Loading summary
Transcript1 lines
- [00:04]
A
This is your host, Nat Town. You're listening to An American Power Extra, a bonus episode with one of our experts designed to go deeper into the energy, military and geopolitical stories driving the news. If you're looking for more context, more history, and a better understanding of the forces shaping global events, you're in the right place. Take it away, Chad. So is the United States going to have to bail out artificial intelligence companies like Anthropic or OpenAI, as similar to how it bailed out the auto industry in 2008, only this time the government is not necessarily calling it a bailout. They're calling it an investment, they're calling it equity, they're calling it national security, which is kind of a new angle here for this type of thing. I mean, the government's frankly gotten pretty good at branding. But we've already seen pieces of this playbook come forward. Companies like intel, the federal government stepping in not just as a consumer, not as a regulator, not as someone handing out contracts, but as a direct investment investor and financial partner in strategically important private industry. And this is not a subsidy, because that can be confusing. And I understand people may try to link the two. A subsidy is basically the government wants this company or industry to do something so they'll help pay for it. What Trump is doing is the government is becoming a part owner of a company. And this is not the same thing as Thrift Savings Plan either, which is basically a 401k style retirement program for federal employees. That is a passive retirement investment. This is different. This again, is the federal government directly tying itself to companies that sit at the center of the American economy. The president directly authorized taxpayer funding to buy stakes in a company, and that was specifically with Intel. This is something they're kicking around with AI companies. And if you ask me, that starts to sound awfully close to the thing people usually call socialism. The government owning a piece of the means of production. But here's the real question. Why would Washington even consider doing this? I mean, isn't artificial intelligence supposed to be the next massive technological boom? Isn't this the industry where everyone is currently getting rich? Isn't this the revolution that is going to change everything? Well, maybe, but that does not mean everyone who invested in it is going to make money. And that's where Scott Galloway's argument becomes really interesting. If you don't know who Scott Galloway is, he's an author. He's been basically a business thought leader. He knows a lot about economics and business and he has some, some great information and he talked about this topic. His point in this arena is not that AI is a gimmick. His point is not that AI is some flash in the pan like 3D TVs. You remember those? Everyone was going to have one. Every movie was going to be in 3D and we all wore those dorky ass glasses. And then like five minutes later nobody cared about that anymore. That is not what AI is. AI is real. It's quite powerful and is almost certainly going to transform the global economy over. It's going to change things like education, coding, scientific research, medicine, intelligence work, military planning, logistics, which is my realm, cybersecurity, the media. I mean thousands of things we haven't even thought of are going to be changed by AI. But revolutionary technology and profitable business model are not the same thing. And history has taught us this lesson before. Think Back to the dot com bubble around the year 2000. The Internet was a huge deal. It absolutely changed the world. Clearly here I am. The long term potential wasn't necessarily hype. It actually was understated at the time. The Internet became more important than most people even imagined at that time. But investors still got way ahead of reality. Companies poured billions of dollars into infrastructure. Valuation skyrocketed. Everyone assumed profits would naturally flow forever and they just did not. If you had.com in your name, people threw money at you. And then that bubble burst. Hundreds of companies disappeared. Trillions of dollars in market value just evaporated. But the Internet did not disappear. The technology clearly survived. The bubble did not mean the Internet was fake or a gimmick. It meant that the market had priced the future as if it had already arrived. That may be what we are seeing with artificial intelligence right now. AI may change the world and investors may still lose a fortune. Those two things can both be true. Over the past few years, AI companies and the companies supporting them have spent enormous amounts of money building data centers, buying advanced chips, securing energy, building cloud infrastructure and scaling models that are unbelievably expensive to train and run. The basic assumption was AI is going to replace expensive human labor. Businesses are going to adopt it everywhere, Revenue is going to explode and companies building this infrastructure are going to be sitting at the top of the next great economic empire. Multi trillion dollar companies. In other words, let's build this machine first and as fast as we can and the money will show up later. And I mean even right now, like that.com bubble 30 years ago. Companies just have to put AI in their name or in their business model and money is just thrown at them. But what if the return on investment doesn't show up fast enough? And therein lies the problem. Because in a lot of cases, businesses are discovering that AI is useful, but it's not magical. It can enhance workers, it can speed up workflows, it can help write code, it can summarize documents, it can answer customer questions. It can do a lot. But replacing humans outright, at least in the near term, and I'm talking probably the next 10 years, that is a much harder problem. And investors aren't going to wait 10 years. And in some cases, it is very much still cheaper, easier and more effective to just pay a competent human being than to build an entire AI system that uses expensive computing power every time it has to think. I mean, Microsoft learned that employees were cheaper than paying credits to try and figure out complex problems. And I saw a tiny, very stupid, very American version of this recently at Dairy Queen. I went there on the 4th of July, and I love Dairy Queen. I'm basically a Blizzard a week kind of guy. And I pulled up with my family and if you know me, I have four kids. So naturally we're all sitting in the car doing what every family does in a drive thru, which is debating what everyone wants, while the person ordering is slowly losing the will to live, which usually is me. And then an AI voice comes over the intercom. And at first I was like, okay, this is kind of cool. We're living in the future. Dairy Queen has artificial intelligence now. That's amazing. And then within about 10 seconds, it became clear the system was listening to the conversation inside the car and trying to add random things we were discussing to the order. It was not taking my order. It was eavesdropping on basically family chaos and turning it into the order. So eventually I basically said, this thing sucks at his job. Please get me a human. And immediately a human jumped on, apologized and helped me. And the way he apologized made me think this was not the first time this had happen. Now this is a very small example. Obviously I'm not saying that the future of artificial intelligence rests on whether it can successfully sell me a Blizzard, but it illustrates a larger problem. A lot of companies were sold AI as if it was a human replacement tool. But right now, in many practical real world situations, it is much better as a human enhancement tool. That is a very different business model. If AI helps one worker become more productive, that is valuable, but it may not justify the wildest expectations investors built into these companies. If AI lets companies replace entire labor categories overnight, that is frankly, a money printing Machine, you've figured it out. But if it mostly just helps existing workers move faster, then the returns, sure, they're there, they're real, but they're slower and harder to detect and probably not big enough to justify the insane valuations that we've heard of and every massive infrastructure bet that they've made. And that is where this potential bubble argument comes in. AI companies may have built infrastructure for a future that is likely necessary, but not here fast enough, like the dot com bubble. It's like a power company building plants for power needs 50 years in the future. They may be right about the destination, but they're wrong about the timeline. And if you are wrong about the timeline when you are spending hundreds of billions of dollars, that's a huge problem. Now, normally in a free market, the answer would be simple. Let the market sort it out. The strongest companies survive. The weak companies fail. The useful products remain. The overhyped products disappear. Capital gets reallocated. Investors, sure, they take losses, but that's capitalism. However, artificial intelligence is not just another consumer product. This is not just some shoe company. This is not an app that lets you order dog food on a subscription model. AI is quickly becoming strategic infrastructure. It's becoming part of military planning, intelligence analysis, cyber defense, scientific research, drone warfare, biotechnology, code generation, managing energy systems, efficient logistics, education, surveillance, even things like propaganda. Economic productivity is going to rely on it. So whoever leads in artificial intelligence may hold a major economic and military advantage for decades. And that is where the free market starts running into the problem of national security. Because if the leading American AI firms stumble financially, if they start to fail, does Washington allow that to happen? Meaning we all say, sorry, you overbuilt your infrastructure and misread the market. You got out over your skis and now the consequences are yours. Or does Washington say, frankly, we can't allow this industry to collapse because if we do, China wins? That's where the debate is right now. Not whether AI is important. It certainly is not whether AI is the future. Again, it definitely is. The debate is whether artificial intelligence becomes the next too big to fail sector, or more accurately, the next too strategically important to fail sector. And China is a huge reason for this discussion. China is not treating AI like a normal consumer technology. China sees AI as a national power project. It sees AI as a military tool, an intelligence tool, an economic tool, and a surveillance tool. And also, frankly, a way to compete with the United States. And to be clear, the United States is looking at those things too. That is why Chinese models like Deep Seq are so heavily studied. The point is not that deep seek is clearly better than the best American models. It it's not. The United States still leads at the frontier. The strongest American models are still head. But China does not necessarily need to be better at every level to create a strategic problem. If Chinese AI models can be good enough, but cheaper, they can gain market share. If they can gain market share globally, they gain influence. If businesses, government, schools, foreign ministries around the world start using cheaper Chinese systems, that creates a strategic advantage for Beijing. And on top of that, there are security concerns. Chinese AI systems may be cheaper, but cheaper does not mean neutral. If the Chinese Communist Party has leverage over those companies, then every country or business using those systems has to ask, what kind of data, influence or dependency are they inviting into their systems? Specifically, what is the Chinese Communist Party going to learn? So Washington looks at this and says, we cannot just let American AI companies collapse because their business model is messy right now. Because this is not just about profit at this point. It is about power. This is about who build the infrastructure for the next technological era. That is how a bailout gets rebranded as national security. And to be clear, the argument is not crazy. I understand. I get it. I do not want China dominating artificial intelligence. I do not want the Chinese Communist Party controlling this and setting the global standard for AI systems, AI infrastructure, AI censorship, AI surveillance, or AI military integration, that would be a disaster. So when Washington says this is national security, they're not making that up out of thin air. They have a point. But national security has always been the magic phrase that lets governments do things they otherwise could not do. And that is where the danger also lies. Because once the government becomes the investor, the customer, the regulator, and the strategic beneficiary of an industry, those incentives start to change. If the government is funding AI companies buying from AI companies, protecting AI companies, and maybe even taking ownership stakes in AI companies, then those companies are not just private businesses anymore. They become a part of the State's strategic architecture. That is a corporatocracy at its core. And Mr. Global talked about that on the last episode of our podcast, American Power. And once that happens, the question becomes, who are they really serving? Are they serving their customers, the shareholder taxpayers now, or the government that keeps them alive and running? This is where the cronyism concern comes in. Because imagine the cycle. A company builds a powerful AI model. The government says that model is vital to national security. The company receives public support, federal contracts, favorable treatment, maybe even an equity investment from the government. Then the company becomes dependent on that relationship. And once a company becomes dependent on on the state, it becomes much harder to say no to the state. Maybe at first the company says we will never allow our AI to be used for certain military purposes, as Anthropic has. But then the government says, well, we've invested in you and we're basically keeping you alive. But also this is about protecting American troops. Then it pivots and says this is about reducing civilian casualties. Then it says this is about countering China, this is about homeland security, this is about. And then we go down the line of this is about domestic extremism, this is about misinformation, this is about election integrity. And you see where I'm going. Every step sounds reasonable in isolation, but over time, the tool expands and starts infringing on rights. The mission expands, the justification expands, the customer changes the product. And that brings us to the concept of the Imperial Boomerang, which my co host Nat Townsend has talked about previously on American Power as well. The Imperial Boomerang is the idea that governments often develop tools for use abroad in war zones or against foreign enemies, and eventually those tools come home. And we've done this before, not only in surveillance, but definitely you can see it in surveillance. You can see it with counterterrorism authorities, you can see it with the emergency powers, but also you can see it with something as basic as policing in your own state. MRAPs were built for Iraq and Afghanistan. These are mine resistant ambush protected vehicles, not running into a lot of mines and ambushes here in the United States. They were designed for war zones. They were designed to protect troops from roadside bombs and rockets and things that you're not going to run into on the streets of Boise, Idaho. It made sense. In Iraq. I use them, actually. This painting picture, whichever side it is on, if you're on the YouTube video, that's an MRAP. Because that was my life for a year. It kept us alive. But then eventually those same types of military vehicles started showing up and American police departments. A tool built for foreign war did not stay in foreign wars. It came home. Why? Well, money. These companies needed to stay alive. So with the war over, they had to sell them to someone. So governments bought them, subsidized by the federal government. And that is my concern with AI. Not that AI is an armored vehicle. Obviously it's not. But institutionally, AI could become the next mrap, a technology justified by foreign threats, funded in the name of national security, built to compete with China, integrated into intelligence, military, cyber and surveillance systems, and then eventually used domestically in ways the public never really agreed to. Because once you build a very big hammer, it has a way of finding nails. Maybe those nails stay overseas. Maybe they stay in cyber defense, as Mythos is right now. Maybe they stay in military logistics, which would be hugely helpful, let me tell you. Maybe they stay in private sector productivity, and I hope they do. But history suggests we should all be skeptical because cameras did not just stay in your hand. They ended up on every street corner. Drones did not just stay on battlefields. They moved into policing border security and domestic surveillance. And again, MRAPs did not just stay in Iraq and Afghanistan. They ended up in American towns. And if AI becomes the most powerful tool the government has ever helped build, we should at least be honest about where this could go. Obviously, this is such an important endeavor that for the first time, the US Government is now buying up the means of production, and a Republican government at that. In the AI realm, whether it is intel chips or anthropic models, this is something we need to pay attention to, because we are seeing a world of firsts. And we aren't regulating, not just for the AI side of the House. We're not regulating even for the government's actions when it comes to AI.