Artificial Intelligence Isn’t Ready for Mass Application

Image of the open AI logo with a wireframe brain above it

Today’s AI technology, while promising, isn’t quite ready for widespread application. I’m not talking so much about AI’s capabilities, but rather the hardware limitations and supply chain challenges that are getting in the way.

For AI to manage vast amounts of data, it’s going to need specialized chips which are still in development. So, give R&D a couple years to figure that out, and then another decade+ for production and supply chains to get sorted out. Without these new chips, power demands are going to skyrocket (because the current, inefficient chips suck up power like nobody’s business). Until those new chips arrive, the US will have to decide which industries will be getting the limited chips that are available, like agriculture, defense, or finance.

While a delay might seem like a bad thing, especially for those who are ready to let AI do their job while they’re sipping Mai Tais on a beach somewhere…it gives us time to figure out how to address all the problems with AI and what its actual impact will look like.

Prefer to read the transcript? Click here

Here at Zeihan on Geopolitics, our chosen charity partner is MedShare. They provide emergency medical services to communities in need, with a very heavy emphasis on locations facing acute crises. Medshare operates right in the thick of it, so we can be sure that every cent of our donation is not simply going directly to where help is needed most, but our donations serve as a force multiplier for a system already in existence.

For those who would like to donate directly to MedShare or to learn more about their efforts, you can click this link.

Learn about medshare here

Transcript

Hey everybody. Peter Zeihan here. Coming to you from Revere Beach, just north of Boston. A lot of you have written in asking for my opinions on AI. So here we go. Pick it apart, however you will. It’s tantalizing. So GPT and the large language models are taking us forward. They’re nowhere near conscious thought. Oftentimes, they can’t even associate their own work from previously in a conversation, with itself. 

It’s basically targeted randomness, if you will. That said, it is still providing insights and the ability to search vast databases in a much more organized and coherent matter than anything we have seen from, search engines before. So promising tech. We had a taste. It’s definitely not ready for what I would consider mass application, but, the possibilities are there, especially when it comes to data management, which, when it comes to things like research and genetics, is very important. 

However, I think it’s important to understand what the physical limitations are of AI, and that is a manufacturing issue. So the high end chips that we’re using, the GPUs, graphics processing units, we’re not designed to run AI models. They were designed to run multiple things simultaneously for graphics, primarily for gaming consoles. And the gamers among us who have logged lots of time playing Doom and Fortnite and all the rest have been the primary economic engine for pushing these technologies forward until very recently. 

It’s only with things like autonomous driving and electric vehicles that we’ve had a larger market for high end chips. But the GPUs, specifically because they run multiple scenarios and computations simultaneously, that is what makes a large language model work. Wow. Got windy all of a sudden. Let me make sure this works. 

Okay. So, GPUs, they generate a lot of heat because they’re doing multiple things at the same time. And so normally you have a gaming console and you have a GPU at the heart of it, and multiple cooling systems typically fans blowing on them to keep laptop from catching on fire. 

So if you take these and put 10 or 20,000 of them in the same room in the server farm, you have a massive heat problem. And that’s why most forecasts indicate that, the amount of electricity we’re using for data centers is going to double in the next few years, to compensate. That’s why they’re so power intensive. 

Now, if you want to design a chip that is for large language models and AI systems as opposed to, that’s just being an incidental use. You can that those designs are being built now, and we’re hoping to have a functional prototype by the end of calendar year 2025. If that is successful, then you can have your first mass run of the chips enough to generate enough chips for a single server farm by the end of 2027. 

And then you can talk about mass manufacture getting into the system by 2029, 2030. So, you know, even in the best case scenario, we’re not going to have custom designed chips for this anytime soon. Remember that a GPU is about the size of a postage stamp because it’s designed to be put in a laptop. Or if you’re going to design a chip specifically, to run AI, you’re talking about something that is bigger than a dinner plate because it’s going to have a cooling system built in. 

Not to mention being able to run a lot more things in parallel. So even in the best case scenario, we’re looking at something that’s quite a ways out. So then you have to consider the supply chain just to make what we’re making. Now. The high end chip world, especially sub10 nanometer, and we’re talking here about things that are in the four nanometer and smaller range, closer to two, really, is the most sophisticated and complicated and, proprietary supply chain in human history. 

There are over 9000 companies that are involved in making the stuff that goes into the stuff that goes into the stuff that ultimately allows TSMC to make these chips in Taiwan. And then, of course, 99% of these very high end chips are all made in one town in Taiwan that faces the People’s Republic of China. So it doesn’t take a particularly egregious scenario to remove some of those 9000 pieces from, the supply chain system. 

And since roughly half of those supply chain steps are only made by small companies that produce one product for one end user and have no competition globally, you lose a handful of them, and you can’t do this at all until you rebuild the ecosystem based on what goes wrong. That rebuilding can take upwards of 10 to 15 years. 

So in the best case scenario, we need new hardware that we’re not going to have for a half a decade and are more likely scenario. We’re not going to have the supply chain system in order to build the hardware, for a decade or more. However, we’ve already gotten that taste of what I might be able to do. 

And since with the baby boomer retirement, we’re entering into a world of both labor and capital shortages. The idea of having AI or something like it to improve our efficiency is something we can’t ignore. The question is whether we’re going to have enough chips to do everything we want to do. And the answer is a hard no. So we’re going to have to choose do we want the AI chips running to say, crack the genome so that we can put out a new type of GMO in the world that’ll save a billion people from starving to death. 

In a world where agricultural supply chains fail. Do we use it to improve worker productivity in a world in which there just aren’t enough workers? And in the case of the United States, we need to double the, industrial plant in order to compensate for a failing China? Or do we use it to stretch the investment dollar further now that the baby boomer money’s no longer available and allow our financial system to be more efficient? 

Or do we use it for national defense and cryptography? You know, these these are top level issues, and we’re probably only going to have enough chips to do one of the four. So I would argue that the most consequential decision that the next American president is going to have to make is about where to focus, what few chips we can produce and where do you put them? 

There’s no right answer. There’s no wrong answer. There’s just less than satisfactory answers. And that leaves us with the power question. Assuming that we could make GPUs at a scale that will allow mass adoption of AI, which we probably can’t anyway. You’re talking about doubling the power requirements, of what is used in the data space. Here’s the thing, though. 

If we can’t make the GPUs and we’re not going to be able to make the more advanced chips anytime soon, we’re still going to want to get some of the benefits from AI. So we’re going to use older, dumber chips that generate a lot more heat per computation in order to compensate, which means we’re probably going to be seeing these estimates for power demand, not simply double, but triple or more. 

At the same time, we get less computations, fewer computations, and generate an AI system that’s actually less effective because we’re not going to be able to make the chips at scale. So is it coming? Yeah. But in the short term, it’s not going to be nearly as fast. It’s going to cost a lot more. It’s going to require a lot more electricity. 

And we’re probably going to have to wait until about 2040 before we can design and build in mass and apply the chips that we actually want to be able to do this for real. So, believe it or not, actually see this as a borderline good thing because it’s so rare in the United States that we discuss the outcome of a technological evolution before it’s completely overwhelmed us here. 

I’d argue we’ve got another 15 years to figure out the fine print.

New Orleans Terror Attack

Photo of Bourbon Street, New Orleans

There was a terror attack in New Orleans today. An American citizen, who has pledged allegiance to the Islamic State (despite its largely defunct state), drove a truck into a crowd; killing at least ten and injuring several others.

Prefer to read the transcript? Click here

Here at Zeihan on Geopolitics, our chosen charity partner is MedShare. They provide emergency medical services to communities in need, with a very heavy emphasis on locations facing acute crises. Medshare operates right in the thick of it, so we can be sure that every cent of our donation is not simply going directly to where help is needed most, but our donations serve as a force multiplier for a system already in existence.

For those who would like to donate directly to MedShare or to learn more about their efforts, you can click this link.

Learn about medshare here

Transcript

Hey Peter Zeihan here coming to you from New Zealand. It’s New Year’s Day. Back in the States, there was just a terror attack in New Orleans where an American citizen driving a rented truck plowed into a crowd, killing at least ten, injuring a few dozen others. He pledge allegiance to Islamic State, which is, you know, largely defunct in Syria these days. 

 Just want to put this in context. You know what it is? What to expect. 15, almost 20 years ago, the folks that represented the Islamic State at the time called upon Muslims all over the world to rise up and butcher Christians, saying that, go get a gun, can’t get a gun, get a knife and you can’t get a knife, get a car and just do as much mayhem as you can. 

Over a decade later in North America, it’s happened three times. Twice in the United States, once in Edmonton, Canada. Two things to keep in mind here. Number one, training operatives to actually do mass casualty events like, say, 911 takes time, takes resources. And when the people doing the training are in a different hemisphere and don’t control a state, it’s kind of a heavy carry. 

And so it just hasn’t happened very often. And when it does happen, it happens closer to the zone in question, like in places like Paris. Second, and I know this is going to be really radical at their core. Most people are not murderous assholes. So if you have a group who claims to represent all Muslims, who says, everyone go out and do something and you know, there’s hundreds of millions of Muslims and it happens three times. 

Math people. So is it horrible? Yep. Is it the start of something new? Absolutely not. 

One Ship Inspection Could Unravel Global Maritime Shipping

Photo of a ship in a port in Helsinki, Finland

The Nordic nations are teasing a new initiative of inspecting ships leaving Russia’s St. Petersburg port for insurance and environmental compliance. This may seem insignificant, but it could disrupt shipping on a global scale.

Russia has relied on its shadow fleet (uninsured and aging tankers) to export oil after the sanctions began; this new initiative aims to curb Russian income and disrupt funds for the war in Ukraine. But what happens when one of these countries completes an inspection and decides to confiscate or detain one of these ships? We’re talking about uncharted maritime territory…and it could get messy very quickly as countries start to take maritime security and trade into their own hands.

A return to localized maritime security enforcement and controls won’t look the same everywhere. The Western Hemisphere and the Nordic countries will experience some shortages but be mostly fine. The Mediterranean’s future would rely on cooperation between Italy, France and Turkey. Places that are heavily reliant on energy and food imports (i.e. East Asia) could face economic collapse or famine.

Prefer to read the transcript? Click here

Here at Zeihan on Geopolitics, our chosen charity partner is MedShare. They provide emergency medical services to communities in need, with a very heavy emphasis on locations facing acute crises. Medshare operates right in the thick of it, so we can be sure that every cent of our donation is not simply going directly to where help is needed most, but our donations serve as a force multiplier for a system already in existence.

For those who would like to donate directly to MedShare or to learn more about their efforts, you can click this link.

Learn about medshare here

Transcript

Hey, everybody. Peter Zeihan here. Coming to you from the Tongariro Crossing in New Zealand. We finally got a break in the weather, just in time for the emerald pools, which are, you know, volcanic and super stinky. Anyway, on the 17th, 18th, of December, representatives from all the Nordic nations, plus Poland and Britain. So that’s, Norway, Sweden, Finland, Denmark, Estonia, Latvia, Lithuania, basically got together and said that they’re going to start checking ships that are coming out of Saint Petersburg port in the Baltic Sea, for things like insurance and compliance with environmental regulations. 

In order to disrupt the shuttle fleet, the shuttle fleet is how the Russians are getting their cruise to market since they canceled by pipeline, and they can’t use Western maritime insurance or navigation aids. They basically, have to get insurance from the Russian government, the Chinese government or the Indian government. And there’s a suspicion that no one is actually issuing policies that pieces are just sailing. 

They get out of the port, regardless of whether or not they have a policy. Not the Indians, the Chinese and the Russians have never actually paid out on one before. So if there was an accident, the idea is that there would be no one there to help pay for the cleanup. But more to point, the Shadow fleet is how the Russians get their money. 

They basically have uninsured or under-insured old aging tankers, that evade everything that the Europeans are trying to do to shut Russian oil out of their systems, and especially to deny, income to the Russians as they’re launching a war in Ukraine. Will it work? It all depends upon how the Nordics and the poles and the Brits decide to handle enforcement. 

So if a ship doesn’t have anything, what does that mean? You confiscate it, you take it into port. That would be getting into some very interesting territory, that it would be uncharted in the modern age. The whole idea of globalization as created by and then guaranteed by the United States since World War two, is that anyone can sail anywhere at any time. 

And you don’t necessarily have to have insurance, although that’s a really good idea. And if the Russians don’t have it, it’s then up to the Nordics to decide what to do. And if the ships are confiscated, because they don’t have something that is not technically illegal, or they haven’t complied with something that just someone said that they needed to do. 

We enter into a new phase of naval maritime transport. If if, if, if we go down that road that there’s no reason for any country really in the world to not take a ship that they like the look of or don’t like the look of unless they fear the consequences of whoever happens to own the ship or whoever happens to, have the ship registry. 

The registry is a joke. Those are countries like Guinea Bissau and Panama, basically places where you can basically file a web address for free, and officially register your ship if you’re going to replace the system with something that actually means something that has to be a country that has a navy that can actually protect the shipping and are only a handful of countries in the world that can even pretend to do that in a regional basis. 

And only one U.S. that could do it on a global basis. And since the United States does not dependent on international transport for most of its economic strength, it’s a stretch to think that the United States would do that unless it was paid a whole lot of money. And so if the Nordics do this, it is the end of the globalized system as we know it. 

Now, keep in mind that I think we’re going here anyway. There is too much shipping. There are too many powers or too many people wanting revisions, and the US has lost too much interest. And we’re probably going to a naval freefall in the not too distant future. I’m not saying that this is the trigger, but I think it’s time to start talking about what the next system looks like and what the consequences are. 

For the United States, if it happened today, it would hurt. Most of the naval shipments that come into the United States are large container ships carrying manufactured goods from Asia. So we’d have to get by without things like phones and computers and all that. The more valuable stuff eventually would be flown. But for all the bulk stuff, you know, your stereos, your cars, you’re going to be having some problems until that manufacturing capacity is rebuilt in North America, something we’re working on, something that is unlikely to be finished before the end of the decade. 

So, you know, timing matters here, too. For other countries, this would be an absolute disaster. Most of the countries of the world, especially in East Asia, import the vast majority of their energy and material inputs. Some of them are even dependent upon significant food inputs, or at least the inputs they need to grow their own food. 

So if this happened to China, for example, we would easily have a deindustrialization, or collapse, complete with famine in a very short period of time measured in months, not even years. 

What takes its place is probably regional groupings, where either the seas are safe or everyone’s on the same side, and agrees with the rules of the games are, that looks really good for the Western Hemisphere. 

That looks pretty good for the Scandinavian bucket. And in the Mediterranean might get a little dicey based on how relations between or among the Italians, the French and the Turks go. If they agree that they can, work together. That looks great. And if they can’t, You get two different mediterranean’s that shoot at one another, which, if you know your history has happened many, many, many, many, oh, so many times. 

So, I can’t wait until the first time that Sweden or Finland or Poland decide that a ship that’s sailing by their coast isn’t doing something right. I want to see what they do. This is one of those many things that could all fall apart in a day. If the stars are aligned. 

So stay tuned. This is probably not something you’re gonna have to wait for me to comment on, because if it does go down, a lot of things are going to break real soon.