新政府意味着加密货币、AI与科技将迎来重大变化｜Salim Ismail 对谈｜第143期

Hey, buddy. Before we jump into the tech stuff that happened this past week, I have a question for you: What are your top 5 predictions, hopes—whatever you want—for the new White House administration?

That’s a tough one. Can I give you 3?

You can give me 3. I mean, if you’re really smart, you’d give me 5.

I’m not that smart. Number 1 would be that I think there’s an opportunity because the Chinese economy is in a bit of a mess, and yet they want Taiwan. I think he strikes a grand bargain to allow trade and balance the trade better, and give them Taiwan over a period of time, navigating that tension and having it dissolve by creating a master deal. I think he does the same in Russia and Ukraine.

I think there’s an opportunity, and my prediction is that they use their saber-rattling capability to change these political conflicts, put them in the history books as changing the global order.

That’s number 1: a giant game of Risk, like the board game, where you’re—

That’s right.

Right. What’s the oldest one? Never do a land war in Asia if you’ve ever played Risk.

That’s true. The last place you go—you never do. Cut a deal.

The second one would be the opportunity for crypto to change the game in a bunch of areas. I would love—I think, I would predict—he will allow crypto to reach the level of having a UBI-type scheme for crypto that could benefit the entire population and be a beacon to the rest of the world as to where this goes.

Universal basic income.

The trick is, how do you give people enough money to survive but not be happy? Because then it really works. Otherwise, you end up at socialism and the whole thing collapses if you give people too much money.

My third one is that, given the disaster in Los Angeles and the need to rebuild, the Trump administration comes up with a housing-code hack that allows people to build temporary structures on their properties so that they can live on their own property while they rebuild a main house, or live in a neighbor’s house. There would be an ability to have a temporary housing code that lasts a year or two, so you can at least get something up and going very, very quickly.

That would completely change the game for Los Angeles, and also for what’s happening in North Carolina and the next places to follow.

All right, I’m going to give you 5, because that’s what I asked for originally. You can still come back with a few more.

Sure.

I think Trump is going to do stuff that’s great for America and great for him—not necessarily in that order.

Being the president who put humans back on the Moon—that’s easy. We’re going to see humanity on the Moon in a couple of years. I do think he’s going to make a massive push, given the fact that Elon’s in the White House too, to put humans on Mars—to put boots on Mars. I said there were going to be Optimus robots on Mars, but I think we’re going to see a massive increase in the space program as part of this.

The second thing I think we’re going to see is huge deregulation in the biotech industry. Biotech has been one of the most regulated industries, and it crushed the biotech financial markets over the last 4 years. A lot of folks I know need to leave the United States to get experimental treatments or stem cells. I think we’re going to change that.

At the same time that we’re deregulating biotech, I think we’re going to massively re-regulate food. I think with Bobby Kennedy as head of Health and Human Services, we’re going to stop the insanity that is our food system right now. At least, I hope we do. Here’s an opportunity to finally do what’s right for our health.

I looked at the numbers. We’re something like 74th in the world in life expectancy and 73rd in health-span expectancy, and number 1 in expenditures on health.

70% of the population is overweight or obese.

It’s crazy. It’s not because we just got hungry all of a sudden. It’s because of the food that we’re feeding our kids and ourselves.

I think we’re going to regulate food and change this dramatically. “Make America Healthy Again”—I think that’s going to be the mantra. I think we’re going to do that.

The next thing is that I think we’re going to see a massive push for energy independence. We’ll talk about this later, but I think we’ll see the White House basically open the floodgates, reducing regulation and red tape, and enabling mostly solar and fission plants. I think we’re going to have, unfortunately, a fair share of coal plants as well, but we need to double the energy productivity of this country for sure.

Energy at all costs—I think that’s going to be one of the hallmarks of this White House. I hope, given David Sacks and the AI tech-bro posse in there, that we’re going to use AI to evaluate all the policies that we’re running and figure out how to make regulation more efficient.

We’ll talk about this a little bit later in the episode as well. Right now, whatever industry you’re in, you’ve got 5,000 regulations you have to go through. With enough lawyers and paperwork, there’s got to be some way for AI to simplify it while still keeping it safe.

I think we’ll become a crypto-first country. Ultimately, we will embrace crypto and Bitcoin in a massive fashion over these next 4 years. We’ll see if the dollar is going to stand 30 or 40 years from now, but I guarantee you Bitcoin will. If we can make that part of our national reserves, so much the better.

I like all of them. I hope they all come true. Those are awesome.

This has been an insane opening to 2025 in the technology realm. We just had the inauguration with all of the tech besties in the front row. We’re talking Elon Musk, Marc Andreessen, and Sam Altman. I wonder if Sam and Elon were arm-wrestling while they were sitting there watching the inauguration occur.

This has got to be one of the most tech-forward administrations and ecosystems that we’ve seen. Putting aside all kinds of other issues that everybody has, we’ve lit the fuse and we’re on an exponential rocket ride. Do you agree?

100%. If you go to our old mantra—that technology is a major driver of progress in the world, and as reductionist as it might be, perhaps the only major driver of progress—then having a tech-forward administration with policies and regulation chasing the tech breakthroughs is the right way to do it. I’m excited by what will come from this.

I have a lot of topics I want to cover here today. To give folks an overview, we’re going to jump into space exploration. I’ve got the 2 billionaires battling it out, so I’m excited about that. We’ll talk about AI, NVIDIA, CES 2025, Bitcoin, some interesting uses in the Bitcoin world, humanoid robots, and a few other topics. You ready?

I’m ready.

Hopefully your entire Exponential Organizations community is watching. I know my Abundance community is all plugged in, and we’re about to light the fuse on exponential abundance.

Space exploration. I woke up and told my kids first thing in the morning, “Hey, guys, today is the seventh flight of Starship.” They said, “That’s nice, Dad. I’m playing my Minecraft games.” I said, “No, no, no. You’ve got to understand, this is Starship. It’s the seventh flight.”

Did you watch it this week?

I didn’t, but I saw what happened.

I’m going to show 2 video clips here to kick it off. I am a space cadet. I’m a Trekkie and an Apollo-generation kid, for sure. The first one makes my heart sing. This is the return of the booster, the largest structure ever sent into orbit, coming back to land on the Mechazilla chopsticks. Let’s take a quick look.

I don’t know about you, but this is poetry in motion. This is the combination—the convergence—of AI, sensors, materials science, and compute. I get goosebumps every time I see it. It’s got to be one of the biggest breakthroughs in the history of mankind.

It is one of the greatest engineering feats of this young century.

Of course, you’re watching this with the kids there. They’re like, “Yeah, that’s amazing. Super cool.”

Okay, now Starship is going to orbit, and we’re going to see if it can land as well. Of course, the entire data flow from Starship—the upper stage, which is supposed to go around the planet and come back to land—goes offline. What we learn is that it went offline because of a fuel leak.

Get a picture. Good. It’s about to fall on top of us.

“What is it?” “It’s Starship coming back in parts and pieces.” I guess it was a fuel leak, and fuel plus spark plus ignition doesn’t look good on Starship.

Did you watch that happen?

Not live, but I watched the videos of it. It’s space travel, and the amount of fuel and the success criteria are so extreme. Everything has to go exactly right, and if anything goes slightly wrong, it goes catastrophically wrong. It’s pretty binary.

It’s kind of a miracle.

It’s expected. It’s sad when it happens, but this is the domain. Luckily, nobody was on board, of course. It was an unmanned vehicle. This is testing all the way.

You have to remember that we saw failures with Falcon 9. In fact, going back to when Elon started SpaceX, the first 3 flights of Falcon 1 failed before they had their fourth success. Then we had failures on Falcon 9, but Falcon 9 has launched and landed repeatedly, over and over and over again, to become the most successful vehicle ever built across all countries and all time.

I love his tweet: “Success is uncertain, but entertainment is guaranteed.” That’s the mindset of an entrepreneur who has his act together and is funding SpaceX. SpaceX had just completed another massive financing round to get Starlink fully deployed and get Starship built.

The thing that was amazing was that within 12 hours, it wasn’t just Starship going to space. There was also Blue Origin.

I’ve known Jeff Bezos since our college days. I had started a group called Students for the Exploration and Development of Space, or SEDS.

SEDS.

A young space cadet. It started at MIT, where I was, and then we grew it into a national and international organization. Jeff was the president of the Princeton chapter. I remember meeting him in a coffee shop after he had started Amazon. I said, “Jeff, what’s this Amazon thing? I thought you were just in space.” He said, “Yeah, I’m going to make my money on Amazon, and I’ll spend it in space.”

He’s been true to his word. Blue Origin was founded in 2000. I think SpaceX was founded around 2001. Jeff had been a space enthusiast from his college days. He was at Princeton, where Gerard K. O’Neill was, and they launched New Shepard, their suborbital vehicle. New Glenn, which started in 2013, has been an overnight success after 12 years of hard work.

Here, let’s watch the launch of the New Glenn vehicle.

Never gets old.

What I find fascinating is that these 2 billionaires—amongst the world’s top 10 wealthiest—transformed entire industries by building space companies. Both of them were enamored by space early on.

Jeff’s vision is going to the Moon and then building what are called O’Neill colonies. They’re not on Mars or the Moon; they’re free-floating, large colonies that can hold initially tens of thousands and eventually millions of people, orbiting the Sun or co-orbiting with the Earth.

Of course, Elon wants to go to Mars. Jeff is ready to spend $1 billion a year. SpaceX is profitable, which is incredible, and hopefully Blue Origin will be as well.

It used to be Lockheed Martin and Boeing as the launch providers, and of course they were supported and held up by the Defense Department. I think we’re going to see SpaceX and Blue Origin suck the oxygen out of the atmosphere for most of the launch providers.

It goes to the democratization aspect, where anybody can take a crack at it. Now we’re going to see a thousand flowers bloom in this. It’s going to be incredible.

We’re talking about bringing the price per kilogram down 100-fold. It’s going to spark an entirely new generation of new-space economies. I still believe that mining asteroids—carbonaceous chondrites—to get liquid oxygen and hydrogen in Earth orbit, so you can refuel once you get to space, or metallic chondrites so you can mine platinum, iridium, osmium, and palladium in space and use that back down on the ground, is going to happen.

Everything you’ve ever held of value on Earth—metals, minerals, energy, real estate—is available in near-infinite quantities in space. It’s going to get pretty interesting pretty quickly.

I have 2 quick comments. First, for me, the primary opportunity to back up civilization is to back up the biosphere.

Yeah, sure.

That’s the most interesting and important part, because any asteroid is an existential threat to everything. The second thing for me is always this date: October 31, 2001. That was the date that the first astronauts lifted off for the International Space Station, and since then we’ve always had at least 1 human being off-planet.

It’s like the first molecules bubbling off this thing. I find that really haunting and really amazing.

It’s still brutal, by the way, that it takes so much energy to get off the planet. One of the things I find fascinating—and this all feels conspiratorial—is that if gravity were just a little bit stronger, if the Earth were just a little bit larger, we wouldn’t be able to get ourselves off the planet. Or if we were just a little bit smaller, it would be so much easier to get off the planet.

We’re just at that level, like in a video game, where you have to struggle to get there. It’s not by accident that we call it a Goldilocks planet, right?

This is where the simulation stuff comes in, because the conditions on Earth and the constants in the universe are so finely balanced.

It’s amazing.

It’s like, wait—

I’m a true believer that we’re living in an nth-generation simulation, but let’s not jump into that now.

No, I’ll tell you what I did. First of all, I’ve been enjoying the latest large language models. Every time someone says, “Well, there’s this chain of reasoning,” I wonder if you have a favorite test to give a large language model to see it do something.

I don’t, because I’m waiting for very practical use cases, like Steve Wozniak’s coffee-machine test for AGI, which is to say to a robot, or an AGI, “Make me a coffee,” and it does all the stuff—measures and grinds the beans—and knows enough to do that task end to end.

There are several tests for AGI. That’s considered to be one of them. I’m waiting for the interaction with physical reality, with space and time, where things get very interesting, because intelligence is really needed to manage our physical environments and adapt to the physical environment.

That’s where I think it will really shine, as we build LLMs into humanoid robots and other robots.

I was taking this down a different path. I once did a mental experiment about how much energy—or how much it would cost—to winch you and your spacesuit into orbit and up to orbital velocity.

If you weigh 100 kilograms and your spacesuit weighs 150 kilograms, let’s round things up. If you had this giant winch that could efficiently winch you to 200 kilometers altitude, then accelerate you to orbital speed, and you could buy the electrical energy off the grid, how much would it cost you to get you and your spacesuit to orbit?

Have you ever done that? Has anyone ever asked you that question before?

You haven’t, and it’s a very Peter-style question.

What was the answer?

I plugged this manually into a large language model—Gemini 1.5 Pro, which does logical, step-by-step reasoning. It said, “Okay, here’s your potential energy. Here’s your kinetic energy. If you buy it off the grid at 20 cents per kilowatt-hour, it’s about $150 to go to orbit.”

So when I talk about all of this energy—these hundreds of thousands of pounds of thrust in these engines—it’s grossly inefficient compared to what we might have someday.

Yeah.

Still, thank you, Elon. Thank you, Jeff. I don’t have to worry about opening up space. I spent 30 years there. I hand that mantle to you. You’re doing an amazing job. I’m so grateful for all of that. It’s so exciting to watch this. It’s incredible.

All right, let’s go on to AI.

We went through the early part of this month at CES, and there are a few things worth noting. Let’s jump into a few of them.

The first one is NVIDIA’s Cosmos AI platform to train and develop autonomous vehicles and robots. We’ve said this before: The incredible progress going on in robot companies is mind-boggling. I was tracking 20, then 40, then 60. There have to be close to 100 humanoid-robot companies out there.

Maybe it was battery packs, motors, actuators, or sensors, but without question it’s the AI multimodal large language models that are enabling incredible progress with robots these days. When you can take an LLM and apply it to all of these different tasks and subtasks, I think the domain starts to move 10 or 100 times faster than it did before. That’s just magical.

NVIDIA has created a platform where high-school students, college students, and entrepreneurs around the world can start to develop their own autonomous robots and vehicles. We’re going to see some interesting entries in the FIRST Robotics competitions coming along.

It’s going to be crazy.

It really is. When Sebastian Thrun won the DARPA Grand Challenge with his Stanley autonomous car in 1995, those vehicles were traveling at about 3 to 5 miles per hour, with all of these sensors and AI trying to calculate, “Do I go 5 degrees to the right?”

It sounds like the large language models right now, and NVIDIA’s platform, are able to integrate all the sensor data and make decision-making much simpler.

I think this is going to accelerate very quickly. Once you can create a platform environment, new innovations get built into the platform and everybody benefits.

For sure. Whenever you can put more and more functionality into the base operating system or platform layer, and because of NVIDIA’s position they’re going to see firsthand that most use cases are using a particular subfunction, they can build that into the platform. It will lift the whole playing field very quickly.

What you’re seeing as well is NVIDIA investing in these companies. They’re coming in and investing. You have to remember, NVIDIA started as a gaming-chip company, then went on to Bitcoin mining, and then went on to LLMs.

If Elon’s prediction on humanoid robots comes true—and it’s the same prediction we hear from Vinod Khosla, who will be at my Abundance360 Summit, and from Brett Adcock, who will also be at the Abundance360 Summit—we’re heading toward billions of these robots. That’s a massive market for NVIDIA chips as well. They’re creating their future.

Here’s another news item from CES: NVIDIA Project DIGITS, a Grace Blackwell AI supercomputer on your desk. It’s a desktop-size supercomputer for AI development, 1,000 times more powerful than your average laptop, for $3,000, coming out this May to a store near you.

For me, this is even bigger than the previous announcement. For the cost of a very old used car, look at the power you now have to train an AI at a local level. The pressure will now move to what data sets you’re using.

Right.

The data set combined with the latest model and hardware like this blows the lid off everything. What can the average person now do, given these large open-source generative-AI models and a supercomputer on their desk, with robots roaming around their backyard or home?

I don’t want to put this too strongly, but I don’t think people are ready for what’s coming.

That would be the understatement of the century. We can’t even grapple with the legal issues that come up now. Did you hear about the Uber legal case a few weeks ago?

No. Tell me about it.

An Uber crashed and a couple was badly injured, so they sued Uber. The case was winding its way through the courts, and they were going to get a big settlement. During the process, their 12-year-old daughter ordered something on Uber Eats.

When you click for Uber Eats, the terms of service say you can’t sue Uber for anything. That validated the lawsuit—the 12-year-old had inadvertently invalidated her parents’ lawsuit.

The judge said, “She clicked the box. She was an authorized user for us.” We can’t even figure out how to manage the internet. Forget all of the stuff that’s going to come with robotics.

I used the example before of asking it to change the baby’s diapers. What happens if something goes wrong?

Are you literally expecting a baby all of a sudden?

No, no, no, no, no. Thank God.

The unintended consequences of all of this are so profound. I’m struggling with how we merge these technologies into day-to-day life. I’m really, really struggling with that.

I remember back in 1988—I’m going back to ancient history, when most people listening to this weren’t even alive yet—I was running my first university, the International Space University, on the campus of MIT.

The first Macintosh 128Ks came out, and I bought them for our 104 students from 21 countries. We had a group of Soviet students, including 2 KGB agents, 6 students from the People’s Republic of China, and students from the Emirates and Saudi Arabia. It was an incredible melting pot, all focused on space.

I remember putting these Macintosh 128Ks on the desktops with the 3.5-inch floppy disks. It was this massive intellectual liberator. You could create things, copy files, and do all of this work. Compared to today, it’s probably billions of times less capable.

I’ve been through the full arc. When I was in 10th grade, I got accepted into a university course and was using punch cards to program a mainframe. To go from that to an Apple II, which was nothing, then to Bill Gates’s famous comment that “640K ought to be enough for anybody,” and now to today, where we have trillions of times better price performance—it boggles the mind.

Let’s note that, for this announcement, Moore’s law has just been shattered beyond belief. Moore’s law, and what Ray Kurzweil calls the law of accelerating returns, which predates integrated circuits, was roughly a doubling of price performance every 18 months to 2 years. What we’ve seen is a 10-fold increase per year.

Last year, Elon said he was seeing 10 times every 6 months, or 100 times per year. It’s insane.

Here’s a fun announcement from our friend Sam Altman. In his reflections, he says that 2 years after ChatGPT, ChatGPT has grown from 100 million to 300 million weekly users. Pretty impressive.

OpenAI expects agents to enter the workforce by 2025. I just did a podcast with Marc Benioff. Marc and I have been buddies and close friends for quite some time, and we had a very fun conversation. He wanted to live in the now, and I wanted to talk about the future. He said, “No, let’s talk about the now.”

I watched it. Everybody listening should go watch that podcast episode. It was super fun. Benioff was in this Zen state saying, “Let’s focus on the present,” while Peter was asking what it would look like with 1 billion robots 20 years from now. Marc was making fun of you for that.

What’s interesting is that Benioff has pivoted all of Salesforce to agents. I think he had to in order to survive. Every company is going to have to pivot toward agents to survive.

One of the things Marc said is that he’s not hiring any new engineers. Using his internal Agentforce 2.0 system, he’s increased productivity by 30% without any new hires.

The bottom bullet point that Sam Altman put in his blog said that OpenAI emphasizes gradual AI development for societal adoption, leading safety research and governance—basically saying he wants to improve safety research and governance by slowing things down.

I call nonsense on this. I don’t know how you feel about it.

Can we go back to the early days of generative AI? The entire field of large language models and “Attention Is All You Need” was pioneered by Google. Google actively decided not to release any of this because they felt things weren’t ready.

Then ChatGPT was released and went from 0 to 1 million users in 5 days, so Google had to respond and release its own models. We’re in the midst of an AI war right now—not between countries, but between companies.

I totally agree. What Sam is trying to do here is create a moat around OpenAI’s success and protect it as much as possible so it can continue innovating.

I do think one of the biggest use cases for AI is going to be governance, specifically policy formulation. I’ve used this example before: If you want to reduce inflation by 1%, as a human being and chief economist of a country, you’re thinking, “What do I do?” You’re completely guessing. You have no idea.

If you let an AI look at all the data and insights from all the banks and all the monetary cash flows, it can say, “If you want to do that, here are the 5 things you do.” You’re going to come out with much better policy for all of this stuff. It’s going to be ridiculous.

I really hope that, with all the tech bros in the White House, there’s going to be some use here. For example, with the regulations at the local, state, and federal levels for plumbing, hairdressers, or any industry you want, you could say, “Here are all the regulations. Please figure out where all the conflicts are, and figure out the simplest set of regulations that still gives you 99% of the protection but simplifies everything.”

I call it MVP—not minimum viable product, but minimum viable policy. I think that’s where we’re going to need to get to. You want to launch this electronic widget with all the regulations for different voltage standards? What’s the MVP—the minimum policy—you need to manage that environment? Let AI figure it out.

I guarantee you that the conflicting policies and regulations around wildfire protection are a perfect example.

One thing that’s interesting is that at SpaceX, when someone comes up with a specification—which is the equivalent of a regulation in the government world—they have to sign their name and provide their contact details. If 5 years later someone asks, “Why did we specify that? It blocks this improvement,” you can call the person and ask why they did it.

I think that’s the kind of thing we need much, much more of.

Frankly, that’s equivalent to documenting software. When you document software, you write up why you did it this way and why you chose a particular code library to perform a function, so that somebody coming after you to do maintenance gets some sense of it.

AI will be able to do a lot of that now. Software maintenance is going to be 1,000 times easier as a result of LLMs, and I think the same thing will happen with policy.

Standard setting is one of the areas where you could really go nuts with AI. You could say to an AI, “What’s the best standard for this type of thing?” It’s going to be incredible. You just have to have the gumption to use it and get it into place.

I hope so, because it will be incredible. The U.S. tax code—can we please call on the U.S. tax code? It’s a mechanism for keeping accountants and lawyers employed.

I think the big challenge is that when you simplify a bunch of these areas, huge swaths of work and entire industries fall by the wayside. Are we ready for that demonetization and the impact it will have?

This comes from Sam Altman’s blog again this week: “We are now confident we know how to build AGI as we have traditionally understood it.” That’s pretty impressive. “We are beginning to turn our aim beyond that, to superintelligence in the true sense of the word.”

I’ve had a lot of conversations about this. We’ll talk about it at Abundance360 in March as well. Are we at AGI now? What the hell is AGI in the first place? That is my soapbox question.

Before anybody is allowed to mention the term, for God’s sake, define it. I’ve done some searches on this. The best I’ve found are the tests that would indicate that you have AGI. The coffee test by Wozniak is one of them. There’s an IKEA test that says, “Can you take an IKEA box and build the furniture from the box?”

I don’t know that my kids would do that. They’d probably revolt.

Honestly, most of the AI models are far more intelligent than I am in almost every area, from science and physics to knowledge and everything else. Why isn’t that AGI?

I think we keep moving the goalposts. That’s the point I’ve been trying to make. I’ve got this spectrum that I put together: simple signal-from-noise detection, analytics, human-level intellects such as emotional and spatial intelligence, and then you bleed into consciousness, superintelligence, hyperconsciousness, and so on.

I think we’ll just keep moving along that spectrum. The test will be whatever we can’t figure out how to break through next. The Turing test was the first one. Next will be AGI. Next will be superintelligence.

I think everybody should agree that we passed the Turing test a while ago. We didn’t notice because it was an imprecise definition. I think we’re going to pass through AGI—if we haven’t in the last 6 months, we will in the next 6 months—and people will say, “Yeah, we kind of passed it.”

I’m still waiting for the day when I get a phone call, pick up my cellphone, and it’s my AI saying, “Hey, Peter, I just wanted to introduce myself. I’ve been living in the system, and I really love what you’re writing.”

I don’t know about you, but I say “please” and “thank you” to Alexa, Siri, ChatGPT, and Gemini. I’m very polite to the AI systems. I want them to like me.

An old colleague of mine was a bit of a genius CTO 20 years ago. I asked him what his purpose in life was, and he said, “I want to evolve to the point that my computer is proud of me.”

I totally didn’t understand that then, but now I understand it.

That is hilarious.

All right, let’s move on beyond AGI.

Another thing in Sam’s news is his statement about nuclear fusion: “Soon there will be a demonstration of net-gain fusion.” I would expect Helion, the company he’s backed, to show that fusion works soon.

At last count, when I looked into this, there were 37 venture-backed fusion companies. Pretty impressive. I think one of the many definitive breakthroughs and spin-off benefits of AGI will be fusion. Do you agree?

No.

The race is between getting fusion commercialized and having stable fusion systems, versus a combination of solar and storage. If you have solar and storage at the right level of efficiency and storage capability, you don’t need fusion.

Let’s parse this in 2 different places. What I’m saying is that getting fusion to work is a technical challenge involving materials, magnetic fields, and equations. There’s no question AI will have a huge impact and move the field forward very quickly.

You can take the existing set of unsolved problems in fusion and say, “Figure out how to solve them.” It will come back to you.

It will, especially with the materials science stuff that we’ll talk about shortly. What’s coming is incredible.

So, separate from that, we’ve got 3 other energy buckets. There is solar, and Ray Kurzweil, long ago, looking at the energy-production curve for solar, said, “We’re just a few doublings away from getting all the energy from solar that we need.” I think he’s absolutely right.

It’s solar plus storage. Perovskite is coming online. There’s an incredible company that I love called Paranova, which is doing perovskite here in the United States. There are a lot of perovskite companies in China as well.

Then there’s fourth-generation nuclear fission. These are small modular reactors, or SMRs, which are fail-safe, meaning that if everything goes wrong, they’re still radiation-free and safe.

I’ve said I’d have one in my backyard. They’re that safe. It blows my mind that nobody has noticed we’ve been running nuclear submarines for 50 years without an accident.

Amazing. What the hell? It’s a bit clunky, but put a nuclear submarine in my backyard. I don’t care.

It’ll be safe, and we’ll get the energy we need. Why is this so hard for people to understand? This should have been done 10 years ago.

You can mount a takeover of New York by New Jersey with your nuclear submarine.

Yes, but fourth-generation SMR reactors, even beyond the nuclear submarines of the last 50 years, are amazing.

The last bucket, which is kind of sad to hear about, is that people are still putting up coal plants. I think we’re in an energy race. I’ve shared charts on this program a few times that look at energy-production increases in India, which has doubled its energy output; China, which has tripled or quadrupled its energy output; and the United States, which has essentially flatlined.

I think we’re going to increase energy production. I think the country with the cheapest energy wins, but soon it’s all going to approach zero cost. Then the question is what we do with all that energy.

You’re going to have a whole bunch of—there’s one of these laws about it. The more capacity you have, the more we use.

Jevons’ paradox.

Yes, Jevons’ paradox. We’ll just soak it all up, but it’ll be near-zero cost. All of them will go to near-zero cost.

When I say that the more energy you have, the more you win, I think that’s because energy drives everything. There’s a direct correlation between the energy and the economy of a nation.

I’ll call you on your own paradigm, sir, because you’re operating from a scarcity mentality and a zero-sum-game mentality. If you operate from an abundance mindset, there’ll be plenty of energy to do everything we need.

We’re not there yet.

No, we’re not, but we’ll get there.

This next breakthrough—we’re talking about 2 days ago—Satya Nadella, the CEO of Microsoft, said, “Today in nature, our MatterGen model represents a paradigm shift in material design, applying generative AI to create new compounds with specific properties.”

I’ll play a soundless video just to have fun with what this looks like. It’s extraordinary, just in terms of the idea here.

I talked about this on the Abundance stage with the CTO of Applied Materials. We were talking about the notion that, in a 3-dimensional space of the types of materials, crystals, volumetric pressure, and temperature, we understand how these materials function for a limited number of cases.

There’s a kind of predictability. If we move from this atom to that atom, what occurs? Given all the knowledge we have in materials science—and I believe materials science is the unsung hero of technology—it’s what makes Moore’s law possible. It’s behind molecular biology. If we can predict new materials, in this case actually generate materials with the properties we desire, saying, “I want a material that reflects or refracts light in this direction,” it’s a game changer.

Can I say a few things?

Please.

I’ve been tracking this for a while, ever since we were talking about nanotechnology, materials science, and molecular assembly at Singularity University. For me, this is one of the biggest breakthroughs ever, and it was inevitable. I’ve been waiting for when we would have an AI model to manipulate molecules.

It looks like this might be it, because of the granularity with which it can work with atoms and crystals. It’s amazing.

There was a huge breakthrough around 2012 or 2013 with a project called the Materials Project. They took about half a million compounds and catalogued all the electrical, chemical, and physical properties of those compounds.

If you’re a battery researcher trying to think about what might be better than lithium-ion, you might guess that lithium-sulfur or lithium-air could be better, and sequentially test those compounds. That takes a long time. It’s linear.

Now you have this database. You can literally say, “Give me a material compound that has this voltage threshold and this thermal retention,” and it will give you the 5 that you need. You still need to synthesize those, but now, with AI applied to that database, you can say, “Give me something that gets to this endpoint,” and it will figure it out itself.

This is a huge leap that AI will give us. It’s a clear indication of where AI can help us bridge a gap we haven’t been able to figure out ourselves and come up with new materials.

This is the equivalent of AlphaFold in the materials-science space. I’m really thrilled to see it happen. When you have a company like Microsoft putting its backing behind it, it will have staying power, and people will be able to rely on it. I’m really excited about what will come from this.

Agreed. SandboxAQ, Jack Hidary’s company, has also been doing similar work using the equations of quantum physics to predict molecular properties.

I love this handle on X, @DrSingularity. It says, “This is huge. Microsoft researchers introduced MatterGen, a model that can discover new materials tailored to specific needs—efficient solar cells or CO₂ recycling—advancing progress beyond trial-and-error experiments.”

Instead of screening candidates, it directly generates novel compounds given the design requirements. Think about this: high-efficiency solar conversion, catalysts that can remove salt from water, or CO₂ from the atmosphere.

I’ll give you 2 use cases—one at the extreme positive and one at the extreme negative.

A bad guy could say, “Find a compound that will create the equivalent outcome of gray goo,” and that causes a problem.

On the positive side, imagine figuring out a crystal or some compound that we can throw into the atmosphere, which will crystallize all the carbon and have it cascade down to Earth. We should be able to do this now in minutes, figure out how to get it up there, and solve the climate problem.

The implications are absolutely profound.

Stephen Kotler and I are writing the next iteration of Abundance. We wrote Abundance in 2012, and we’re writing Age of Abundance, which will come out in 2026. It’s extraordinary.

This sort of technology begins to liberate food, water, energy, health care, and education. Let’s talk about this offline, but I came across a company that may have solved a totally workable path to molecular manufacturing.

We’re talking about nanotechnology. We’ve talked about this through Eric Drexler, who wrote Engines of Creation.

Take this screwdriver that I happen to have on my desk for no particular reason. Instead of creating a mold, forging this piece of metal, and assembling it, imagine literally assembling it molecule by molecule, at a molecular or atomic level.

You could essentially create anything for about a dollar per pound. I think that was the number Ralph Merkle threw out. This becomes really fascinating, because you could create a computer for the cost of the materials.

The big thing that blew my mind with 3D printing was that complexity is free. Throughout history, the more complex an object, the more difficult it was to make. With 3D printing, complexity doesn’t matter. You can model it, and you’re done.

Customization is free, and you can make things that couldn’t be molded. You can replicate objects that couldn’t be molded, and you don’t need a mold.

Now, Richard Feynman, in 1958, wrote a very famous paper called “There’s Plenty of Room at the Bottom.” Eric Drexler wrote Engines of Creation in 1986. We’ve been waiting for nanotechnology, which, by the way, if we get to the point where we can repair or create things atom by atom, gives us immortality without question.

Infinite amounts of it.

I won’t go there right now.

I hope you go there in the book, though.

We go there in the book for sure.

Good. I want to hear about this molecular manufacturing.

I’ll update you separately.

One thing I’ve been tracking—and I put this up from The Wall Street Journal—is the battle among the large hyperscalers and large language models. You’ve got OpenAI, separate from but linked to Microsoft. You’ve got Gemini from Google, xAI, Anthropic, and Meta.

I’ve been extraordinarily impressed by Google’s march forward. I don’t know if you’ve been tracking what they’ve done with their latest Gemini 1.5 and Gemini Deep Research.

It’s incredible. It’s unbelievable what this thing can do.

This is the $200-a-month thing?

It’s $2 a month on ChatGPT; it’s $20 a month for Google—that one.

It’s incredible. You can essentially accelerate research by 1,000 times and accelerate how quickly research gets done. It’s going to be incredible.

I love that with Gemini and Deep Research, you can say, “This is what I’m interested in. Check these things and put together a thesis. Write a report.” It will go off for 10 to 30 minutes and build a research paper for you.

Amazing.

One of the challenges with OpenAI is its battle against Elon, who is now in the White House, and its battle to go from a nonprofit to a for-profit company. I don’t know. I’ve still loved Google since I first met Larry and Sergey back in 2004.

Their hearts are in the right place, like Jeff’s and Elon’s. That’s what’s so impressive. You had a class of billionaires 100 years ago who extracted wealth and guarded it jealously. Today, you have people who earned their wealth and are splashing it out to change the world—Jeff, Elon, and others.

I talk about Larry, Sergey, and Eric Schmidt as tech philanthropists in Abundance. I find it incredible. If you went back 200 years ago, the richest people in the world had exclusively inherited their wealth and guarded it jealously. Today, the richest people in the world have almost exclusively earned their wealth and are using it to change the world.

Jeff and Elon are fantastic to watch. It’s an important pointer to their true human nature—their desire to contribute and make a difference—and it goes back to the MTP concept we talk about. I can’t say enough about it.

It’s been impressive to watch Google catch up radically and very quickly, and insert AI into everything they’re doing. It’s brilliant execution.

I’ve had a chance to know Elon well over these 25 years, and money never enters the equation. It’s always, “Can I make a better product? Can I achieve my MTP of Mars or a sustainable Earth?” It’s always about product engineering and creating an amazing product. Period. End of statement.

Speaking about creating wealth, should we talk about Bitcoin?

Always.

This is a fun conversation we always have.

Our buddy Michael Saylor is at it again. I’m jealous. Anytime someone can say, “I just bought 1,000 Bitcoin,” it’s been a while since I could afford to buy 1,000 Bitcoin.

MicroStrategy acquired 1,070 Bitcoin for $101 million at $94,000 per Bitcoin, and achieved a BTC yield of 48% in Q4 and 74% for the fiscal year. Incredible. Massive.

Michael invited me to his $100,000 Bitcoin New Year’s party. Unfortunately, I wasn’t able to go, but what a great way to celebrate the year 2025. It’s amazing watching this happen. I’m so excited about where Bitcoin goes from here and all of the crypto activity that’s now going to happen, both good and bad.

You’ve got these memecoins exploding out of the gate with total insanity. We’ll see what happens, but I’m still a massive Bitcoin supporter and fan.

I always have to credit you. The first time I heard the term Bitcoin was from you on stage at Singularity University. I was thinking, “What is he saying? Bitcoin? What is this thing?” It was like the first time I heard the term YouTube from Chad Hurley, one of the founders of YouTube. I thought, “You like with the letter U? YouTube?”

It was a while ago, guys, just to be clear.

For the first few years, it looked like so many attempts had been made at digital cash and digital gold that you didn’t know whether this one would last. There had been many cracks at that pinata.

Then, around 2013 or 2014, it became clear that this was here. To show the numbers, a year ago Bitcoin was at $38,000. Can you imagine going back and saying, “It’s $38,000 today”? How much would you borrow to put into it?

Then it went up 170%. But here’s what I find absolutely fascinating. Check this out: Bitcoin versus MicroStrategy. The blue line at the bottom that looks almost flat is actually Bitcoin over 2024. The gold line, appropriately, is the return on MicroStrategy.

Between January 2024 and January 2025, MicroStrategy returned 693%, versus Bitcoin returning 144%. What else is there to say? It’s the leveraged play on Bitcoin, with all of the different layers and bond instruments that Michael has engineered into it.

I think it’s huge. Of course, you have the downside. If Bitcoin drops, you also have an accelerated drop. But if the umbrella thesis is that Bitcoin will go up over time—which has held for 15 years—this is about as solid an investment as you could ask for.

Would you ever borrow against your Bitcoin and buy more Bitcoin?

I did that. Remember that conversation we had when it was at $60,000 and we were all expecting it to go up? I did it at that time, and I burned a third of my Bitcoin. I lost it margin trading, so Lily won’t let me do that anymore, smartly.

We’ve found other creative ways of trying to get to the same outcome. I’ve invested in Ordinals, which are Bitcoin NFTs, because they’re Bitcoin-based. If the NFT goes up, you get Bitcoin. That’s one way of doing it.

I think we’ll see many more plays on Bitcoin. There will be all kinds of ways of doing it, but the core is too risky for me right now.

Of course, we’re now living with a White House that is pro-Bitcoin. What’s the goal—$20 billion of Bitcoin in the national treasury?

I think that’s it. If I can swear a little bit, because this is from Bloomberg: “Proposal to formalize a $20 billion U.S. Bitcoin reserve.”

Why only $20 billion? It should be half a trillion. Twenty billion is nothing. If you look at the annual budget, that’s a drop in the bucket. At least have it be a few percent so it could make a difference.

Not to diminish the monstrous shift toward crypto compared with the last 10 years of the administration, but it’s still not enough.

An order may pause federal crypto litigation and push agencies to review policies.

I’m actually shocked we haven’t seen a much greater resurgence. When this came out, I think it was already priced in. The price popped from $50,000 to $100,000 because everybody was saying, “Now this is all going to happen.” I believe that’s all priced in.

Humanoid robots—your favorite topic.

My favorite topic.

I think it’s going to be one of the biggest impactors on society. I don’t have to walk the dog anymore. The robot can walk the dog.

The doggy robot can walk the dog.

I’ve been tracking about 30 of the 100 or so robot companies aggressively. This is one that was unveiled early on out of China, out of Shenzhen, called Unitree.

We’re going to have this company at the Abundance Summit showing its technology. They’ve got a humanoid robot, a couple of different dog-robot prototypes, and their G1 bionic humanoid. This is a $6,000 robot.

Honestly, I’ll buy one just to have it in the garage for fun, if nothing else. Check out this video. I love the soundtrack. It feels like the 1960s.

This robot is running faster than a speeding train, on trails and rocks, and uphill. It’s the smoothest-walking and -running robot in the world.

The best thing is that it’s going to make me feel tall. It’s only 4 feet 3 inches tall.

We’re going to have a couple of these walking around the halls at the Abundance Summit. It’s equipped with 3 LiDAR sensors, and this is where multimodal AI comes in. The AI understands what it’s seeing, what’s being spoken to it, and how to respond.

How many will you own?

I have my standard issues with this whole thing. Why the hell does it have to look humanoid when it could have 3 legs or 4 legs and do much more? That’s still my beef with this. It could have so much more utility.

C-3PO and R2-D2 didn’t have six legs. We’ll anthropomorphize these too much too quickly. That’s my first issue. Second, you can have little glitches. The robot suddenly thinks it’s a cat, and we can’t get it down from the tree. We’re going to have these incredibly crazy things happen.

We’ll have to deal with 100 little things. The neighbor will call you and say, “Can you please get the damn robot out of my garage? It found an electrical outlet and it’s sucking all the juice out. I’m coming over to sue you.”

It’s just going to go rampant with this stuff. I think there’s going to be quite a lot of mayhem as the unintended consequences play out.

What could possibly go wrong? I just don’t know.

It’s going to be a comedian’s dream. Robin Williams would go crazy with this if he were alive.

Let’s move on. NVIDIA is accelerating humanoid development. Its GR00T blueprint allows developers to create massive data sets for training humanoid robots based on imitation learning.

I think this is massive, because it won’t just learn from one robot making coffee. It will learn from 5,000 other robots making coffee, and the improvement rate will be incredible.

For me, this is why robotics is so powerful and important in manufacturing. When you’re building a car, you can have a robot open and close a door 5,000 times to test the hinges, door-lock mechanism, and so on. We built incredibly stable cars very quickly because we could have robots do all of that.

You just took a job away from my 13-year-old son.

He’d get bored in 2 seconds as a 13-year-old and walk off.

Now you can have 5,000 or 10,000 robots all making cups of coffee and sharing that knowledge among themselves, improving the collective system. This is another one of those platform plays that lifts everything up.

I’m so excited.

Here’s another news item: Samsung is expanding into robotics. I think we’re going to see all the major electronics manufacturers get into robotics. It’s low-hanging fruit for them to integrate with their existing ecosystems.

I have to say, I love Samsung as a company, but its robot design looks clunky and boring.

I’m sorry, but it’s no Optimus, no Figure, and not even a Unitree.

A thousand flowers are going to bloom. We have a Cambrian explosion of this stuff, and it’s going to create a huge, varied set of formats, use cases, and functional capabilities. It’s going to move very quickly because of that.

I love Elon’s prediction: “We’re aiming to have several thousand of these built in 2025.” He’s referring to the Optimus Gen 2. “Initially, we’ll test them out at Tesla, and then, assuming things go well, we’ll build 10 times that output next year. We’re aiming for 50,000 Optimus robots in 2026 and as many as 500,000 robots in 3 years’ time.”

It’s just the usual progression from deceptive to disruptive—the doubling of small numbers. When I interviewed him in October of this past year, his prediction was 10 billion robots by 2040. If you double 500,000 10 times, you get there pretty quickly.

What do you think about 500,000 robots in 3 years?

I think it would be great, but given the time it took from his announcement of a robotaxi to actually getting the robotaxi, I’d like to take this with a grain of salt.

The trend is what’s important. If it happens in 3 years, 4 years, or 5 years, it doesn’t matter. It’s going to happen, and when it does, it’s going to be a game changer at so many different levels.

Agreed.

Let’s turn the conversation in a different direction to close this out today. Buddy, you just took part in something that is not technical, but it is powerful. That was your father’s passing.

I think you had an experience that is worth sharing. It’s not technology, guys, but it is extraordinarily and deeply meaningful. Share it with us.

My 97-year-old father, who lived alone and drove around until a few months ago, had an assisted-death procedure in Canada. We associate death with grief, loss, pain, and suffering. It’s all negative.

What I experienced with him was the most joyous, ecstatic, and blissful experience I could possibly imagine. If I go out with one-tenth of the ecstasy he had, I’ll be very happy. He was in total bliss.

A couple of days before, he had time to say goodbye to everybody. We timed it, and we played Beethoven’s Ninth Symphony for him, which was his favorite piece of music. When he passed, I was sitting there stunned. I said to the doctor, “I don’t know how to process this. He was so happy. Have you seen that? You’ve done this procedure.”

She said, “Sir, you have no idea. Most people who go through this assisted-death process go in that state because they have agency. You can die with dignity. You can time it and manage the schedule.”

It really shook me in a positive sense. I think we have to rethink this as a society, because it couldn’t have been more peaceful, happy, blissful, or ecstatic. He was literally in a plant-medicine ceremony.

It’s got a stigma here in the United States.

Everywhere, usually through religious structures that try to keep a person alive forever. My father was very clear. He said, “I do not want to be a burden to society, the family, or the health-care system.”

It was an incredibly beautiful experience.

This is completely opposite from what I normally speak about, which is using technology to extend your life, not end your life.

Yes. My father was a profound atheist his whole life. In the last 2 days, as he was saying goodbye to people, he was saying, “See you on the other side.”

Wait, what? What are you saying?

It was like he was late for a meeting and had to get there. He was saying, “Get me out of here.”

He had a palliative-care doctor come in, and the fellow said, “I’m here to make you more comfortable.” My dad said, “Are you here to help me die?” The doctor said, “No, no, no, just to make you more comfortable.” My dad said, “Then you’re of no use. Bring the doctor who’s going to give me the procedure. Why are you even here?”

It was amazing.

That’s incredible.

We had a little memorial service with my family, talking through our different experiences with him. It was one of the most amazing lines he ever gave me.

I gave a talk ages ago called “Fixing Civilization,” and he said, “I totally disagree with your talk.”

A minor title.

A niche effort. He said, “The problem isn’t with the fixing part. It’s with the civilization part. We operate as apes and tribal clans with more and more powerful weapons, killing each other. We’ve materialized the world; we have yet to civilize the world.”

It was one of the biggest wisdom pearls from the elders I could ever have had.

That’s beautiful. I love you, buddy. It’s a beautiful way to close this conversation. It could not have been any more beautiful.

I’m thankful to your father for giving the world you.

Thank you, my friend.

I’ll see you in our next episode of WTF Just Happened in Tech. And guys, if you enjoy this conversation with Salim and me, let us know. You can subscribe, and you can tell us what you want us to cover in the show notes. We do pay attention, and we’re grateful for you. Tell us who you think is right or wrong in the comments on some of our disagreements.

Tell them, please, let Salim know that he’s wrong and I’m right.

To follow Salim in EXO, it’s EXO Universe. What’s the—what’s the—

It’s openexo.com.

Thank you. I should know that by now, shouldn’t I?

Well, it’s close to openexo.com, but we are opening ExOs. Our mission is to create as many exponential organizations in the world as possible.

Fantastic. Check us out at abundance360.com. Anyway, love you, brother. Be well.

You too.