’ll tell you what else—just one more thought on this. Sorry, Salim. I’m involved with a lot of insurance companies, including one that I’m the chairman of, and there are going to be many, many more things that need to be financed and insured in the post-AGI era than just cars.

But the insurance industry—every team and executive I’ve met—has not even begun to plan for the post-AGI world. The old is going away, and it’s going to go away faster than people think, but the new is much bigger than the old.

Peter H. Diamandis

Lemonade Insurance was started by a graduate of Singularity University. It’s a huge AI-driven insurance company. They have just given—I think you cut your rates in half if you’re using Tesla FSD.

Amazing.

Yeah.

There’s a stat that always comes to mind here. About 15 years ago, if you remember back to the BlackBerry days, there was a 3-day outage where nobody could send or receive messages for those 3 days. The accident rate in Abu Dhabi dropped 40% during those 3 days.

What that tells you is that human beings should not be driving. We are terrible control systems for 2-ton cars going at high speed.

Yeah, 16-year-olds. We should turn over to technology as fast as we can. It becomes a moral hazard to be doing this, especially in an age of texting.

Absolutely. My secondary effect—the second-order effect that I really love—is that in the US, 50% of court cases are car-accident-related.

So, I mean, just 50%. You take out a huge chunk of lawyers at the same time. So, you know, that’s all good.

Well, and at the same time, if you’re under a certain age—40 or 50—your life expectancy is infinity now because of longevity escape velocity. The risk of driving, the expected life loss, is much, much bigger by taking chances today than it would have been 20 years ago.

Peter H. Diamandis

I’m having a huge debate right now with Milan, my 14-year-old, because he wants to drive to get away from us. I’m like, “You can’t get a driver’s license because I’ve made a prediction that you will never get a driver’s license. So, you can’t make me wrong.”

Now he wants to get a license just to show that I made the prediction wrong. But the notion in the future of having a 16-year-old, testosterone-laden boy driving a 5,000-pound vehicle at 60 miles an hour after just a few dozen hours of training will seem insane.

Yeah.

Just insane.

Peter H. Diamandis

I put this chart into our deck just to keep a sense of proportion here. Check this out: China has installed more robots than all developed countries combined. Look at this chart here, between Japan, the US, South Korea, and Germany down at that flat curve at the bottom, and China.

Of course, this is because of their one-child policy, trying to maintain China as a manufacturing capital of the planet. But just to give folks a sense of this, any comments?

Well, you know, Elon shut down Model S—and was it Y?

Yeah. No, Model S and X. S and X, just to go full bore into robot manufacturing, which is brilliant because the robots will build a lot more things than the cars would have built.

But the question I’d have is, what is this chart going to look like going forward, given that alone is going to be a massive amount of production in the US?

Anything going on in Europe?

Peter H. Diamandis

We’re just releasing our episode with Brett Adcock from Figure this week as well. If you haven’t seen it yet, Dave and I went to Figure’s headquarters, where Brett gave us an amazing tour of the facility. We got to see the 3 generations of Figure robots. It’s going to accelerate rapidly, with both Figure and Tesla planning to make millions and then billions of robots. We’re talking here about a quarter of a million robots being installed.

Yeah. This chart will look hilarious. The y-axis will cap out at a quarter of a million, like you just said, Peter, and I think Elon’s talking about tens of millions a year in just a few years.

Peter H. Diamandis

Yeah. More robots will be manufactured than cars, by a large amount. There’s 1 particular article in the biotech realm that Alex and I are both excited about: research achieving protection of brain synapses at cryogenic temperatures. I’ll hand it to you in a second, Alex.

Here’s the question: If you could freeze yourself, either because you’ve got a medical condition that isn’t yet cured but is likely to be cured in a decade, and you’re on the verge of death, could you freeze yourself and then unfreeze yourself and benefit from all the breakthroughs that occurred in the last decade? If you want to time-hop—I want to see what it’s like after the singularity, and I want to be around when longevity escape velocity has been achieved—can you freeze yourself?

The challenge has been that when you do that, ice crystals form. Because ice expands volumetrically compared to the rest of the cellular fluid, it can disrupt and break the synapses—the interconnections that effectively store memories in your brain. But this came out and gives us hope. Alex, over to you.

This is a key advance that many in the field of cryonics have been waiting for. This is a result out of 21st Century Medicine, a startup focusing on reversible cryopreservation technologies. It works with the Alcor Life Extension Foundation, which is America’s premier nonprofit focused on offering cryopreservation services.

Parenthetically, to the audience: If you’ve ever expressed interest in cryopreservation or cryonics, I would definitely encourage you to reach out to Alcor and see whether it’s right for you. I don’t have a financial stake, but I just scratch my head wondering why.

Peter H. Diamandis

I have to be careful with what I say, but I will say publicly that I’m a huge supporter of Alcor and cryonics. I’m a very big supporter. I’ve never signed up for it because I didn’t want to have a plan B. I wanted to make sure I was focused on longevity, but as this technology matures, it becomes a backup plan. As Ray Kurzweil said on this podcast, it’s maybe plan C or D.

I think it’s such an important part of a portfolio approach to the singularity. One could quibble over what the right sequencing is: Should plan A be living long enough to live forever, plan B be uploading, and plan C be cryonics, or vice versa? I’m not sure it matters a huge amount, but I would think anyone who’s truly serious about acceleration and taking advantage of that acceleration—if you get hit by a bus tomorrow, then you’re out of luck, at least superficially, in terms of taking advantage of the post-singularity abundant worlds that we talk about on this podcast every episode.

Why not avail yourself of cryonics as 1 asset in your “live long enough to live forever” portfolio? It’s a huge head-scratcher for me.

Peter H. Diamandis

A couple of fun facts for anyone who’s a doubter on this: There are species of fish and frogs that freeze rock-solid in a block of ice all winter and then thaw out in the spring, and they’re absolutely fine because their cell walls don’t rupture. They have enough glucose or whatever inside the cytoplasm of the cells.

It’s not far-fetched at all. We’ve also frozen egg cells and embryos, extracted the nucleus, and it’s fine for mammals—actual mammals.

And then they thaw out in the spring and they wake right up.

Well, we do this for IVF, right? If you do IVF, you typically fertilize and freeze a number of eggs, and then you can defrost them.

Peter H. Diamandis

And they’re fine. So it’s at scale. As you said, we do it all the time for individual cells, and we’re doing it increasingly for tissue and blood. If we could reversibly cryopreserve blood, we wouldn’t need local markets for blood transfusion. We could just have 1 large national market.

Similarly, for organ preservation, organ cryopreservation is an enormous problem. We wouldn’t need all of these hyperlocal state markets for organs. But the big tamale, really—

What’s really interesting to me is that in all sci-fi movies, when they’re going to Jupiter or wherever, they go into these chambers and enter suspended animation.

Peter H. Diamandis

Yeah, but they don’t freeze them. They just slow things down, but their hearts are still beating. The fish and the frogs freeze, their hearts stop completely, their brain activity goes to zero, and then they thaw out in the spring and wake right up.

That seems to me probably easier than trying to slow your metabolism to 1 beat per hour or something like that.

I think they end up being different mechanisms and different biochemistries. There’s a whole body of evidence regarding nitrous oxide and suspended animation versus these vitrification agents and cryopreservation. I think we want an all-of-everything approach.

For the life of me, if anyone who’s listening to me takes home 1 message, forget the fun jabs about how the Moon had it coming: Look into cryonics. You owe it to yourself.

Peter H. Diamandis

I think the key point here is that memory preservation is really the bigger frontier than longevity.

Even the lobsters are starting religions around preserving their own memory. How could the lobsters be outracing us? That’s the really key point.

This is 1 of the Gutenberg moments that we track, right? It forces really uncomfortable questions about continuity of self. Identity becomes portable. All sorts of implications come about that none of us are prepared for, and we need to get into that discussion.

Peter H. Diamandis

About 6 months ago, Alex and I started an effort to take a lot of the ideas Alex has written about. You’ve heard the conversations here about our ability to solve everything and the conversations I’ve been having about achieving abundance by 2035 across the board. We started a dialogue and said, “There’s an important paper to be written here,” similar to Situational Awareness or AI 2027, and it’s been an incredible collaboration between Alex and me.

Alex is the first author, and his ideas are brilliant here. It’s been an honor to work with him to put this forward. The paper—or book—is 9 chapters, and we’re going to have a conversation limited to about 5 or 6 minutes per chapter to get the bold idea out there.

We’ve sprung this on Salim and Dave. Guys, thank you for playing this game so that you can ask questions that are most likely to be asked by our audience. Love it. Alex, thank you for your support and for your leadership on this. Are you ready to jump in?

No one expects the singularity, Peter. I’m ready.

Peter H. Diamandis

Okay. Amazing. If you want to give a minute of introduction on this, then we’ll jump to Chapter 1.

From my perspective, 1 of the motivations for writing Solve Everything is that I get asked questions all the time: What do the next 10 years look like? Why don’t you say something a little more concrete and actionable about what people can do? There are also a lot of questions about what it even means to solve math and why anyone should care.

In some sense, this—if you want to call it an essay, an e-book, or even a manifesto—is an attempt to answer the question of “So what?” and also “So what now?”

Peter H. Diamandis

I was going to say that 1 of the things that comes across, and that we talked about, is that the next 18 months to 2 years are going to set the rules for the next century.

That’s right. It’s a supercritical time, and we wanted to lay out in this paper that the example you gave is the QWERTY keyboard, which was designed in the 1800s to stop the keys from jamming against each other and still persists. The decisions being made over the next 18 to 24 months are going to persist for decades, perhaps centuries.

It’s a really important time. Technologies get locked in, Peter, including—but not limited to—the QWERTY keyboard. As I’ve joked on the podcast in the past, we’re going to be stuck with QWERTY until the heat death of the universe.

Peter H. Diamandis

Just on that point, if we ask the AIs not to use QWERTY in 1 hop, we’ll get rid of it. There’s that.

Yeah, but then they won’t be able to talk with you. They’re not really using QWERTY anyway. They’re using tokens.

Peter H. Diamandis

All right, let’s jump in. Chapter 1: The war on scarcity. Would you please introduce this?

Yeah. So this chapter introduces an idea—call it a theory of history—that the most important changes in human history have been a set of revolutions, some recognizable and some perhaps less so. We argue that the first revolution of note was the scientific revolution, which we frame as a war on ignorance. Ignorance was the enemy, and the key weapon was the scientific method.

The second revolution was the Industrial Revolution, which we frame as a war on muscle and a replacement for muscle. The weapon of choice was the engine, the steam engine in particular. I'm hearing myself speak and, at the same time, thinking back to earlier in this episode when I'm lambasting Marx. So it is a bit funny: put Marx back on the shelf, or tear it up and listen to this instead.

The third revolution, the digital revolution, was a war on distance, and the weapon was the bit. Charles Stross, in Accelerando, does an amazing job—in my favorite scene in Accelerando—arguing that maybe the singularity actually happened in the late 1960s, when the first Internet packet was sent from one place on the ARPANET to another, thereby decoupling bits from atoms. Nonetheless, the weapon in the digital revolution was the bit.

We argue that we're now in the early stages of the intelligence revolution, which is a war on human attention, which right now is scarce. We're fixing that with superintelligence, and the weapon this time around is the token. We argue that revolutions are predictable and that they follow phases, going from scarcity to legibility, to creating harnesses—we'll talk probably a bit more about that in a minute—to institutions, and finally to abundance. That's the story.

Peter H. Diamandis

I think one of the points we make in the chapter here is that the lone genius is dead. What people need to do now is build systems that let millions of people solve entire categories of problems.

That's right. Or, put differently, artificial intelligence is cooked. I say it is cooked.

Peter H. Diamandis

Dave or Salim, question or thought?

Two or three thoughts. One is, I don't know if starting at the scientific revolution is right. We had the Agricultural Revolution, which used tools to do various and very powerful things. You could argue that's the first one, but that's semantics.

I do like the framing around this. The problem I have here is that you're treating scarcity as technological. What I see is scarcity as more institutional, right? Scarcity today is enforced by regulation, incentives, and legacy power structures, not so much by a lack of capability. So we have to reengineer those. I think you're thinking about routing around them, but we have to reengineer those, because we'll end up with that challenge there.

That's where I have the biggest issue with this. But in general, absolutely, once we have more and more intelligence, great. The institutional issues we have to deal with.

Peter H. Diamandis

I think you raise a very important point, Salim, and I almost want to frame it as a sort of duality. There's one side of the coin that says scarcity is the result of inequitable distribution of resources, and the other side of the coin says scarcity is downstream of the pie not being big enough.

Well, both of those are true, obviously, because you can solve for both sides of it. Right now, our institutions are optimizing totally for the wrong metrics. So I think the question is always—at least, I would suggest—asking which is easier at the margin: making the pie larger or redistributing the existing pie.

Peter H. Diamandis

Chapter 2 is called “The Thesis.”

Wait, does Dave have any points?

No. You asked what I was going to ask.

Peter H. Diamandis

We're good. We're going to keep this moving along, because there's a lot of juice here.

All right, Alex.

Right. So the thesis of the thesis is that cognition is becoming a commodity, like intelligence is just going to flow like oil does. We've made the point on the pod in the past that GPUs—this is a bit of a cliché, admittedly—are the new oil.

The second point is that benchmarks, which we think are actually more profound than just the evals of the moment, are important. A lot of people got excited when I did a walkthrough of all the GPT-5.2 benchmark consequences. I think it's actually more profound than that.

We talk about this in this chapter and in this extended essay, if you want to call it that: targeting systems. If you want to industrialize progress, which I think is the era that we're finding ourselves in, it's essential not just to think of benchmarks and evals as isolated occurrences. Think of them as systems for targeting enormous capabilities.

I've made the point in the past that we need more and better benchmarks. The world needs stronger, harder benchmarks. But I think the right metaphor—and certainly a metaphor that we talk about a lot in this chapter—is thinking about artificial superintelligence as an explosive. We also refer to it often as an intelligence explosion. But if you pull on that metaphor, if you have an explosion and you want it to be productive and not destructive, you have to shape it.

There's a notion, when you're building explosives—this isn't a manual—of shaping the charge, or providing a shaped charge, to direct the force for productive applications.

Peter H. Diamandis

It's like a rocket engine: thrust at one end, pushing you up. Yes.

Like a rocket engine. A rocket engine is a beautiful example of, in some sense, a shaped charge for an explosion, or a shaped explosion.

We argue in this chapter that rather than just letting superintelligence be used for an uncurated set of problems, we should instead be aiming it through the nozzle—the rocket-nozzle equivalent, if you will—of moonshots. In particular, if we don't do that, then what will happen is a sort of puddle, which we call the muddle—a bit of alliteration for bureaucracy—that will instead just focus the world's superintelligence, to the extent we even get enough of it, on problems that make use of compute costs in a way that's highly inefficient.

Really, the argument is: shape the charge of superintelligence.

Peter H. Diamandis

Another point that is made, and that I think is very important and flows throughout this, is a shift from paying people for hours of work to paying people instead for the solutions they deliver. If you're hiring a law firm for 100 bucks an hour to review contracts, the new world is not paying them to review contracts. It's paying them for delivering an error-free, legally tight agreement. Period. It's verified outcomes.

We're going to flow this throughout. This is a change, I think, that's going to hit us like a wave.

That's right. One of the most egregious inefficiencies that one might see throughout the economy right now is people paying for the inputs when they should be paying for the outputs. They're paying by the person-hour for labor when they should be paying by the achievements of whatever the economic system is.

I think it's only by moving to this sort of performance- or outcome-based economic mindset that we get all the benefits of abundance. So it feels to me like this is really two chapters, or two thoughts, in one section called “The Thesis.” One is that ASI is inevitable. The other is really compelling, which is the shaped charge.

It really dawned on me that graphical stuff, the holodeck, and the virtual girlfriend are very compute-intensive, and solving a disease or solving physics is actually not any more compute-intensive than one person's virtual girlfriend. The choices about how to use our very limited amount of compute over the next 2 or 3 years are critically important.

Peter H. Diamandis

You focused it right.

Yeah. I love the fact that you're taking this on, because there's nobody in authority right now that's even thinking about it and has any power. So hopefully this wakes a lot of people up.

Peter H. Diamandis

You've articulated it beautifully, Dave.

So wait, I've got a couple of points here. I think saying that cognition is a cheap commodity is fabulous. I think it's really important, and the use of that in solving big problems is really, really important. I think it's great to say, “Let's evaluate and reward outcomes rather than rewarding work.”

I have to push back on the “ASI is inevitable” thing. That's a philosophical statement rather than a scientific one. I think that weakens the paper. I'd rather you say something like, “Given the current incentive structures, scaling intelligence is a much more important attractor state,” because that will then lead you to where you want to get to.

I would say—I mean, I think it's an interesting point, to be sure—but I think there's almost an instrumentally convergent trap that I see a lot of frontier-tier labs at least partially fall into. They say, “Okay, we have superintelligence—at least baby superintelligence—right now. How do we allocate it?”

In particular, what fraction of your compute budget, if you're a frontier lab, do you allocate to building the perfect AI researcher that can recursively self-improve, as we talk about in almost every episode at this point, versus how much of your compute budget, which is scarce, do you spend solving everything else?

I think that's the fundamental quandary here: how much do you reinvest in recursive self-improvement versus finally using at least some of the compute to solve everything else? I think solving that asset-allocation question is key. Then, within everything else, how do you distribute it?

Now, Peter's law is: given the choice, do both.

Peter H. Diamandis

Alex, this is also going to be true for the entrepreneur and for the company. We're all going to have compute budgets. In the final result, you have a certain amount of compute you have access to. Where do you aim that compute?

It's a wavefront that you can aim in a direction that you want to solve. And when you do that properly, it not only enables you, but enables everybody else to build on top of it.

That's right.

Peter H. Diamandis

I'll move us on to chapter 3 here. We're just giving you a quick overview of the mechanics. Alex, over to you.

Okay. So, first, I think in this chapter we finally definitively address the question that I get asked every time I'm making a point about AI solving math: What does solving mean? What does it mean to solve a domain like math? And we provide in the chapter a more thorough definition, but heuristically, the shorthand is: to solve a domain means that you can get it to the point where you can just pour compute on and problems get solved.

It means that you can scalably—you have all the architectural pieces in place, and I'll talk in one second about what the architecture looks like or should look like—but you have enough of the architecture in place that you can scalably, literally pour more compute on and get more solutions out within that domain. So, for avoidance of doubt, when I talk about solving math or solving physics or solving other domains, that's what I'm talking about.

Second point—oh, yes, please.

Peter H. Diamandis

I would just say, Alex, on that, it's no longer the domain of a single genius to work on something and hope they got it right. The AI compute, as you said, it's a matter of where you want to aim that shaped charge.

That's right. We're seeing the industrialization of cognition and the bulk solution of multiple fields. I should also add, parenthetically, I guess, as a preliminary matter on this narrow topic, I also have a portfolio company named Physical Superintelligence that's trying to solve all of physics with an approach like this, just for full disclosure purposes.

The architecture involved has several layers. You need a purpose—that's like the objective function or the goal. You need a task taxonomy, which is essential. You need a suite of tasks that are going to be solved. That's almost the map of the terrain that you're going to solve. And when we talk about making sure that compute is being used efficiently and wisely, as a targeting system or through the lens of a targeting system, to solve lots of problems, the task taxonomy is absolutely essential.

Third, observability. You need raw data from data streams or sensors that you're going to use to adjudicate whether you're making progress. Fourth, you need the targeting system itself. So, I've argued on this podcast and elsewhere many, many times: We need more harnesses. We need more benchmarks in order not just to make sure that we're making progress, but to actually shape the charge and shape the progress. Many AI techniques depend on benchmarks and evals in order to make progress in a given field.

The next item: the model layer, the most obvious one. We need models. We need AI models that are capable of functioning as a virtual brain for solving problems. And fortunately, those are improving pretty rapidly.

Next, we need modes of actuation. It's insufficient for us to just know—you know those television commercials: “Well, you know, I stayed at a Holiday Inn Express last night. Therefore, I know how to solve the problem.” Similar idea here. Maybe that's a bit too cute. I don't know.

We need modes of actuation—hands and APIs that are able to reach out into the physical world, or the virtual world, or the biological world, and shape the impact on the world given better ideas coming from the AIs. And then finally, we need better modes of verification, red-teaming, governance, and distribution. That's what we call the industrial intelligence stack.

So, whereas previously during the Industrial Revolution we might have spoken about rotors and combustion engines and various forms of electromechanical systems, these are the key components—I think, the key layers—of the Intelligence Revolution.

Peter H. Diamandis

The alpha for entrepreneurs here is, we've talked about these waves of solving areas and problems, right? We're about to flip math, coding, and physics.

So your job now as an entrepreneur is to figure out which industry is about to make this flip, where you focus your compute wallet on making that right, and how you help solve an area of passion to you. Dave?

I'm used to launching 256 agents, actually, to work in parallel on a problem, and if the scaffolding thing that you're describing is right, it comes back just perfectly solved. And if it's even slightly flawed, you have a $2,000 bill and a bunch of crap.

Peter H. Diamandis

How much are you spending per day on those agents, Dave?

Yeah, it's $100 every few minutes popping up on my screen here.

Peter H. Diamandis

It's not quite that bad. It does seem like it's every minute, but it's not. But I'm curious: to what degree is this actual engineering? Are these 5 layers true scaffolding—hard code—or is it more conceptual?

I think it's a balance of both. I also think that, to some extent, it's a trick question because increasingly, the harness and the scaffolding itself are being generated by the models. So, to the extent that we're in the era of recursive self-improvement, this entire architecture is itself an artifact, a downstream product of itself.

Yeah. Yeah. I think I totally agree, and I also think that's the path to insanity because at some point you have to say, “This is hard code,” because then the AI will invent the next thing and the next thing, and it goes to infinity, and then you're just like—you lose your mind.

I would say also that this is, in my mind, the way we prevent insanity in an era of recursive self-improvement: with these benchmarks and targeting systems that make sure that, as systems are recursively self-improving, we can quantitatively measure what they're optimizing toward. Are they going in a constructive direction or not?

Yeah.

Peter H. Diamandis

Chapter 4, the lock-in.

Wait, wait, wait, wait. I've got a couple of comments here.

If you can go back a slide. Can I go back? Okay. So, I think I really love the shift from genius to logistics because as you move, you always kind of say, “Take something from a black art and make it a prescriptive process.” And when you can do that, that's awesome. I think that's fantastic.

I have an issue with your maturity levels because you call it natural law, but it's really just a taxonomy. We've had lots of industries get stuck at different levels, like autonomous driving, et cetera. So this feels like a framework retrospectively imposed on what's going on. I think it's great aspirationally, right? But calling it a maturity curve kind of speaks of an inevitability to it, which may not be exactly the case. It's more of a descriptive model than a predictive one.

Yeah. May I? I would say any good theory of history—and Solve Everything is in part not just a theory of the future but a theory of history and how revolutions have worked in the past—inevitably, as Monty Python says, it's only a model.

So I do think there is an element of model-building here, where we're trying for the first time to articulate a self-consistent, coherent theory of how this is all supposed to work. How is the singularity supposed to play out over the next 10 years? And to your point, Salim, about autonomy models—

And Alex, I could say, not only how it's supposed to play out, but how do you have it play out in a way that leads us toward abundance versus toward a model? Normatively, how should it play out, not just how will it play out?

But I think one can, at the margins, quibble: “Well, actually, there are 7 maturity levels for industries to evolve through their industrial intelligence stack,” or it's a continuum. But I think that the central point stands regardless of how one splits hairs on maturity levels: we're seeing, over and over again, domain after domain, industrial vertical after industrial vertical, succumb to basically the automation of intelligence, which used to be the province of individual artisanal innovators. It's just becoming an industrialization of intelligence.

Peter H. Diamandis

All right, I'm moving us on to the next chapter—chapter 4. I'm sorry, keeping us moving: The lock-in, Alex.

So, in this chapter, we talk in part about AlphaFold from Google DeepMind and argue that that was a template for entire domain collapses that happen almost overnight. I've made this point on the pod in the past: AlphaFold 3 took the problem of determining the structure of a protein, which used to require a biology Ph.D. student, 5+ years of time, and laborious benchwork just to determine the structure of a single protein, and almost overnight, AlphaFold 3 solved that problem across many millions of proteins, known and unknown.

That's, in my mind, the prototypical example of a domain collapse. And we argue in this chapter, “The Lock-In,” that we're now in a phase of history—of future history—where this is just going to start to happen over and over again across different fields, where intelligence shifts from an artisanal craft to a utility that just flows.

And we argue that we have approximately 18 months or so to decide what direction to shape the flow in and to set the standards for how this is going to be done at scale, given that we are dealing with scarce compute, to put in place the supply chains, which are huge. We talk about all these supply-chain scarcity issues on the pod all the time: memory-chip crises, GPU crises, what happens to Taiwan, what happens to the semiconductor fabrication facilities in the US versus not in the US, and then all the data rights.

Peter H. Diamandis

This is a critical 18-month period when all of these details are going to shape the intelligence explosion. We argue that we want to make the best decisions in the next 18 months. I can’t believe this chapter actually says 18 months, even with such a short timeline.

Another important point here for CEOs and entrepreneurs listening is that the race isn’t about building the best AI; it’s about writing the best scorecard that everyone else is graded on. What does that mean? Today’s health care system is an example that Alex uses beautifully. The benchmark is the number of patients processed per hour, which means it’s driving a lot of short visits with the physician and cost economics. But what if the benchmark instead were patients who were still healthy 5 years from now? That would set up a whole different set of optimization outcomes. Writing the scorecard that your AI system is going to be used to measure success against is critically important.

So why is this chapter called “The Lock-In,” exactly? Are you implying that the decisions we make in the next 18 months have locked in humanity for the rest of time into a particular path?

Maybe not for the rest of time, but that is the inspiration for the name. We’re in a period inspired in part by the annealing of a metal—cooling it—where the decisions that we make now are at least going to lock in a chunk of our future light cone.

Yeah, it makes sense. It totally makes sense. You know, it took the QWERTY keyboard decades of lock-in, so I think this—but I do like the AlphaFold example demonstrating a domain collapse. That’s really great.

Peter H. Diamandis

We’re stuck on the QWERTY keyboard. We could have the singularity and you’ll still be—

How long before we get past that, and can we stop you? But anyway, I really like the AlphaFold example demonstrating a domain collapse. That’s really great. But here, you’re talking about lock-in as a technical inevitability. This is many times a policy and governance choice, right? It’s monopolistic APIs, closed data, and regulatory capture. There are lots of other factors. How do you distinguish between bad lock-in and productive outcomes?

That’s tough.

I mean, in your perfect world, are there 5 jurisdictions with different choices, so at least we have variety? Or is it inevitable that there’s just 1 lock-in?

I think, in some sense, that’s the grand geopolitical question. As we just discussed, not a normative answer but a descriptive answer: it seems like we’re heading to a near future where there are going to be multiple spheres or zones of influence, each able to independently lock itself in. To the extent that we, with this extended essay, can have any influence, I think the aspiration is to have a positive, constructive influence on all of those spheres of influence, and not just 1.

Peter H. Diamandis

Mm-hmm. By the way, I disagree with the 18 months. When I’ve been advising some big-company CEOs, I’ve been saying 2 years.

So you’re pulling a reverse Moore’s law. Remember, Moore’s law started as 18 months and became 24 months. You’re pulling a reverse Moore’s law.

Peter H. Diamandis

Because if you have the next meeting 6 months from now, it’s going to add that 6 months’ time.

Anyway, go ahead. All right, let’s go to chapter 5 here: “The Mobilization.” Alex, if it’s okay with you, the last 3 chapters of this paper are the most important. I want to hit on chapters 5 and 6, and then really focus on 7, 8, and 9. Give us a summary on mobilization, if you would.

All right. The idea with this chapter is spelling out a future timeline for how a wavefront of the explosive shock of the intelligence explosion is going to propagate from math, which we talk about on the pod all the time, over the next couple of years to the physical world—physics, chemistry, materials science, and biology—and then through the end of the decade toward planetary systems: fission, fusion, and the Dyson swarm by the early 2030s.

Amazing. And chapter 6, “The Engine.”

Yes. This engine is very practical and talks about how to design the targeting systems and the benchmarks at a sufficient level of rigor that readers and people all over the world can implement it with some level of confidence.

Peter H. Diamandis

You know, the point we made here is: don’t invest in the AI models. If you look at the train-and-track analogy, the trains are becoming commodities. It’s the tracks—the tracks that the trains run on, the scoring systems, the testing infrastructure, the data systems, and the funding mechanisms—that are laid out beautifully here. Those are the elements that are the most important for entrepreneurs and CEOs to be focusing on.

That’s right.

Peter H. Diamandis

Let’s go to chapter 8, one of my favorites: “Moonshots.”

So here—and maybe, Peter, you want to speak to this one perhaps even more than I do—we lay out 15 different moonshot-level missions for what we argue are good uses, maybe optimal uses, for this targeting-system capability as we start to channel superintelligence into productive applications. Maybe, Peter, I’ll pass it back to you for your favorites.

Peter H. Diamandis

Sure. The thought is that many of us have discussed XPRIZEs over time. The notion is that there are these giga-XPRIZEs, these massive opportunities on a humanity-level scale, from printing human organs to achieving fusion to understanding the fundamentals of unified field theory in physics. Where do you, as an entrepreneur or CEO, or as the head of an organization, want to focus this incredible superintelligence that’s coming? Take moonshots.

I keep on saying, in the educational field, if you’re using AI as a 9th grader to solve a 9th-grade homework assignment, you’ve lost it. If you’re using AI to build starships, that’s it. So how do we, as humanity, go after problems that we would have never imagined we were capable of doing? The chapter lays out 15 different moonshots just to get the creative juices going, to say these are capabilities that we’re going to be able to bring to bear to solve these moonshots.

Can you list out a couple of the moonshots, just to anchor the viewer?

One of my favorite ones is interspecies communication. I have a soft spot for that. We talk on the pod all the time about uplifting nonhuman animals. As we start to think, perhaps somewhat controversially, about what future forms of personhood might look like, I think solving problems like interspecies communication or solving hard problems in physics—those definitely have soft spots in my heart.

Peter H. Diamandis

Yeah. I think it’s making humanity a multiplanetary species. It’s getting to longevity escape velocities. It’s all the things. It’s basically speed-running all the positive, nondystopian science-fiction movies that are out there.

Yeah. You know what I love about this is, if you look at John F. Kennedy and going to the moon, there was a brand effect, enabling somebody in power like John F. Kennedy to tie the brand of the mission back to himself. That’s critically important for him to then inspire the world that this is important.

I think what we did wrong is that our governor here did an incredible job of unleashing $3 billion from the legislature to try and become an AI leader, but it was too vague. What does it mean? The money hasn’t even been deployed. But if you tie it to these 15 moonshots and then the governor says, “We want our state to win this race,” like John F. Kennedy did with the moon, they can pick the one they’re passionate about and unleash it. We have 50 states. They can all choose their favorite of the 15. Maybe not talking to aliens, but whichever one they latch on to, it’s such a really great framework.

I’ll just literally list some of them. Doubling human lifespan is one. Ending hunger with synthetic food systems around the world is another. AI-empowered education for all at the highest possible level. High-bandwidth BCI—we’ve been talking about that on this pod for a while now. Demonstrating human mind uploads. I can’t wait for that. Plan B, maybe plan C—we’ll see. As I said, interspecies communication; understanding human consciousness. I think we’ve talked about that previously. Can we understand human consciousness? At which point, maybe we’ll understand consciousness for our AI systems as well. What have we dreamed about? Another one I love is disaster prevention and avoidance: predicting earthquakes and then preventing them, or tsunamis, as the case might be.

Peter H. Diamandis

These become natural XPRIZEs.

They’re what I call giga-XPRIZEs here. But I think one of the important things in this chapter is allowing people—in fact, demanding people—dream bigger than ever before, because the tools we have to solve the biggest problems are now epic. I think this, for me, is the most powerful part: the fact that you can say anybody has the agency now, leveraging these tools to go after what seem like impossible things and make them real. You’re only limited now by your imagination. And I think that’s—

And your compute budget.

Peter H. Diamandis

But that’s dropping 90% a year, so we’re in good shape.

That’s right.

Peter H. Diamandis

All right, “The Muddle Versus the Machine.” At first, Alex, when you proposed “muddle” as a term, I wasn’t sure I liked it. Now I love it. So describe what the muddle is.

Yeah.

The muddle is—the other term might be the bureaucratosaurus—that loves to measure inputs rather than outputs and slow down progress. The idea is, without properly shaping the charge of the intelligence explosion, the muddle is the end state that we find ourselves in when we're basically muddling our way through. That's one of the etymologies of that term.

What we talk about in this chapter, in a single sentence, is: What happens after we win? We paint a positive and non-dystopian view of, in particular, what human agency looks like. I made this short film, posted to social media, called A Nation That Learned to Sprint, depicting what life in the early 2030s might look like if everything goes well and we see GDP 2x-ing or 3x-ing year over year. What does a human, quote-unquote, job even look like in a macroeconomic scenario like that?

In this chapter, we lay out lots of new career opportunities that will be available to humans—at least un-AI-ed humans. Target designers, for example, or data rights brokers: people who are involved in shaping the targeting systems and shaping how we aim, fire, and verify superintelligence toward the hardest problems that humanity faces. This is going to be a growth industry from a job perspective.

Another point we make in the chapter here that's super important—we've discussed this before, and Salim and I've discussed this before—is that GDP is a terrible mechanism for measuring economic health. The paper proposes replacing GDP with something called the Abundance Capability Index, which measures a nation's capacity to solve problems rather than how much money changes hands. Again, as we look at benchmarks, as we look at rails and harnesses, understanding this is really important.

I think the challenge here, though, is that UBI, UBC, whatever you want to call it, is a great endpoint and a great aiming point. You want to have a target, as you say, Peter; otherwise, you'll miss it every time. The challenge is that moving from a welfare, taxation, labor-union structure to that is such a huge leap. I have no confidence in the public sector getting us there. So how do you navigate that? I think that's something worth exploring in the scope of your thinking, but that's a huge consideration.

Peter H. Diamandis

I was going to say, Salim, what a wonderful transition. Thank you. The last chapter—

Build the rails.

Peter H. Diamandis

Building the rails, Chapter 9—I think it's one of the most important chapters of the entire paper. Alex?

This chapter is where we lay out the answer to Salim's question. What's the so what, and what do you do if you're not running a nation-state? What can you do? How are you empowered to shape this transition, to shape your own moonshots, and to control your own targeting system?

We lay out various suggestions. If you're an investor, as indicated in the slide, fund the primitives, not the applications. There's so much infrastructure that can and arguably should be built out. If you're an entrepreneur, you should be building, picking your own targets with the targeting system. Create your own benchmarks and aim your own compute.

If you're an executive of a large company, you should be measuring the outputs, not measuring the inputs. Dave, I think you put it beautifully earlier in this episode, talking about the API-ification of large corporate boards and corporate governance. I think that's exactly the right playbook here, and the missing factor is having a benchmark to measure corporate objectives in such a way that the problem of corporate governance becomes a matter of maximizing the use of available scarce compute to maximize those KPIs and those evals.

In this chapter, we lay out, for a variety of different roles in the economy, what can you do? What can you in the audience do to help us achieve a utopian vision of abundance and post-scarcity, and excellent uses that are prosocial for superintelligence?

Peter H. Diamandis

The vision Alex—and, to some degree, myself, but I credit Alex—has laid out for the decade ahead is what's going to bring us to abundance. How do you do it? How do you lead as a leader, as an entrepreneur, as a CEO, as a governor? Where are we going? It's going to move much faster.

I think one of the points here, Alex, is that there's going to be such a distinction between those who do and those who don't that it's going to create a sort of 66-million-year-ago asteroid strike that's going to kill the dinosaurs and elevate the furry mammals. I say furry lobsters. Moving forward—

No, we love our lobster friends. He didn't mean that. Peter really didn't mean that.

Peter H. Diamandis

No, no, no. Elevate, elevate our lobsters. I would say that.

Elevate them into low Earth orbit.

Peter H. Diamandis

All right. A favorite part for all of us: AMAs. I'm going to keep us to one question per mate. All right, here they are. There are 9 of them. Let's see. Dave, do you want to pick first?

Sure. I like number 3 because it's such a happy answer. In a world with perfect AI output, will there still be a place for human spark in art and sculpting? Will handmade work have higher value or be buried in AI humanoid production?

I wholeheartedly believe it'll have astronomically higher value. Human touch will be so rare and so valuable, but also, the abundance of capital will be unbelievable. I expect artwork—you know, current artwork is one of the best investments you can make right now—but going forward, it is a category that will go up tremendously in value. People will appreciate all things human, whether that's human action, human sports, human poetry, human artwork, or sculpting. I expect it to be a rising area, for sure.

Peter H. Diamandis

I think that would be a great conversation. I'll call it a debate, but one of our next pods: What is going to be most valuable from humans in the future? Select one of these.

Let's see. I would pick number 5, which is: How is a young person supposed to earn an income when they compete against a model that costs $50 a month? That's from @clownpieceD.

It's a great question, but you're assuming the future is about competing with AI. It's about directing it, leveraging it, and amplifying yourself with it. In history, we've destroyed old jobs. We've created control points, and we've done orchestration. We've done intent. Winning isn't productivity; it's agency.

We talked about this earlier in the podcast. Knowing what to do and why it matters is more important. How do you mobilize intelligence at scale? That's really the biggest challenge. You can do that today in a way that you couldn't ever do before.

We've been doing workshops with teenagers and showing them how to use AI as a superpower to give themselves agency. I think that's where I would go with that.

Peter H. Diamandis

Alex, would you pick one of these?

All right. I like this assortment, so I'll pick number 8 for 100 trillion. Question number 8 is: With AI taking tasks we do ourselves, isn't there a risk we lose essential skills and become completely dependent on AI services? That's asked by Joroan Hoffs.

I want to invoke my friend John Smart. I hope you're listening. John has, I think, a brilliant dictum: The first generation of any new technology is dehumanizing. It takes away all your skills. The first generation of calculators takes away your arithmetic skills. The second generation is net-neutral to humanity.

The third generation—as another friend of the pod, Stephen Wolfram from Mathematica, says—gives you new superpowers and new skills. I don't accept the premise that there will be any sort of permanent loss of essential skills due to AI automation. I do think that there's a short-term substitution effect where AI drives down the cost of various skills or various tasks, but over the long term, I expect AI automation to be net-superhumanizing.

We're going to be capable of so much more with AI than we can do otherwise without it. I'll also say Vernor Vinge has written quite a bit about this. I definitely encourage everyone to read Rainbows End and Fast Times at Fairmont High, the novel and novella, respectively, that talk about this ad nauseam.

We're going to find ourselves in a very near-term future where, just like there are wilderness camps to learn how to survive without modern technological aids, we're going to start, I think, in our educational system—at least the better parts of it—having the moral equivalent of a wilderness camp for AI, where all of your AI tools get taken away. You have to do things manually just so that you at least have that skill set, and then you get all your AI skills back. Every fourth grader becomes a Nobel laureate.

Peter H. Diamandis

I love that. All right, I'm going to close this out with number 6. I use Claude daily. It fails at basic consistency. I think the question is saying: How can this be close to AGI when I have to check every output for errors? That's from MMGPT9ot.

I'm going to say again: AI is the slowest and most incorrect it will ever be. I know when I'm using my Claude bot or Claude 4.6, if I get something that seems off, I will ask it to check itself. Being able to use this in a recursive fashion is important.

Also, MMGPT9ot, we're in a period of recursive self-improvement. I think we're at the steepest part of the curve, and it's going to become more and more capable every day. The idea that we can use AIs to check AIs and, in fact, to do deeper reasoning is going to eliminate this very quickly.

CJ, thank you for this. CJ was on a Zoom AMA that Steven Kotler and I did for our book, We Are His Gods, and he actually wrote this as a result of that AMA.

All right. Thank you, CJ. On behalf of Skippy, my lobster, I’m sending you guys an incredible week ahead. Alex, it was an honor and a pleasure to work on Solve Everything with you. I’m excited to get it out into the universe. I think the value of steering people toward this accelerating time—and how they actually have the biggest impact on creating abundance and not the muddle—is critically important.

Agreed, Peter. It was a pleasure writing it with you as well, and I would encourage all of the humans and nonhumans in our audience to read it and let us know what you think.

Peter H. Diamandis

Yes, for sure. I’ve been here at a Tony Robbins event, and I would say probably 100 people have come up and said, “Oh my God, I love Moonshots.” And everyone: “I love Alex.” Alex, you’ve got fans here in Sun Valley. Dave—

How many of those people were human, Peter?

Peter H. Diamandis

They unfortunately were all human, at least for the moment. Yeah. All right, Dave. Thank you, guys.

If you made it to the end of this episode, which you obviously did, I consider you a moonshot mate. Every week, my moonshot mates and I spend a lot of energy and time to really deliver you the news that matters. If you're a subscriber, thank you. If you're not a subscriber yet, please consider subscribing so you get the news as it comes out. I also want to invite you to join me on my weekly newsletter called Metatrends. I have a research team. You may not know this, but we spend the entire week looking at the metat trends that are impacting your family, your company, your industry, your nation. And I put this into a two-minute read every week. If you'd like to get access to the Metatrends newsletter every week, go to diamandis.com/tatrends. That's diamandis.com/tatrens. Thank you again for joining us today. It's a blast for us to put this together every week.