Room-Temperature Superconductors and Nuclear Fusion with Andrew Cote

 
 

Room-temperature superconductors are here! Or are they? Today we’re going to explore the latest findings on the LK99 experiment. I’m speaking to Andrew Cote, a stellarator engineer.

(For anyone who isn’t a nuclear physicist, a stellarator is a type of fusion reactor. Andrew’s 

basically a nuclear fusion engineer who uses superconductors in his daily work.)

In this episode, we talk about the importance of superconductors and how they could impact everything from electronics to energy generation. Andrew discusses the legitimacy of the recent superconductor breakthroughs. We also talk about nuclear fusion, quantum computers, and… invisibility cloaks?

Here’s what I learned from the episode:

  • LK99 is a potential watershed moment for science. It could be compared with the invention of the transistor.

  • A room-temperature superconductor could revolutionize electronics and technology broadly. It could have a massive impact on various fields, including power transmission, microelectronics, and quantum computing.

  • We delve into the idea of commercializing room-temperature superconductors, through an entirely automated AI robotic material science experiment farm.

  • Despite the disappointments in some aspects of LK99, the experiment led to interesting findings like the doping effect in a lead crystal. These could have applications in material engineering, possibly leading to innovative designs like a new transistor or other materials unrelated to superconductors.

  • Fusion energy is the "last energy source humanity will ever need." We talk about the timeline for fusion becoming a reality, its current challenges, and its potential to revolutionize energy production.

  • Superconductors transmit energy without loss, improving the potential for large, centralized nuclear power plants and the idea of solar power from the desert powering distant cities like New York.

  • Andrew does a deep dive into the complexity of fusion reactor design and how they produce energy. 

  • There’s huge international collaboration in fusion research, such as the ITER project. Andrew shows fusion's symbolic role in "thawing the cold war."

  • There’s a strong correlation between increased access to energy and improved quality of life.

  • Failing to transition to abundant energy sources like nuclear fusion or unlimited geothermal would lead to dire consequences for all of us.

  • Science fiction has anticipated many technological developments. There are huge second and third-order impacts technology can bring to civilization.

Invest your money into high-growth startups at the earliest stages

When new technologies meet the market, the world changes for the better. That's why we invest our money into the best founders in our network.

We write small checks to 15-20 very different startups each year. Previous investments include Aalo, Gently.com, Omella, Driv.ly, Weavechain, Stell Engineering, Ouros, Solve Data, and more.

Our Website has background on the fund, our past deals, and more.

Accredited Investors: send me an email, I'll send you our deck, and we'll get you into our deals starting this quarter.

Learn more about Andrew Cote:

Additional episodes if you enjoyed:

Episode Transcript:

Eric Jorgenson: What's it like trying to keep up with breaking news worldwide? That's like happening in real-time in your specific industry. 

Andrew Cote: Yeah. That's funny. It is the first time archive has been like a competitive sport or something, or people are actively racing to refresh, like the material science section or the condensed matter physics section of the archive.

It's what is going on. But yeah, it's been interesting. It's excellent, and it's great that people are interested in these kinds of fundamental science things. They're out of the spotlight. 

Eric Jorgenson: Yeah, it feels like this slow-motion Michael Bay movie of people worldwide, like coming together, finding each other on Twitter.

I can't imagine how many new connections have been made of people that didn't were just lurkers before that are suddenly like, Oh, my God, this is my spot. It's my time. I got to speak out. 

Andrew Cote: Yeah, it's funny. It is an interesting social phenomenon for people to rally behind something like a fundamentally optimistic narrative that's very hopeful about something.

Usually, the kind of viral news stories are just horrible. It's just awful news. It's like some disaster or something. So yeah, that was interesting. 

Eric Jorgenson: Yeah. Was it you that said this? We're seeing the same information patterns, but. That happened with Covid, a negative phenomenon now happening with a positive phenomenon.

Andrew Cote: I don't know if I said that. But that's interesting. Yeah, it's an interesting comparison. It's true. Both are; here's this potentially colossal deal unfolding in real-time. And so you're very incentivized to be aware of it. But this one is all about more basic material science rather than Hospital rates, and they're very different phenomena and scales, too. It was mainly confined to some people on Twitter.

I think pretty much. It's funny. It's just people tweeting on this stuff. We're just like, Hey, this is what I read. It is cool. I didn't expect to find that. And then all these newspapers are like running with headlines and trying to keep up like the reporters are trying to keep up because it's all happening in real time somewhere else.

Eric Jorgenson: Yeah, I want to set the context for this conversation and the rest of it by reading one of your tweets, which is, "Every company builds products using the same limited set of available foundational technologies. Physics produces something fundamentally new every few decades, changing what's physically possible.

If successful, LK99 would be a watershed moment for humanity, easily on par with the invention of the transistor, overnight revolutionizing all of electronics and technology."

Sothis LK99 is the superconductor that has been setting this little corner of Twitter on fire. Of all the people I've followed and all the threads I've read, you strike this fantastic balance of a balanced perspective on optimism and scientific cynicism.

Approachable, but not afraid to carry forward the implications into why this is important and put it all into a vast context, which is why I wanted to have this conversation in a longer form and try to pull on a bunch more of those threads because it's so hard to Jump around on Twitter.

It's just a lot more work to assemble it. And I feel like if we can do it all in one big body of work here, it'll help people wrap their heads around superconductors in general, as it's an example of the technological progress that we're seeing that's like right on the edge of like raw science turning into holy shit, there's a new technology turning into, oh, here's the societal change that comes out of that.

Andrew Cote: Yeah. And that's a theme I pursue in much of my writing: we never really know how great things might be soon because we can't see revolutions coming. Yet we keep experiencing them. And so we have this weird short-term memory where it's like, things have constantly been changing again and again.

Yet at the same time, we look at our problems right now and think, Oh my God, we're never going to solve these. It's impossible. Reminding people of the dramatic changes that can occur through advances in fundamental science is essential. Isaac Asimov had this quote, which was great, which is science can wonder and amaze, but engineering changes the world.

So really understanding the connection between science as it is pushing the frontiers of knowledge and capability engineering as it relates to the practical application of these things in our material, social world. And then one more step, which is how is the world different? How is life different?

What's it like in a world with nuclear rockets, new spacecraft engines, faster computers, brain implants, or superconductors? So it's funny. The supernatural thing is something I've worked on for a while. And it's always been exciting. I first wrote about it about a month before all this happened.

I just had this thread. I was just like, Hey, look, we've had progress so far. We keep breaking expectations. Nobody thought we could have high-temperature super inductors 1986 came out of nowhere, broke our theories, and a Nobel prize enabled all this new stuff. So, what if that happens again?

Like, what would happen? And then a month later, this story breaks out, and I was like, Oh, Hey, that's cool. 

Eric Jorgenson: Yeah. Let's do the superconductors as the first one of these, like science turn engineering turn societal impact, because yeah, you were writing about superconductors before they were excellent, but let's talk about the paper that started this whole phenomenon.

Andrew Cote: So that paper came out, it lingered for a while. Nobody saw it. My buddy, Ryan McKenna, put it online on Twitter, and it got some attention and just said, Oh, we're so back like this paper, right? And then another guy picked it up and then became more popular.

That paper was fascinating. If you look at the field of publications in that topic area, it would only sometimes stand out. So it was selected for by this crowd's behavior in some way; the first time I wrote about this, I just called it the good, the bad, the ugly, and the good is okay; there's some reasonability here, some viability like there's this and analysis of the crystal structure. It uses compounds like, so it's using copper and oxygen, which are previously found in superconductors and stuff. So it wasn't like saying it's like an impossible material out of the blue.

And it was also easy to reproduce, right? So it would have been easy to check because they gave you synthesis methods. They made some specific predictions about the crystal structure. The bad part was it needed to include a few key measurements. You usually look for both things in a superconductor, like a very high transition temperature.

They had a more decaying transition temperature, which means at what temperature does it have finite resistance and then drop suddenly to zero resistance? Some other things were quirky like the scaling on their plots hid essential details.

And yeah, so there's a bit of thing missing there, it was like 80% there, it was 70% if they had filled in the details, and it looked like it was rushed story came out, it was rushed. Some politics were going on with the authors, which had this plausibility thing to it.

So that was part of it. But, what struck a chord with people's imagination was that there's nothing physically impossible about this, but the impacts are enormous. And so even Paul Graham said it well. He said this is an excellent lesson in expectation.

Cause, you know, even a conservative MIT professor said this is a 5% chance. It is real. It's, oh, a 5% chance of this is a big thing. What is the 1% chance of an asteroid? That's a big deal. The paper triggered a lot of interest, speculation, and replication attempts. It was an exciting phenomenon in people's imaginations.

Eric Jorgenson: Yeah, there are a few exciting things in there. One, the ease of replication kicked off this whole; I'm a big believer in popular science; everybody should be able to replicate anything. Replications aren't hostile. They support science and verifications.

And this was something that people could only do quite in their garage if they had some badass equipment. Still, indeed, in many workplaces, many universities could do it. And so we got a ton of replication attempts and a ton of iterations on the method and the materials and things like that.

And it's interesting in there. As you pointed out, every time we discover a superconductor discovery, it usually represents a whole class. It kicks off a series of similar improvements. I'm playing Battleship. We discovered a new zone to iterate within.

Andrew Cote: Yeah, that's very true. I can respond to that. Look at this plot of different material compositions. You can see these kinds of little branches in a plot versus time climbing to higher and higher temperatures. People find new compositions that are different from each other.

And then they improve them a bit. They optimize, and they tweak them. And it's an exciting process. Suppose you think about the atomic composition and crystal structure. In that case, is the space of possible materials ample and suitable? I was looking at numbers in this recently.

Suppose I have a little crystal unit cell that could have ten atoms or 20; it's 10 to 11 combinations and 20 10 to the 28 combinations, and 30 as 10 to the 37. Now it's just it just blows up so fast. Like the number of combinations, all have a unique electronic structure. And electronic usually means circuits, but in the material science sense, it usually means what are the allowable energies for electrons? What are the orbitals that you call, so the shape of those orbitals, and where do they overlap with each other?

And that determines many exciting properties that scientists simulated for this study. So finding a good starting point it's like finding gold in a river. The mountain range is so extensive. Okay. Everyone's looking for gold, and you need help to search the mountain range.

It just takes a billion years. So a bigger gold deposit is nearby if you find gold flakes in a river. So you can start with the material. It's not promising to optimize it, tweak it, get better within that range of mountains, and then find where the real gold is.

Eric Jorgenson: That idea rhymes with the phase shifts in computing. We started very mechanical and then moved to transistors and chips. As that has changed, we could do another one going to the biological computer, optical, or quantum.

And this feeds into quantum computing, so we'll return to that later. Something else you said, I didn't realize that this paper had lingered before somebody picked it up.

Andrew Cote: It came out on July 22nd, a few days ago. It just wasn't right away. 

Eric Jorgenson: That made me worry about what else might be out there that people aren't snapping up, publishing, and trying to replicate and see how efficient is the market of archive and new scientific papers being read and spread. 

Andrew Cote: The science market is an interesting one.

For reference, something like 80% of publications can never get replicated. The people that try can't replicate them. So it's a meager replication rate in general. Scientific findings are unique and one-off. There are a few effects here.

So think of the publication like this tree branching out all these different directions, people trying different things. Sometimes, people will get damaging results on a topic, and one or two things could happen. One is they don't publish that, so no one else knows. And they think they'll try it too.

So you waste time trying things that have yet to work. Another one is they think it doesn't work, but it could have worked if they had done it differently, and then you trim that branch, and people don't pursue it. So a wealth of knowledge is embedded in our legacy of publications open to reinterpretation and experimentation.

As scientific careers unfold, people are trying to claim new territory to build their publication record history. So there's less incentive sometimes to pick up those old things to find diamonds in the rough.

You could say. 

Eric Jorgenson: And on your brute force point, I think someone did the math on this and was like, it would have taken half the computing power in the United States, 10 000 years to brute force find this configuration, but once we had the configuration from experimentation, they could the simulation, as you pointed out, could do experimental iterations nearby and discover a lot of cool stuff or at least validate the possibility of it.

Andrew Cote: 100% Yeah, that's a fascinating little sub-topic. I had a speedy idea for a startup when all this was blowing up. How would you approach this and eventually back business? How would you commercialize room temperatures? We can talk about that later if you want, but this is an excellent thing in this field of simulations.

So what are these simulations? You take a bunch of atoms, right? And they all have energy levels around them, electron orbital shapes. And what's happening is that those atoms are like defining the potential energy function for an area. And then, you do what's called solving the Schrodinger equation, which is given a specific different equation, seeing where you can find solutions. Those are where electrons are allowed to live. So it's like a mind-body problem. If familiar with physics, like this mini body problem, meaning it's like impossible to get analytic solutions to. So the standard software it's called this Vienna AbInitio Simulations package.

It uses a bunch of approximations to solve the shorter equation in these different situations, and it has varying degrees of approximation and accuracy. And people develop modules or abilities to include other kinds of effects into it as well. So the simulations aren't deterministic of materials' properties.

They're trying to approximate it in some way. But what people have found, and I was reading some articles on this, is when you have a starting location, you can use an evolutionary algorithm, which is defining some fitness function saying, I want this, I want a square watermelon.

That's my fitness function. And now, I will start reading watermelons over time, picking the most square ones. I can breed a square watermelon. And it's the same thing with the simulation package, where I can start with a configuration and then evolve from there.

Towards a desired direction is a pretty effective means of exploring or traversing the state space with issues. How do I get over local minima? Does a solution have to converge smoothly in quality because you can never get over a hill into a better region?

So there are other things there, but yeah.

Eric Jorgenson: Many simulations could be better because we need to understand them all precisely, the quantitative, like complete formulation of what we're replicating or what we're trying to simulate.

Andrew Cote: Nature is intractably complicated and has unlimited floating point accuracy.

There's this common phrase in physics. Sometimes you hear it like the only perfect simulation is the object itself, right? Because that has the full fidelity.

Eric Jorgenson: Let's unfold your idea for commercializing room-temperature superconductors. 

Andrew Cote: Okay. Yeah, Totally. So back to the mountain analogy. So we're mining for gold. And some guys find some flakes in the river and the prospecting one Oh one, it's all about where's your claim. Where's your plot? So the first thing to ask yourself is. This guy found a river in the Sierra Hills. Is he standing on the best or right away? Is it? Does he have the IP already? And the answer is probably not. There are other ways of getting there.

Another thing to think about is that the last time high-temperature superheroes were discovered was in 1986 took 30 years for them to get engineered into a commercially viable package too. So there's lots of competitive landscape to Conquer in the engineering optimization of that stuff.

So the strategy has got to be based on this. Look, how do we strip mine the hills? Okay, fine. Do more than use picks and shovels. People always talk about picks and shovels, the gold rush. I want that machine from Germany that scoops up the ground. Okay. And giant buckets at once. So what does that machine look like in material science?

And this is just; this is all super off the cuff. So people would say that's whatever. But you get an incredibly high computing team. People that are used to running simulations. And then you start high throughput, stimulating these shorter equations, solving them evolutionary search items, all this stuff.

You can make ML feedback loops within that, which accelerated like evolutionary algorithms, one of them. And on the other side, you have to have a material component, which is this field of robotic self-driving laboratories. And this field emerged at a conference in Toronto later this month in a couple of weeks. However, it's 500, 000, 000 was granted by the Canadian government, the University of Toronto, to investigate constructing these things in a productized fashion or a high throughput, more capable fashion.

My buddy. He just did his Ph.D. on the topic of self-driving robotic laboratories. And yeah, the gist is, so it's like a super multi-axis arm with a bunch of different interchangeable tool heads, and then around it, it could have a rail, it's moving around, but it has just a bunch of laboratory stations. It can mix materials, grind compounds, sample them, put them in an oven, take them out, and time everything. You're using computer vision, like inverse kinematics, to control the motors, all this kind of stuff.

As well as then the materials testing, right? So you can imagine, think of a warehouse filled with those like sushi restaurants where the sushi chef is cooking at the grill. Everyone's sitting around something like that size, right? Except it's just a robot arm.

And instead of a sushi grill, it's like a bunch of laboratory stations. And the robot arm is just its share of the Tesla factory, right? It's just like trying stuff grinding through to have a warehouse full of that, maybe 20, 50 of those, the cost of a few million bucks each, and they're just speed running the material science. They're just speed-running 30 years of academic research as fast as possible. And so those robots are also amenable to ML optimization called Bayesian learning, which is more straightforward. It's when you have a black box model because the robots aren't positing causal mechanisms behind why the materials act a certain way. They're just exploring the configuration space of preparation methods.

But then you have the simulations team with the theory of why materials behave as they do. So you have these two areas of operations that can also be in feedback with each other. So that's like heavy computing, heavy robotic enabled. Why now? Because these tools are just available, we will raise this kind of stuff.

Three hundred billion or 300 million dollars to start is one of those things where look at this company; you could have raised throughout that company's life. You could have put 2-3 billion dollars in it and have a thousand X return. Cause the market size of these application spaces is in the trillions of dollars.

And that's an annual market size, right? So it's nuts.

Eric Jorgenson: In the output of that, you conceived to be like a new material science breakthrough where you own the IP of the material. Imagine drug discovery could be accelerated this way to like that.

That feedback loop is generalizable, but material science is just because the market is enormous. 

Andrew Cote: The market is so big. Yep. That's the right, perfect analogy in the material. It's right. So many startups in the space have raised a lot of money for silicon simulations, which are trying to find ideal adapters or antibodies, therapeutics, etc. 

In the simulation, that kind of stuff saves a lot of time before they start trials and testing on cell or animal models. The funnel for pharma development is so wide at the beginning. It narrows down to such a thin point, the more you can do to reach the top of the funnel and get better viable candidates. Instead of 1 working, you have 120 working, right? That saves a lot of money upfront.

R and D drug cost is like a billion dollars or something to bring to market or ten years. That was about right. Yeah, it's unbelievable. But a lot of analogies too. There are also lab companies, like transcripts, that tried to provide laboratories of service, which was like, we can send you our protocols.

It's all the life science stuff. Send you our protocol. They have robots that can do the operations and give back the results. Those have found mixed results. Partially, life sciences can be challenging because there's a much higher degree of chaos and complexity with the temperature.

Material science is similar, but both are very high; they're very skilled, complex disciplines, right? Cross contaminations, brutal, all this kind of stuff. 

Eric Jorgenson: The startup idea and paper background are good lead-ins to the discussion around the engineering and impact of superconducting materials. It may be an example of why this market is so big. Let's talk about the value of superconductors and the impact of them. 

Andrew Cote: Totally. Yeah. Interesting. So 1 of the tweets I put out there, I was trying to expose all this thinking to people.

To get them thinking about it. So you can think, and I'll reason by analogy. 1986, we have the atrium barium, high-temperature superconductors. They came out, and the performance wasn't super strong right away. It's gotten a lot better over time. It was initially challenging to package them in a format that would be resilient.

And three numbers matter for this. The two main ones are how much current it can carry before it craps out and become a regular conductor. And then also, how much magnetic field can it live inside? How much can it withstand? Both those things affect its ability to conduct super. So they're like the performance envelope.

You can think of it like that. Yeah. A motor performance envelope is like its RPM versus its torque, right? So at high RPM, you have low torque; at low RPM, you have high torque. Same with this stuff at high fields, low currents. At low current, sorry, at high current, low fields, right?

They trade off against each other. So we've gotten better at that over time. And the thing that is indeterminate with any new material is how good it can get over time. So the way to think about it is, given its engineering performance envelope, where would be the applicable areas you could use this?

Okay. So there are three cases. First is Low current low field, right? So a very delicate, sensitive material. So, it is only applicable in things like, a lot of microelectronics applications. A lot of electronic sensors also make sense. 

There's an integral part of quantum computer design; the Josephson junction is like a small loop of superconducting material, and a current travels in a loop around that. And then there's some bridge where the current is tunneling across that gap or this resistive bridge.

And it's like a fundamental building block. It's like a diode or a logic gate equivalent for typical circuits. So people build stuff out of that, but that would be a low current, low field application. And the thing is, such tiny antennas for IOT devices and Bluetooth and cell phones, as you make those antennas smaller and smaller, the limitation on their ability to pick up signals comes down to what you call the surface resistance, which is like, how much resistance is there in the surface of the copper?

If you eliminate that, you can get smaller, more sensitive antennas. So that's, again, shallow currents—a low signal like the low field. And I think a similar line of reasoning. So those would be easier to make because they're what you call disaggregated.

They're separate from the rest of an integrated circuit. So they're like a standalone piece that could be put on a PCB board or its little board, like an antenna. Another application, though, much harder, longer lead time is to start building integrated circuits directly out of the material itself.

And that's the biggest win. But it's the hardest to make, right? Because the silicon wafer production process is highly optimized, right? TSMC, their billion-dollar machine, it's like this 300-millimeter CMOS wafer fabrication process. It's super advanced. It's optimized.

And so it's very tough to fit a new material substrate into that and make it work at a performative level. That said,  you could probably catch yourself in the long term, and where that would get you is incredible. One of the most significant limitations on compute density, like why can't chips be cubes, is removing all the waste heat, correct?

So a superconductor doesn't lose energy to waste heat. So you could make 300 times more energy efficient. These are the numbers I looked up. It was a study looking at an existing computer. One example is only to represent some sizes of chips. Still, an X-scale computer is 300 times more power efficient and ten times faster.

And it's like chip refresh speed. So definitely some big ones there. And yeah, the low field, low current is like a huge win, like all computing and sensors. And then, as you increase the field in the current Karen capabilities, you unlock. New applications, like you, still have those base cases, but the next one is looking at things like power transmission, So high current, but maybe low field because the wires are in a straight line that's still huge. That saves like 50ish nuclear reactors worth of power lost.

Eric Jorgenson: I was surprised by the energy loss from power generation origination to like end user is often 60% plus. So even if we just 60, oh, that's big.

Andrew Cote: I saw 5 to 7% or up to 5 to 12% for transmissions just carrying it. 

But you're right in that. There are also lots of losses. Yeah, in the transformers, which step between voltages and the generators, which are, like, turning the turbine that generates electricity too. So, yeah. Total system losses are enormous.

Eric Jorgenson: Yeah. I'd imagine they get worse. The closer they get from a nuclear power plant to like my iPhone is like the most significant steps down are between the actual grid and the phone, but that's huge. So even just replacing a few of those intermediate steps with zero resistance, things could theoretically not that there still won't be, but immediately one-third of the energy cost without building anything else just with our existing base load.

Andrew Cote: It does a fantastic couple of other things too. Which is it? So now more extensive applications. You get more efficient generators. So they take less material. So smaller, right? Smaller footprint. And that's like the thing that you turn that generates electricity like it could be a turbine, nuclear plant, coal plant, whatever.

It's all the same kind of generator idea.

Eric Jorgenson: Do you prefer nuclear around here? 

Andrew Cote: Yeah. I'm a super proficient guy. Funny. My uncle was one of Greenpeace's founders, a big anti-nuclear organization worldwide. Yeah. In Vancouver, British Columbia.

They were against nuclear bomb testing, which was the right call. And then, they expanded the scope over time. So I won't comment on that; just a funny anecdote. So you can do exciting things that are counterintuitive. Like right now, solar panels, energy is always local, right?

Because transmission losses suck, and no one wants to run it too far, but why not have solar power in the desert that's powering New York? So you can get there are no losses along the way. So you can, and there's another thing to like reactor efficiency in terms of like material costs versus power wattage output.

It gets better as it gets bigger. So you build super giant centralized nuclear power plants that spread everywhere and power everything instead of like tiny plants everywhere. 

Eric Jorgenson: Very cool. Okay, the middle bucket is a lot of like power transmission, power generation, that transmission.

Andrew Cote: So generation a lot of that requires high fields too. So then, now you go into the high-field bucket. Okay. So rough numbers. The first bucket was one and a half trillion. I did the math earlier. These are 2022 numbers, very rough approximations. The middle bucket was 750, 500 billion, like five, five to seven, two, two. Still, I am trying to remember the third bucket was another one and a half trillion or one trillion. 

Eric Jorgenson: So, annualized GDP increases?

Andrew Cote: These are the total market sizes for those markets. It wouldn't replace the whole market, but it's the size of the market taps into, right? You could get 20% of the market selling the components into it or something like that. So we're talking about which currents, high field applications now like fusion reactors. 

Maglev trains, MRI, and medical imaging, correct? All those require high field and high current. You might be thinking; we have two variables. We should have a two-by-two matrix of applications. And so, why are only three cases right? 

You can't have a high field without a high current because the fields induce currents. And so that's a more complex state to access. You can have a high current with a low field because the wire is straight. So it only has itself field, which can be small. But if you make a coil out of that.

He experiences his fields, but making coils, how you make magnets. So nuclear fusion would be significantly enabled by this power generation, power transformers, maglev trains, MRIs, and medical imaging in general. 

Eric Jorgenson: Let's dive into fusion because that's what you're working on right now, right?

Andrew Cote: Yeah, that's right. 

Eric Jorgenson: Cool. Are we making progress on that? Are we going to have fusion? 

Andrew Cote: I'm optimistic about that. We've had consistent progress with it over time. We're not looking at this point for Deus Ex Machina, Salvation breakthrough in any one area.

We're within reach of steady improvements with what we have currently. And much of that was enabled by the HTS magnets we have today. Who would have thought, right? But so that high temperature super just every from the 80s. That's why we have magnetic confinement today, right?

Confinement fusion companies are because now we can have superconductors in 10 Tesla fields carrying like a thousand amps per square millimeter, which is crazy. Not both of them simultaneously, but those kinds of performance limits are impressive. And they work with liquid nitrogen, which was a huge breakthrough.

So that's made it way cheaper. Otherwise, you have to look at helium, which is way more expensive and rarer. Nitrogen is like 80% of the atmosphere. So that's been exciting. There's a whole, and there are three general classes of companies, magnetic confinement, like token max accelerators.

It's like the plasma donut. Some companies are called magneto inertial, like collapsing plasma gas balls. So general fusion used to work there. They're trying that trial for energy has that kind of design. Helion has a similar thing to her. Compressing them together, but starting with plasma. And then there's inertial confinement, which uses solid fuel pellets to compress those, usually with laser light.

Although some companies use more different kinds of shock impact methods, like this first light fusion, it's cool because there's a zoo of different reactor types out there, and they all have different trade-offs, right? They all have different advantages and stuff. So it's like incredible.

It's like the 50 states. Each one has a different government, and you see which one you want to join. 

Eric Jorgenson: And you moved to the plasma donut. 

Andrew Cote: I went for more donuts. Yeah.Funny enough. So now I work with stellarators, which are like.

Somehow the lesser popular magnetic assignment type is the donut type. Historically, in the United States, 90-ish or more percent of funding has gone towards token Mac development. And for a good reason, suitable? They're easier to design. They're easier to operate. It's interesting.

So stellarators were invented 1st, right? They were in the 50s at Princeton, but the 1st ones they made needed better efficiency and poor performance. And people were trying other things, field, reverse configuration, Z pinch. This company is now pursuing those, by the way, in more modern, updated forms. But a few different designs are floating around in the States.

People are trying, and they could be working better. And then the Russians, it was 56 or so, they published their data on token max, which was their design. And it just smoked everyone else. It was just like, look, these are crazy. People didn't believe in numbers.

The Russians invited them over. It was like a good thawing of relations. And everyone in the States just switched token backs, robust, which drew a lot of attention for a long time. Now, token backs are well researched and developed, like France will be the world's most enormous 35 billion dollars. 

Total civilization-scale engineering, right? Like totally insane. Like dozens of countries, the eater is funny.1986 will come up a lot in this podcast. That was the year Reagan met Gorbachev in Reykjavik, Iceland. And they wanted to have some project that would improve relations.

And that's where the eater was born. And that was a collaboration point scientifically because they had now a couple of decades of token Mac collaboration, right? So this is fusion bringing the countries together. Fusion's thawing the cold war, right?

We are fusing our worldviews. Yeah, totally. All right. You can edit that out. I'm just kidding. 

Eric Jorgenson: I won't. We love bad jokes around here. 

Andrew Cote: Hell no. Yeah. Okay. But so, really cool stuff on this fusion front, stellarators. So they fall into the backseat for a long time. And then also in the 80s, it was later 80s, this guy, Alan Boozer, came out with a new theoretical model of stellarator plasma, right?

Of a model of looking at this plasma in this different frame of coordinates, but that would exploit critical symmetries, and that's started this whole kind of gold rush in the field in a sense because people now could think, wow, like this would perform even better than a token back. And that's the kind of allure, right?

So the quick, the classic thing token max is easier to design, harder to run stellarators, more complex to design, easier to run. Tokamaks are pulsed operation. We need to turn it on and off. Okay. Like a four-stroke engine, only a quarter of the strokes are like detonating fuel. And then still later is more like a steady state thing.

It's more like a furnace like the fire keeps burning. It's constantly producing energy, and it's less liable to disruptions. Disruptions are when the plasma goes haywire inside the reactor. It's pretty wild. So eaters are donuts. You can think of this current as traveling in a loop around the donut like a racetrack, and that has tons of stored energy.

So it's about 60 megajoules in the eater. Something like 15 mega amps of current, and when the plasma disrupts, it becomes unstable. Imagine you're like balancing a broomstick on your palm. And once it starts to fall slightly in one direction, it keeps falling faster and faster. So it's precarious.

When that disrupts, it disrupts in the form of this beam of current slamming into the wall. And it's 60 sticks of dynamite, right? So it can melt a solid steel block with no problem. Vaporize it. So the inside of token max is often armored with tungsten tiles. It's like they're serious.

Look at these inside photos. It's a super cool fusion in general. It's one of those badass areas of engineering that I am very fortunate to be a part of in any small way where the extreme limits of engineering are all brought together at once. 

You have a plasma, a million-degree plasma, right? You have neutronic radiation, like neutron particles being generated or rating stuff. You have massive magnets, magnetic fields, okay. You have cryogenic systems and vacuum chambers. You have all these crazy esoteric plasma diagnostics.

It's tough to tell what's going on. Plasma is just this big smoke ring glowing hot for a millisecond. They use lasers. You shine lasers at it. You take Measuring the rotation of magnetic fields and polarization, and it's all fantastic.

Something people don't know about. So a lot of these reactors use what's called a neutral beam injector. What the hell is that? It's a particle beam that blasts fuel in there. How do you like shoveling coal into a furnace? How do you load a fusion reactor?

Use a particle beam, right? Of course. It's just the coolest thing ever. But the funny thing is these particle beams, these neutral beams, and they're so giant and massive. They're as big as the device itself. Like you, they don't get any attention. I've worked on a, so I used to work in accelerator physics, like more accelerator engineering.

So that's like radio frequency, acceleration systems, vacuums, and cryogenics. And it's a similar tech stack as fusion skills-wise. Yeah. That's some excellent overlap. 

Eric Jorgenson: Yeah, you've had a few exciting things. You had a stint in bio as well, didn't you?

Andrew Cote: Yeah, man, that was wild. Boy, I tell you what. So I wonder if I'm, I wonder if like right guy, right place, right time sometimes, but a couple of times things have happened that seemed was dental. Yeah. So 2019, me and some friends and I tried our first little attempt at a startup together.

We didn't go big or anything, but some friends were working on the school project, and it was like for a fry racing team that we thought was super cool. And it was like, we're just like, whatever. And before that, I had mostly done applied physics research. So it's an incredible look at a new world of startups and entrepreneurship.

It didn't entirely take off. It was like too much of a science project, not much of an engineering project. So that's different from the kind of company you want to have because it can take forever. So I was thinking of what to do. It was in Vancouver. I was supposed to work at Lawrence Livermore national lab.

Okay, which is where NIF is, and I was going to work on this laser system. It's cool, I met this brilliant scientist there, and she was, we were in touch for a while. And it's a field called laser Wakefield acceleration. So laser wake, what is that, right? So it's like the next generation of particle accelerators.

Right is using plasma as the accelerating chamber, and it's wild. It's like basically wake surfing, right? Like behind the boat, right? So about toes, a person, and the person riding the wave behind it. Instead of the boat, it's like a late laser beam. And the person surfing is a bunch of electrons.

And you blast this. It's just this is nuts, right? It can't be. It's real. You blast the plasma with a super powerful laser pulse—a crazy amount of petajoules or petawatts instantaneous power. And the energy might be like 10 or 20 joules, which is a lot for a laser pulse.

Holy crap, and then trailing just behind it is the electron bunch, and you get energy transfer effectively from the laser to the electrons. So this is wild because it means particle observators could be like. We need two hundred times or 20 times shorter to hit higher energies in fundamental physics.

So CERN is amazing. CERN found the Higgs boson, right? The next generation of proposed accelerators is a hundred kilometers long to reach the next energy level. So this is unbelievable. So where was I? Okay, so I was going to join. Yeah, it's nuts. So these future circular colliders and these international colliders propose things like future circular colliders.

It makes CERN look like a little corridor next to the basketball. It's like a megalomaniac. I love it. That's great. How else are we going to make black holes? Okay, great. So I was going to join this accelerator lab, and there was little startup space there.

Like it's not like a, whatever, and I saw this post for this place, Chan Zuckerberg, biohub on LinkedIn. And it was a guy that had done the same undergrad as me, not the same year as me, but he said, Oh, this is great for people from that background. It's multidisciplinary. It's research focused.

And so I looked into it and was like, it's in San Francisco. And it's infectious disease research. It's a nonprofit. It's October 2019. It sounds great. It's a great pitch. It's building stuff for developing countries, helping them find new diseases. I thought that was awesome. And it gets me to San Francisco, just the place to be.

And it's also a nonprofit. As a hedge, I could be more research guy after that. I could go into industry. It was an excellent bet. And it was a fantastic place to work. I can't recommend it highly enough. Like the bioengineering team there with Raphael and Paul, these guys were just like, and they were just like super engineers.

They were just totally, and it was one of the most competent groups of people I've worked with. Overall, everyone, there was top of their game. And then a few months later, the pandemic breaks out, and it's holy crap. And so suddenly I'm like, with all these like professional biologists and stuff, and everyone's yeah, okay, we're going to set up this testing lab.

That's a big sprint, a big hackathon. So seven-day hackathon, instead of whatever B2B SaaS, you develop a QPCR testing lab. So that was pretty wild. That was one of the more interesting, several-week periods of my life where we talked with regional county health officials in these meetings.

I got roped into some meetings in DC, too, with similar engineers and doctors trying to figure out how we will make ventilators. Are we going to run out of hospital beds? If we project the epi, like if you project the hospitalization numbers statistics from other places onto our local capacity and forecast a shortfall in facilities and equipment. It was drastic, and those projections were not as bad as we experienced in many ways. So we, so that those didn't manifest, but still facilities were massively overwhelmed.

So yeah, you go from building this lab in 7 days to Michael Lewis writing this big piece on it and Bloomberg COVID lab that could save America. That's this dramatic story. It's a great piece. And then, after that, we transitioned to building like ventilators. Digital electronics and a bit of software design, C plus and stuff like that, using finite state machines and signal processing to look in real-time, monitoring breathing patterns as we picked up like a pressure sensor and detect if they're going into if they're falling out of sync, if they're outside of some limits, right?

Yeah. So that was like, really okay, wow. Life sciences are essential stuff. And then, actually, I was there for about two years. And then I tried my first real startup out of there with some friends, the same guys from the Ferrari team. 

Eric Jorgenson: Cool. What was that company?

Andrew Cote: Yeah, it was called Ada bio. It was super fun. So that was like 2021. I'd been the CCP for a couple of years. There was a lot of talk about it. Thinking significantly about new projects that would help the organization and a common thing that had come up in many folk's discussions was data management. It was interesting because the life sciences have advanced a lot.

We're still early days of understanding how to write DNA. We don't know it at all, really, or that stuff, but we're learning the cut-paste copy commands from CRISPR and other stuff. So we now have like single-cell sequencing devices. They have mass spectrometry.

They have transcriptomics, which tells you it's like the stack trace of the cell. What are all the proteins like? What are all the protein programs running at that time? A lot of really cool tech that generates massive amounts of data. And it takes a lot of work to organize that. And the quick gist is that the experimental design is constantly changing.

So every time you experiment, it's different, so the data format also changes. It's like a mundane problem, but it just kills people. So you can only agree on a schema for your data set after. At the same time, you need to leverage all this information to make an informed analysis because it's so much information.

So the whole point of ADA was that we could use large language models To extract relevant information from all these unstructured text documents. People can keep working in their messy Google folder collaborations where things are all over the place and in random spreadsheets and random notes to each other and emails and Yada. Still, we could synthesize and digest all of that. Put it into a knowledge graph, a more flexible type of database that exists as a subject-object predicate triple.

It's like I have brown hair. You have a blue shirt, whatever these things, Samantha did this experiment. John did that experiment, something, and that would be the database. So it's super organic. It's apt to be adaptive. You can incorporate a lot of different knowledge sources into that.

That was the kind of dream that the startup had; you learn a lot trying to go now into the sales process and working with other companies. You learn about market dynamics and what people like to spend money on and what they don't. The Pharma industry it's 15 big companies that like to spend money on new diagnostics, new ways of getting data, and then new ways of analyzing it.

Like simulations include. So the productivity tool was a tough sell, but I learned a lot in the process. You learn a lot by trying that kind of stuff. 

Eric Jorgenson: You decided to jump back into nuclear after that.

Andrew Cote: Yeah. It was a great learning experience, and many startup failure mode analyses are always complex. Still, I learned a lot, and it's always there's proximate, and there are ultimate causes. So then, after that, I was trying to think about what was next, and I was bummed out.

Like the startup, it didn't work. That sucks, and then I thought, okay. I always liked fusion technologies and wanted to return to California. I was in Canada at the time. I had a storage locker full of stuff down here. So I want to get back to my life right in San Francisco.

And yeah, I joined a small company in Redwood City that was building called beam-driven fusion devices. You're just accelerating charged particles and causing fusion to produce neutrons. And that was for treating like it's a sporadic form of early childhood cancer that affects the brainstem. It's inoperable.

What you can do, though, is design an adapter molecule that binds to those tumor cells and carries boron, which has a favorable reaction with neutrons. It's called a capture cross section as a good capture cross-section. So it'll react with neutrons, form an unstable intermediate isotope, and then radioactively decay.

And when it decays, it'll emit high-energy particles that destroy the surrounding cancer tissue. So that was cool. That was like a 20-person company like R and D tech selling that stuff, primarily to other labs. And that was fun because I got to be an engineer again, building stuff with my back to save radio frequency engineering, physics, etc. 

Yeah. So I had just my two-and-a-half, three-year biotech tour of duty. Detour of duty, then I'm back to the physics stuff, and I was there for a while, which was a lot of fun.

And then someone I worked with at John Fusion got in touch and said, " Hey, I'm, we're starting a new fusion company. It's based out of Princeton. We're going to make stellarators. I was like, Oh, that sounds pretty cool. So they have signed me up for that. 

Eric Jorgenson: That's awesome. Do you consider yourself more engineer or more scientist?

Andrew Cote: That's a great question. Usually, scientists have PhDs, and I was like, not, I'm always, I was like debated, Josh, I get the Ph.D. or not. And so I didn't write. And s I can't call myself that. Then again, I've met lots of people. I met many people at Stanford linear accelerator with the job title engineering physicist.

And I was like, Oh, that's cool. How'd you get that job? My degree is called engineering physics. And that's. The name of my degree, too, right? I've had that job title before. It's a badass job title because it's ambiguous, but I'm not a real theoretical physics guy.

And I'm way more of an engineer, but I've always worked in experimental scientific settings. So you get, you are in that environment anyway. One thing I've done, I have yet to optimize for many publications over time. I now have a few patents and stuff like that, but only a few academic journal submissions. So that's scientists do that, right? So I probably identify more engineers a lot, but I like to research and develop engineers in these experimental capacities. 

Eric Jorgenson: That's an attractive academic scientist versus scientists.

The general populist scientist is like a fascinating distinction, and I think like an important one to make, right? Anybody who's like developing new stuff or replicating things. It's a fine line. It's not just theorists with PhDs who are scientists, right? 

Andrew Cote: I see that this is funny, right?

Yeah, this is a thing you see often in these kinds of. Fields where it's, are you good at math? Like you asked me. And I think, Oh no, man, like my buddy from school, he went to get a Ph.D. in pure math at Oxford and is now on the road to being a math professor like that guy's good at math.

So I'm probably not that good at math, but generally speaking, I'm closer to science than many folks reading journal articles and thinking about them. So in that sense, science is adjacent, but. Scientists, it's like you're running a lab, your principal investigator, you have research grants, you have postdocs, you have grad students, you have this, to me, the kind of science archetype.

Eric Jorgenson: So yeah, interesting, that's just a context definition for you to me. From my perspective, you're critical. You sit in the middle of you're the significant overlap of the diagram of engineering and sort of science. Meaning that you are the person who develops new technology and pulls something from the realm of barely possible into, Hey, look, we built it.

And the more people that do that, the faster and the better they do it. That's where improvements for civilization are actualized in that process. 

Andrew Cote: Oh, Yeah. That's I'm flattered to hear that. That sounds great to me. I would like to know if I experienced it that way.

I would say. To solve engineering problems, I'm frequently reading things in research literature in the last few years. So there's this translation application component, which is excellent. Other people do the hard work of discovering the thing, and I just read it, read the results.

Eric Jorgenson: It's a bit like the cook and the chef. There's probably one for every 100 engineers, and there's probably less than 10, maybe less than five. They're on that frontier of trying to engineer something that has never been done before versus like we're trying to build a small engine with a slightly different mix of trade-offs and scale than has been done before.

You're still close to a frontier in the broad engineering world.

Andrew Cote: Yeah, I like to think like that, too. That's for me always been what's interesting. And that's the strength of this kind of engineering physics education. And that's the actual name of degrees called engineering science.

But the whole point, the whole kind of ethos behind there, there's a lot of applications where the design rules don't apply because it's not as well studied if you're designing like a gear train, right? There are many design formulas to follow that will predict failure stresses, failure modes, lifetime reliability, etc.

And so it's like. The whole point is, what are the physics principles behind those design rules? So I can extrapolate to application areas that haven't been tested before. So that's where it's like understanding the physics. You don't need to push the frontier of physics and string theory.

And that's a different skill set, but understand the physics intuitively so that you can operate with a design intuition and an eye for good design in a challenging environment with many things going on simultaneously. High electrical currents, magnetic solid fields, super high temperatures, cryogenics, radioactivity, like this is fusion, right?

It's 5 or 6 physics disciplines on their own. Like they're all specialties, so you must learn enough to be valid. You got to learn enough to know the 1st and 2nd order considerations, which is fun. That could be one thing. I've always enjoyed jumping into different topics, diving into them, and seeing how they work.

And then the third part of the science engineering thing is understanding the market dynamics that can determine whether your amazing breakthrough gets used by people. And it's part psychology, part economics, too, right? 

Eric Jorgenson: Yeah, I was going to do that segue because you had another great tweet that I pulled in here to my notes.

Economics is where psychology meets physics. It would only be easy to understand the trajectory of technology with all three.

Andrew Cote: The point I was trying to make there was this. Economics is the transmission where the wheels meet the road. Our driving society through the world has rules of governance, and we have the way we want to work things and how much parking should cost, but also what, yadda, yadda, right?

And we have to interface with material reality, eat food, be warm, and produce clothes and stuff like that. And so economics is that transmission where we have, or it's the engine and transmission, if you drive badly, you'll crash into a tree, and societies have driven badly in the past.

If you look at them, that's like space lasers and tank armies didn't win the cold war. Full supermarkets and grocery stores won it. And Michael Jackson, eight tracks and blue jeans kind of stuff, right? Like the productive wealth of the economy when it's under stewardship, it is well adapted to providing for its people.

And that's sometimes it comes down to the physics of stuff. You must produce more energy than you consume, just like the simplest. That's why I'm so pro-nuclear. You'll get energy ROI, like how many megawatts hours to produce a megawatt hour or a megawatt hour of solar energy.

It's 4 to 1, and nuclear is 50 to 1 or 40 to 1. It's like same with materials costs, all this kind of stuff. It makes sense on that basis. So there's a physics fundamental to economics. Still, there's also this weird psychology behavioral component where people try to think about what's fair and proper.

We want to live in something other than a physics-first universe. It's pretty cutthroat, and it looks a lot like a sort of fascism or totalitarianism or something like that. It looks terrible. So you must understand charitable compassion and acknowledgment of different circumstances and everything.

So there's always a balance of yeah, this kind of, I'm not going to get the politics of that. Still, technology; has to be physically viable and produce economically sound functions, but sometimes it doesn't like beanie babies, right? What's the ROI on a beanie baby versus the fusion reactor, and why do people spend a billion dollars on beanie babies and not fusion energy?

Eric Jorgenson: It's hard to quantify joy, but that's what we have orbited like my current rough draft of the thesis of this show, which is like with technology, capitalism, and friendship like all will be well that's what makes the world go around. That's what makes life good and worth living.

Andrew Cote: I agree; Man, a quick comment on that. I used to study sociology, and it was all Marxism all the way, like all Marxism all the time. And it gives you a certain mindset and is very critical of capitalism; there are many things to improve. But now I think of it like, what is the profit incentive?

Sure, it has many forms, but it's the fundamental physics equation that the total value coming out of an activity area has to be bigger than the value of stuff going in. And so long as that equation holds, the value of stuff increases over time, and that's just so basic. It's so simple. It only explains some profits; only some of the right things are profitable according to specific rules.

But that's what I think of as a good principle.

Eric Jorgenson: And things get muddier with subsidies and things like that, too. So we've made a very complex system out of a simple fundamental formula. 

Andrew Cote: But you can only get to the right places sometimes with subsidies to overcome this initial; only some technological enterprises are Q1 profitable.

And most of the best ones aren't. 

Eric Jorgenson: Yeah. I'd like to know if I've heard of one yet. 

Andrew Cote: Yeah. Beanie babies were profitable.

Eric Jorgenson: I have yet to study that as a technology enterprise. 

Andrew Cote: I wonder if that's the term. I just made it up, but is it? Does it make money? Cause fusion is a 10-year bet, It's a ten-year bet. So 2035 is when all these companies try to get on the grid. 

Eric Jorgenson: Yeah, and that is a realistic timeline for you as a front-line engineer.

Andrew Cote: looking at the performance improvements over time in the field. If you draw the line out, then it's reasonable in terms of magnetic confinement energies, that kind of stuff. Different approaches to fusion have different degrees of scientific risk. Magnetic confinement has just engineering risk. It's a matter of design and testing, but the science is way more de-risked than other fusion branches. 

Eric Jorgenson: I know we had some big headlines over the last couple of years about fusion, but it's still the non-tech-focused people I talk to are, oh yeah, that's not real, or it'll never happen.

And so it was one of those things; it's always 20 years away. We turned it on for a millisecond at net negative energy costs. It's not quite, but we need to see the progress you see from the outside. Ten years is incredible. And this could be a fantastic couple of decades.

Andrew Cote: Oh yeah, for sure. And it's such a long-run bet the first reactors will be expensive and challenging to produce. But what you're doing is you're getting an unlimited fuel supply forever. You get the fuel from seawater and dig it out of the ground. A lot of the fuel types are very abundant materials.

And they have a similar energy density as stuff like uranium and plutonium, which are toxic, hazardous, and harder to extract. So it's like the last energy source humanity will ever need. 

Eric Jorgenson: Yeah, Fusion is; people talked about the holy room temperature semiconductors being like the holy grail of material science and super superconductors and fusion, like the holy grail of energy production. And I don't know, like we can keep riffing on the, like the next industrial revolution forming between these things, my summary, trying to wrap my head around this in the big picture of sort of civilization, societal, technological changes like if this is real, which is I think we can come back to the current state of a step forward in superconductors means it unlocks things that are currently constrained by energy cost.

And it unlocks things that are currently constrained by computing costs. Both of those and we can talk about why, are much broader than people think. 

Andrew Cote: I'd add a third: it unlocks currently impossible things. Like, you can't do it. You just, it's just, you can't even do it.

So it's part of it is everything gets better and is cheaper. It's like unbelievable; it's going to use fewer materials, it's going to run faster, higher performance, it's going to use less energy, it's going to be cheaper to make, okay, all four of those things that you usually can pick two or something, and it's going to enable fundamentally new things that you can't do otherwise.

Holy crap, right? So that's why it's like a real, like a, yeah, it's a LK 99 five spice. Really? There were five categories of improvements. 

Eric Jorgenson: That's something else. It is one of your tweets on this topic. Not realizing how much is impacted by decreased energy costs.

You said half the input cost of materials like aluminum is energy. 

Andrew Cote: A lot of times. So one thing about aluminum, it's almost infinitely recyclable. So you can, yeah, that's good. Still, the extraction of ore and the refinement is incredibly energy intensive, and the forming and manufacturing process.

So energy is one of those things where it's like. It's like the base price of something. If you could think of it this way, think of a high throughput factory that achieves perfect economies of scale, meaning the cop, the capital costs per unit produced the zero are somehow amazingly impossible.

But suppose it's all robot, fancy stuff, whatever von Neumann probes assembled themselves out of the luminous ether of our imagination and formed this factory producing beanie babies. And it's fantastic, right? The long-run cost was beanie babies. They're going to be two things. It's me, materials, raw materials, and it's the energy, right?

The price floor on stuff is the energy and material costs, right? And now, looking at just materials by themselves, like they have extraction refinements. So many metals; it's like a double-digit percentage of the price. When you look at it is the price of energy. And it's tough. It's tough because our energy economy is now susceptible to these occasional wild price shocks due to geopolitical events.

And it drives damaging cycles in the economy. It makes it hard for consumers. Inflation is so regressive, right? Because it's assets that get inflated relative to dollars. And most people aren't rich in assets. Only some people, only wealthy people, have tons of assets usually.

So devaluing cash relative to assets and energy inflation, it's one of the great destroyers of the middle class, is this inflation stuff? And it's weird because energy has gotten cheaper a while ago. What's up with that? How are we as a species supposed to be crushing it, and energy is getting more expensive?

Eric Jorgenson: Yeah, which gets too close to the political fire because we said we wouldn't do too much of that.

Andrew Cote: Oh, that's a noticeable man. That's not politics. That's physics. The physics says don't be an idiot.

Eric Jorgenson: That's a good bumper sticker. So many people believe that more significant energy usage is wrong and humans should have less access to energy because it's terrible for the environment. That's what a lot of environmentalism has turned into: make energy more expensive and generate less without realizing we have.

They have also slowed the creation of renewable clean energy that we now have all the engineering to do. And that the more energy there is to your point, there's no more fundamental measure of abundance than the cost of energy. The more energy we have, the capacity to generate, the cheaper it will get, and the cheaper everything else will get absolutely everything else.

Andrew Cote: A hundred percent. Agree. I wrote this little essay thing. It's on like a sub stack. If no one, if anyone reads that, whatever, but it's this kind of the history of energy. And it's like, when did the population boom happen? And when did our quality of living get a lot better?

It perfectly coincides with when we started having access to more energy, and if you think big picture, look, fossil fuels is a one-time subsidy for our industrial economy. And we don't use that to get to an abundant energy future. In that case, we will have squandered that inheritance on political favors and bickering over small fry stuff.

And it's not that social problems are small fry. I'm sorry. They are essential, and we should have those discussions. But we are burning the candle at both ends as if we run out of cheap energy. We have yet to get to a new abundant, tremendous source, nuclear fusion, whatever it could be, orbital solar arrays, or unlimited geothermal.

We're screwed, man. 

Eric Jorgenson: It lights out figuratively for all of us.

Andrew Cote: Yeah, man, that's right. Everything gets worse. Like quality of living so directly proportional to energy consumption. 100%. 

Eric Jorgenson: If you have any off the top of your head, I would love to know when you added a category just like things that are entirely impossible today.

What are those? I'm intrigued.

Andrew Cote: Yeah. Okay. So this is where we go back to that very initial thing. Where every product is a combination of available things, right? And so the number of new things you get is combinatorial in the capacities you have. So we're good at looking at all the first-order applications of superconductors, energy transmission, cell phone sensitivity, and computing speed.

And it's hard to determine the second and third-order applications. That would be enabled down the road. So look at transistors. It's a really, and there are many analogies here in some ways where transistors were thought about a lot before they came out and the same with room temperature.

But in transistors, before they were invented as a material at Bell Labs, it was like, okay, we use vacuum tubes, and they suck. They're big and bulky, and they break all this stuff. It takes a lot of work to tune the performance. They could be better. And the vacuum tube transistor, it's just a Current switch or a current amplifier.

Can I use a small signal to control a significant signal? So it's like a valve on a pipe. And there are many ideas of what we could do if we had this thing. Oh, we can make all these relays and switches shorter. We can make an automated telephone system or something like awesome stuff.

But you can't predict it. Large language models like VR headsets or this kind of stuff, phones, the internet. Yeah. We had phones before the transistor, but I know you're saying smartphones. It's super funny.

So this is one reason I'm a huge science fiction fan. And I think that's been. Why I got into the things I did because it was like the chance, however, small to work on something that's sci-fi is so crazy. 1911, this French guy had this cartoon, and people have iPhones and have iPhones with video screens talking to people, except they have the little earpiece clamshell thing held to their ear, right?

They didn't have the AirPods. They had the video. They had FaceTime without the AirPods. 

Eric Jorgenson: But it needed to be visible in cartoon form, so that makes sense.

Andrew Cote: Yeah. It's like digestibility. People aren't going to get that wireless headset, but science fiction is fantastic.

Because it's trying to imagine boldly, like what are the near-term second, third order impacts realistically are change the boundary conditions of society, throw in some weird tech. How does everything else readjust and change and stuff like that? And so they've been very anticipatory of future developments.

And you look at cyberpunk and the metaverse. Arthur C. Clark, a sci-fi author, first wrote about tie-ins, the personal computing revolution, and even telecommunication satellites. So there's a long history of this kind of stuff. And yeah, superconducting is the future. We can get there.

When it happens, it's such a worthwhile goal, and it's also something that we can build the tools to get there faster in computation and experimental throughput. And so there's enabling text. You don't have to jump the climb the wall all at once.

Eric Jorgenson: So on the topic of 2nd and 3rd order effects, it's also hard to imagine sometimes, not just It's easy to imagine replacing existing technology with new technology. It is hard to predict the second-order effects on the economic side of things, like how market sizes change when things take one or two or three orders of magnitude leaps in either quality or price or Anything like that.

So the exponential can take off quickly in the feedback loop between these co-developing technologies. It's a very unexpected way when the feedback loops. We saw this in a great book called Where's My Flying Car. I did an episode with the author, and my book notes as a separate episode.

And he did an excellent job of comparing the previous industrial revolutions and the feedback loop between internal combustion, steel, gas, railroad, telegraph, and electrification, all co-developing with, Hey, we're going to have nuclear, we're going to have AI, which is going to accelerate the development of nuclear, the decrease in the cost of energy and AI are both going to bring nanotechnology much closer, much faster than it had been possible to do before.

 Superconductors are another, it's something I don't remember him talking about specifically, but it's another kind of lever that unlocks stuff but compute. AI might unlock the superconductor, which helps us unlock nuclear fusion, which helps us unlock the nanotechnology, which is just Material of God, like power over all material things, which is just insane.

Andrew Cote: It's like Civilization Five, like endgame technology.

Eric Jorgenson: And it's to your point earlier like that seems like absolutely impossible magic. We have no precedent for imagining that, and it's harder to imagine. It's hard for a lot of people to imagine, it was easier to conceive of air conditioning as an invention that might happen in your lifetime than to believe that we're going to be able to will air into a bowl of food and send rockets to Mars-like that seems insane, but it's not like we could see it in our lifetimes if the proper steps happen in the correct order.

Andrew Cote: Yeah, it's true. So there's this great concept from biology called punctuated equilibrium, which is tons of species will evolve out of some big event, like the KT extinction, the Permian trance extinction, the Cambrian explosion, whatever. And you'll have all these new creatures emerging simultaneously, based on some big upset.

And then things become more static. It's like niches get developed and minor performance improvements and different animals. Business models, something else comes along, and everything changes again. And there's a vast reshuffling, right? So I'd recommend another book, The Structure of financial capital and technological revolutions, by Carlotta Perez.

Eric Jorgenson: I almost thought it might have had an arm's reach, but that is an incredible book.

Andrew Cote: Yeah. It does an excellent job of tracing out. I agree with the book in large strokes. Look, one revolution enables the next, and now you have railroads, cheap freight.

What does that enable? Okay, like International, intercontinental shipping and specialization and more urbanization and, everything builds. And it's enjoyable to think about what are things enabled by the most recent revolution, right? The I. T. information revolution. So one of them we're seeing already, which is biotech.

And so DNA synthesis and sequencing, just a quick story on that. So the human genome project was like 2 billion to sequence the genome. And they were using a sequencing technique called Sanger sequencing, which was expensive and slow. We're reading a thousand-page novel one line at a time.

Okay. And now Illumina has this new sequencing, which it's called; what is it called again? I need to remember the next-gen or something. As a broad class of ideas, but it's fundamentally different. Instead of reading the book in order, right? You have a thousand copies of the book, chop them into random sentences, and read millions of sentences in parallel.

And then, you look for the joint overlap with high computing and reconstruct the whole book. And so that's enabled by computers. Computers are fantastic, so biotech is built on top of the computers. Let me put my cards on the table.

Biotech is one of those things where we think medicine is excellent and diseases and get a hair loss treatment. It is important. I don't mean to say whatever. That's awesome, and it's a programmable matter. It's stuff that grows. It's a von Neumann probe like.

What's the difference between a tree seed and a von Neumann probe? It's like you just put it in the ground. It unpacks instructions that are the size of a molecule into a t complicated thing, something more complicated than anything we could produce. That's a tree, let alone a person and a brain, and so forth.

The ability to design in DNA is the programming language of molecules and the molecular world. And so the program will matter. We can invent nanobots the hard way, like reengineering them from semiconductors. And that's tough. That's tough.

But evolution has this 4 billion year head start or 3 billion year head start of exploring the possible combinations of molecules. And it's done so massively parallel. In all these different little tide pool experiments, we've come up now with an optimized, complicated byzantine Rube Goldberg ask series of interactions and things in the body like, oh, this one molecule turns your eyes blue or causes you to forget something like what, like hell, like all the that's a made up example.

It's so weird. These pathways are so interleaved with each other. So multiplex, there's so much economy of use in a single molecule in the body and all places it can manifest and show up.

And that's why diseases are so weird, where it affects some small cellular mechanism. And people always get a cough and a rash on the back of their neck, and their knees start hurting. And you're like, wow, okay, those are all related, but having a factory that grows out of a pumpkin or a colony that unpacks from a seed chip.

And it just starts growing a thousand years in advance, like Expanse is great Sci-fi. Because it's like, what would happen to humans if you had Fusion rockets? You can do lots of stuff. And then let's think bigger. Why not just have von Neumann probe life that you send to a planet, which unpacks into an orb gate or something?

I just ruined the first season. I'm sorry. But biology is underestimated in its ability to be impactful beyond just life science and agriculture. Those are the highest-leverage applications now because producing these molecules is so hard. And so you need to have a million dollars per kilogram kind of price or a million dollars per gram, which is valid for antibodies because the effective dose is so tiny; it's like micrograms or something would impact someone. So whatever, but in the long term, like learning to natively speak the language of biology through synthetic life for synthetic living organisms. And then your computer heals itself.

Eric Jorgenson: Yeah, I agree with you. However, I want to see us pursue both paths toward self-replicating nanotech stuff. There's no reason not to, But the bio stuff is fantastic.

Andrew Cote: Okay. Okay. I can think of 1 reason. The gray goose scenario is just rapid.

Yeah. 

Eric Jorgenson: Yeah. I do Marism in nanotech. It's the gray goo. Yeah. What is the current verdict on the superconductor, like the breakthrough? How much of a breakthrough do we have? Did we kick off a long cycle? Is everything immediately different? It has been like a rapidly developing situation.

We're recording, probably a week before we'll publish, but just your rough estimate and go forward with prognostication. 

Andrew Cote: So the sentiment has shifted overall through in the public community, as well as the sort of academic research community, and academics are a lot better at being reserved, reserving judgment.

So all the facts come in because that was only a little happening. That's a virtue. It's true. And I'm likely to blame for that, but I always was trying to say, Hey, by the way, this is not confirmed. 

Eric Jorgenson: No, you did a fantastic big-picture balance.

Andrew Cote: But it's why this is important. Why do people care? How does this stuff work? That was part of the story. The initial weaknesses that the good, bad, ugly, the thing I tweeted, they just never got strengthened over time, like that no one, there was never a replication that came in that really oh, that measurement that was missing or wrong.

Yeah, here it is. We nailed it. There we go. So just never the longer those things didn't fail to materialize. And the most obvious one was zero resistance. That's the biggest deal. Lots of different stuff floats, right? Only some things good float. Not all things that float are good.

And so yeah, you know, the lack of replication confirmation of those findings in the material experiments And then now and that so the waves of optimism, right? So the first wave Of reactions or simulations, which did point toward this electronic band structure, was promising and was, in the author's own words saying, look, this is thought to be enabling super connectivity in a few different interaction pathways like a few different mechanisms could be related here.

It doesn't predict anything, but it's not really about it. So those were the first things to get done that was the easiest. And then the next was like the visual cues, like videos of it levitating. So those were both positive indicators. They did not deny it so that optimism could increase.

But then the stuff that takes longer to get done, like the physical measurements and the physical replications, started to come out, and they needed to be more promising, like they were not confirming the thing. And then the more difficult thing to get done was Measuring what's going on and figuring out what's going on.

So now those are coming out, and they have a much more mundane explanation of effects like this. You thought the resistance drop was just a phase transition in the material. So the resistance did change but never to superconducting levels.

There's still science to be done in this whole theme. So it's a done deal for sure, but when you think of your, my long tail odds of 5%, 10% being authentic, and that's 10% of a huge deal. And that's shifting back down to one person, 10, or less. I don't say number to tie myself to it, but it does tiff the relative expectation value. Some exciting things surprised people who did the simulations and experiments, right? One was that this doping effect of introducing copper into a lead crystal caused a structural deformation in the crystal. That deformation produces these otherwise absent favorably flat energy bands near the Fermi surface, like the sea level of energy in the material.

That's an excellent effect. I am still determining how well understood that was before. I'm not a condensed matter physicist, but that may be partially anticipated. Another thing is, here's this material that, at room temperature, undergoes this phase transition. Materials engineering could be done to leverage these effects in a different material substrate.

These might become parts of your toolkit for thinking of new materials and might not be superconductors. It might be other things that juries out on, whether this leads to fruitful research efforts. It could be bittersweet and fruitful.

People explore it and come up with new stuff. Hey, here's this excellent transistor design. That's better now. We didn't think we'd get there, but. We got there by accident, and it might also distract people from what they were doing, which was more practical. Yeah. But that's always part of the game, right?

It's always this random walk-through of possible experiments and semi-random, right? It's informed. So I'd like to see if those things led to helpful research or if it was just a., This was a false full of school or something.

Yeah. 

Eric Jorgenson: So we added a leaf to the tree of knowledge, indeed, and hopefully, it's something that points us in the direction of something that's like a lot more fruitful, but it doesn't seem like this was, oh, my God, we can make cheap room temperature superconductors now.

Andrew Cote: Not all these leaves catch the sunlight, either.

So whether it's the sun's going to shine on this leaf.

Eric Jorgenson: Okay, where should people go to get more of you to follow and learn from? 

Andrew Cote: Yeah, sure. I sometimes tweet threads about cool stuff. It is under development. People might not have heard about it and try to tease it out. Hey, what's the science here?

What's the engineering application? And then what's the big deal? What's why does it matter? That's a fantastic trifecta. So I'll try to write more on Twitter as, like, here's the overview, right? Kind of the topic. Here are the key points, and then I might do more longer pieces on the science behind it.

So the subset part of the section is called the physics of industry. And just the first extensive article is all about the fusion energy industry. So really, honestly, here's the shameless plug. If you want to get up to speed with fusion, read that article because it starts here's the first part history of energy and why it matters.

Where it comes from. The second part is an extensive overview of all the different types of fusion in general. Some specific examples, the companies working in them, what are the technological hurdles to those things? And what are the pros and cons of different approaches, and how does it work?

That's a vast topic. And so you can't cover it in exquisite depth. So an expert might be frustrated that some points are crossed over. But it's trying to be an overview for someone literate in science and technology but different from a fusion expert. So it's what's the first-order intuition for things.

And on your podcast again. We'll see. 

Eric Jorgenson: That'd be awesome. Yeah. I've had a ton of fun doing this. Do you want to, in case you need to be more of a Renaissance man between bio and fusion and everything? You also host AI salons in San Francisco for people.

 

Andrew Cote: Yeah, that's right. Totally. Okay. So I've had a couple of little startup attempts, and they're really, it's always easier to start software companies, right? It's the laptop and an idea. And some Red Bull and solar and subscription.

So AI salon it's Yeah, what are the second, third order social and economic and sometimes philosophical impacts of AI, as it's currently manifested in large language models, and how we think it might get in a few years or even a longer time horizon. And, been a lot of interesting discussions around that.

So we've had ten sessions so far; 100 people have come to the doors on that. And the format it's a single-threaded conversation. So just eight, 12 people to hang out usually in my living room, the space, but sometimes we get more significant events, 60 people having several conversations at once.

And people talk about how it will affect dating relationships, right? And loneliness and depression and therapy, people will want to marry their AI, whatever's or, healthcare and medicine or science. And so different topics every week, it's fun. Hobby, right?

My friend Ian Eisenberg and I run it. So the next session might be too late, but it's transhumanism, so AI ascendance. Okay. So yeah, I linked that on my Twitter bio too. So anyone's welcome to join; sign up. We get a good mix of people each time. And yeah, we'll keep doing that for fun.

Eric Jorgenson: I will put your Twitter front and center in the show notes. So people should follow you, taking a tour through all your threads, and your pin tweet is Thanks. It's a perfect time. There's all kinds of cool stuff in there. Some of it we touched on today. For some of it, we didn't even use super hot plasma lasers to dig deep, super deep geothermal things through the earth and excellent thread on nanotech. Yeah, I enjoyed the hell out of this.

So nuclear rockets. Yeah. I would love to have you back, and we can just run through whichever, whatever you've been tweeting about, whatever you're curious about because there's a lot of fun, a ton of this. Thanks for coming. Andrew. Appreciate you. It has been a lot of fun. 

Andrew Cote: Thanks to you.

TechEric JorgensonNuclear, Superconductors, ScienceComment