Oaktree Capital Management · Q4 2026 · Published 11 November 2026 · Memo
Howard Marks puts eleven questions to investors in AI, from who will keep the profits to the nearly $600 billion of AI-related debt issued in 2026, and argues that so little about AI is predictable that position size and price must carry the decision.
Howard Marks, who says he is not a technology investor, sets out eleven questions an investor in artificial intelligence has to answer: how the models work, what they will be able to do, who will profit, how lasting the leaders' positions are, what the companies will earn, how data centers will fare, how much the economy depends on the sector, what its borrowing means for credit, how predictable any of it is, whether it is a bubble, and how to invest. He asked Anthropic's Claude to score parts of the industry for predictability on a scale of 1 to 10; it gave regulated power utilities serving AI load a 6 and frontier labs such as OpenAI and Anthropic a 2. Marks cites Goldman Sachs's count of nearly $600 billion of AI-related debt issued so far in 2026 and J.P. Morgan's finding that 42 AI-related companies produced about two-thirds of the S&P 500's price gains since January 2024. He concludes that whether AI will earn adequate returns on the trillions invested in it is largely a matter of faith, and closes with Warren Buffett's line from Forbes in 1979 that uncertainty is the friend of the buyer of long-term values.
In that connection, I’m struck by the fact that today’s leading model labs, OpenAI and Anthropic, were unknown to most of us just a few years ago.Show in the letter ›
(The big chip producers – Nvidia and TSMC – are racking up the big profits these days, demonstrating the value of a strong competitive position.)Show in the letter ›
FactSet reported that in the second quarter, Alphabet and Amazon notched a total of $151 billion of gains on equity investments, mostly on Anthropic and SpaceX (which now includes xAI).Show in the letter ›
AI-related debt issuance has totaled nearly $600 billion globally so far this year, according to Goldman Sachs, comprising debt issued by the hyperscalers, in data-center/project financings, and by other AI-linked borrowers.Show in the letter ›
Note, however, that it came from one of my oldest files: timeless wisdom from Warren Buffett that appeared in Forbes magazine on August 6, 1979.Show in the letter ›