Is This Time Different?
Mapping Capital Returns to Artificial Intelligence
Two weeks ago, my wife Leah and I took a quick trip to Newport, Rhode Island. It was my first real visit and I had a terrific time.
I had heard about how much of a sailing community it was, but I didn't realize how obsessed the town is until I visited. You can take sunset cruises and see the actual ships that won the America's Cup in the 1960s and 1970s.
I'm far from a sailing fan. The only real thing I know is that Larry Ellison, the founder of Oracle, has spent ungodly amounts of cash on America's Cup campaigns ($750 million, according to one estimate).
He can certainly afford it. Last week, Oracle's stock popped more than 30% after it reported an eye-popping revenue performance obligation of $455 billion (up over 350%). The skyrocketing RPO was from massive upcoming cloud infrastructure spend from one specific client (OpenAI).
Oracle's stock move was astounding. Simply put, companies worth hundreds of billions of dollars don't have such intraday moves like this.
For a moment, Ellison became the world's richest person. Not only does it show the power of sustained ownership (Ellison still owns more than 40% of Oracle's shares), but it shows the modern-day race to gain as much market share as possible.
This current environment inspired me to revisit Capital Returns, one of my favorite finance books. The book contains investment letters from Marathon Asset Management, but the most well-known part of the book is the introduction by Edward Chancellor (one of my favorite financial writers and historians).
The introduction summarizes Marathon's capital cycle theory. The crux of the theory is that attractive capital returns in a particular sector inspire more aggressive capital spending by firms, which eventually leads to excess capacity, collapsing profits, and slashed capex. That reduction in profits and investment eventually seeds the ingredients for a new capital cycle, and the cycle continues.
It's a compelling theory and one that we saw during the 90s tech bubble, the 2000s global shipping industry, the 2000s housing market, and more. There's a good argument that we're in the middle of one of the largest (if not the largest in terms of nominal dollars) capital cycles in our lifetimes.
The ingredients are certainly there. Capex among the major AI labs, Magnificent Seven companies, and AI-adjacent companies continues to accelerate. The music is still playing and everyone is continuing to dance.
Eventually, this hyper-intense spending and competition will hit an apex. Economic returns will fall below the cost of capital and lead to lower share prices. It seems inevitable, although the exact timing is basically impossible to forecast.
But is this time different? While those are the most dangerous words in finance, we're dealing with the pursuit of the ultimate prize: artificial general intelligence.
Arguably, the first firm (or even nation state) that achieves AGI will win perhaps the final economic, political, and social race. So perhaps it's hyper-rational to go all-in on getting as close as possible to AGI.
It's fascinating to watch. Like the dot-com days, there will probably be some type of severe pullback. But also like in the dot-com era, the underlying technology isn't going away. Whether we like it or not, it's here to stay. And that can lead to immense riches and power.
I'm focusing on one tiny part of this universe. At the same time, it'll be interesting to see who the winners and losers are. In all likelihood, the winners will be companies that we least expect.
Prompt of the Week
Another introspective prompt from me this week. I don't use LLMs as a therapist, but I do find it interesting to see what a somewhat "objective" machine thinks about me. Try this prompt and let me know what you think:
"What do you know about me that I might not know about myself?"