Anthropic’s IPO Is Very Silly.

A 100-Year View

Our little family traveled to celebrate my father-in-law’s 99th birthday this past weekend. This is the same man who rolled out his 20-year plan for me just a few years ago, so I suspect we’ll be making this trip for a few years yet to come.

He polished off his meal, had cake and ice cream and even danced with his wife to Sinatra in front of the close gathering of friends and family. The youngest attendee was his three-year-old great-grandson who danced the night away and chased a balloon to the joy of everyone in the room. The 100th is already planned, as he made sure to remind us all.

Today, Anthropic released its bare-bones prospectus detailing its vision for the future. Little has changed since we reported on the early leaked information, and in fact there are more questions than answers after going through it. But reading the company’s aspirations and risk factors, it occurred to me that my father-in-law will either outlive it or it will discover a way to keep him alive for another 99 years. That is to say that Anthropic will either go up in flames having burned the greatest amount of capital in history or it will discover what Ponce de León failed to locate. As with all of these big AI swings there appears to be no in between.

After 99 the producer, my 99-year-old father-in-law and his wife have likely listened to more UNFTR episodes than anyone. (If she’s reading this, I apologize to her for having to endure this much of my voice.) While I don’t hear from him after each episode, I hear from him a lot—particularly when he disagrees with my conclusions. Which is often.

We agree on a sliver of time. The here and now. He believes in my research and thought process and therefore my perspective on the present. He disagrees on everything that came before and even more so on what comes next. He defines his life by optimism. It’s his creed. As a child of the Depression who sold yarmulkes on the street when he was little, he’s familiar with dread. As a Jew who fought in World War II, he’s familiar with the worst of what mankind is capable of. And yet, he chooses to see the good in everything. Tomorrow isn’t yet written, so get busy writing and don’t worry about how we got here.

He’s gotten to me more than he realizes. It’s why I recently wrote about my disdain for white liberals who decry the Trump administration as the worst time in history. It’s why I respond with “for who?” Things have been worse. Like, way worse for most. But now history has come for the rest of us with economic precarity and fascistic surveillance.

Through ups and downs, things have worked out well for my optimistic father-in-law. But he said something in a quiet moment with just a few of us as he held court that I can’t shake.

He was talking as usual about his unwavering faith in the future and what technology will deliver to us as a society. Coming from someone who has given me copies of Ray Kurzweil’s “The Singularity Is Near” and Norman Vincent Peale’s “The Power of Positive Thinking,” it’s very much on brand. The man who still trades crypto every day on his computer. At his office. Where he still goes every day. To trade crypto. I digress.

He was admonishing us for not understanding the moment in which we live. His first great disappointment in my view of the present. When we pressed him to offer a perspective he said, “It’s 1932.”

As a stockbroker once upon a time, he too believes the market is inflated beyond reason as it was in the 1920s. All the same instincts of greed and leverage are at play. The difference now is that we can print money to prop up whole systems. That can only last for so long, on this we’re aligned. But he perceives this moment to be the same as his first real sentient year on this earth. The throes of the Great Depression. On the precipice of world-altering war. Fundamental changes ahead that will throw us into chaos before the dust settles into something new. Something better. And the delivery mechanism will be artificial intelligence and all it stands for. He is the living, breathing epitomization of Anthropic’s more optimistic vision for the future. The only crack in his outlook that I was able to sense is that he’s legitimately upset that he won’t be around to see the grand recovery 40 years from now.

That’s how long he thinks it will take to actualize the promise of this new frontier. Plenty of heartache ahead, but how wonderful things will be if we hang on. It’s also on brand for him to offer the greatest possible hedge. Hard to attack the premise if the payoff is four decades from now.

This is Anthropic in a nutshell. From a pure investment standpoint, OpenAI is fucked. Anthropic is the one to beat and its forthcoming haul will allow it to put even more distance between the two companies. Meta has already pivoted to align with Apple on a different AI system play. SpaceX is a joke. It will need to buy intelligence, compute and market share because it’s not a real company. It’s a giant special purpose acquisition company, pure and simple.

Anthropic knows what it is and where it’s headed, and that’s why it needs all of the money right away. The battle now isn’t about competing interests; it’s about which one of these systems will be integrated into the other tech giants and platforms. Anthropic isn’t a product on its own. Claude on its own dies out in the cold and the elements. In order to live and to thrive it requires host bodies. You know, kind of like a virus.

What the prospectus revealed is a company that needs cash like we need oxygen. The cost of compute continues to skyrocket and its revenue growth, while impressive year-over-year, has already begun to decelerate. For anyone keeping score, that’s the opposite of what needs to happen. It also has enormous concentration risk with a handful of enterprise customers, some of which are also heavily invested in Anthropic and themselves experiencing negative cash flow because of their own need for compute. It’s all such a clusterfuck. The company also said that it needs to achieve investment-grade debt status quickly because of its capital needs.

So now we have two companies (SpaceX and Anthropic) claiming TAMs the size of the entire U.S. economy and the highest valuations of any company ever. They can’t both be right. Hell, neither can be right on its own.

Famed investor George Noble’s assertion that AI writ large will be “the greatest misallocation of capital in history” can be true. In two ways.

Anthropic might someday be a good company. It might find certain economies of scale and build the best enterprise models. That’s where the value ultimately lies. It might have hundreds of billions in revenue and a few billion in net operating income. It might. Or it might not. Either way, the return on the multitrillion-dollar investment into the infrastructure designed to support it will likely never come to fruition. The total addressable market for AI as a standalone function remains a fantasy. That’s why a respected investor like Noble can make this claim.

AI is a powerful supercharging tool for the big boys. But here’s what it’s not.

It’s not the railroad. Or the interstate highway system. Or the internet. There is nothing so universally beneficial about AI to make a claim of democratization. Especially since its stated purpose is to displace workers. Conversely, it’s a powerful tool that will enable other market leading systems and companies to widen their respective leads. The only way this all pans out is if the biggest companies in the world can pay enough to incorporate Anthropic models into their platforms in order to widen their lead and pad their stats.

Alphabet and Microsoft will get bigger and more powerful. Amazon and Walmart will as well. Massive platforms and systems providers will get bigger and more efficient and the rest of the service providers will be subsumed by them. Or gone. Anthropic is greasing the wheels of monopolization. It’s not democratizing anything because it’s too expensive to implement. It’s pressing the advantage of those with the advantage.

There will be more losers than winners in the pure AI game and even the winners will likely never produce a respectable ROI that deserves this level of investment. Our leaders and the investor class have made such an uncompromising bet on the future of AI to reveal the other aspect of the great misallocation of capital. The starvation of other industries.

That’s where my father-in-law’s math might be spot on. It’s 1932.

Not every company failed during the Great Depression. The great technology company of the day, the Bell System (now AT&T), thrived during this period. According to its own company history, “In fact, during five of the 10 years, 1930-1939, the dividend was paid out of surplus, thus adding emphasis to the stock’s reputation of being one best suited for ‘widows and orphans.’” As bleak as everything was, Americans still craved entertainment as an escape so this period also proved fruitful for a handful of movie industry giants such as Loew’s, which earned a profit every year of the Depression. At the beginning of the automobile revolution in the United States, there were hundreds of car companies, but only a few remained after the Depression, most notably the “Big Three”—GM, Ford and Chrysler.

Anthropic is an accelerant to monopolization. We cannot eat AI. We cannot drive on it. We can entertain ourselves endlessly to AI slop, I suppose, but that’s hardly a substitute for food, energy, healthcare, education and all the other areas of the economy we’re starving in pursuit of a vision that belongs only to a handful of tech entrepreneurs. Perhaps the Bell System is Alphabet, Loew’s is Netflix and Tesla is the “Big Three” in this scenario. Rewarded by the investor class for their ability to soak up cash during this fiat frenzy we’re all financing without our permission.

The circumstances today are similar in that there are giant corporations that enjoy market and mindshare. And if the economy truly goes upside down, the ones with the greatest share of both and the highest capitalization will remain and potentially thrive when competitors are scorched. They won’t be better. They’ll just be bigger.

The Giant Cash Sucking Machine That Is AI

Here’s a great visualization for what we've been talking about. Only hyperscalers in Big Tech could afford to go cash flow negative to this degree in order to build a moat. The most profitable companies on earth are giving it all over to “compute.”

Sources: Bloomberg