Analysis•

Oaktree's Howard Marks Tells Investors That Being First in AI Is No Guarantee for OpenAI and Anthropic

In a November 11 memo, Oaktree co-founder Howard Marks argues fast AI progress, big tech rivals and a projected $1 trillion yearly capital bill could unsettle today's frontier labs.

Written By :MINRK
Oaktree's Howard Marks Tells Investors That Being First in AI Is No Guarantee for OpenAI and Anthropic

Photo by Batyrkhan Shalgimbekov on Unsplash

1. Howard Marks AI Memo Puts Durability in Question

Howard Marks, co-founder of Oaktree Capital Management, released a memo dated November 11, 2026, titled “A Layman Thinks About AI,” questioning whether OpenAI and Anthropic can hold on to their leads. He argues that rapid technical progress, well-funded big tech rivals and an AI capital requirement he projects at $1 trillion a year make early leadership far from permanent.

Marks has built a reputation over decades for urging humility at moments when other investors feel most certain. His latest memo carries that habit into artificial intelligence, the theme that has dominated markets for the past several years. Rather than predicting winners, he concentrates on how much is unknowable and how much money is being committed in the meantime.

The core message is simple: being first does not guarantee staying first. In a field that changes as quickly as AI, he suggests, investors should be careful about treating today's frontrunners as tomorrow's inevitable champions.

2. Why Marks Calls Himself a Layman

The modest title is a deliberate choice. Marks presents himself as an outsider to the technology rather than an expert in machine learning. That framing lets him step around engineering debates about model architectures and benchmarks and focus on the questions he knows best: risk, capital and whether investments will ultimately be repaid.

It is a familiar posture for Marks, whose memos to Oaktree clients have long been read well beyond the firm's own investors. He typically approaches new market themes by asking what could go wrong, what is already priced in and how confident anyone can reasonably be about the future. Applying that lens to AI means setting aside the excitement about what the technology can do and asking instead what it will take to earn a return on it.

By declining to claim technical authority, Marks also makes a broader point. Investors do not need to understand every detail of how large language models work to recognize when uncertainty is high and the amount of money at stake is enormous. Those two conditions, in his view, call for caution regardless of how impressive the technology appears.

3. Can OpenAI and Anthropic Keep Their Lead?

Marks does not say they cannot, but he raises the real possibility that they may be challenged or overtaken. His central concern is durability. In a field improving at this speed, any advantage a company holds today can be erased by the next breakthrough.

Crucially, he does not assume that breakthrough must come from a small startup. Large technology companies have the money, talent and computing resources to compete head-on with the current frontier labs. Marks treats that rivalry as a genuine threat to the leaders, not a distant theoretical risk.

OpenAI and Anthropic are the two names most closely associated with the frontier of AI development. Both build large general-purpose models and sell access to them to consumers and businesses. Their positions rest on staying ahead technically while raising the vast sums needed to train and run ever larger systems. Marks's memo asks how secure that combination really is when the biggest companies in the world are pursuing the same goal.

4. A Predictability Score of 2 Out of 10

In an unusual twist, Marks used Anthropic's own AI model, Claude, as an input for the memo. Claude assigned frontier AI labs a predictability score of 2 on a scale of 1 to 10. A score that low signals that the future of these companies is close to unknowable, even to a system built by one of them.

The score is not a forecast of failure. It reflects how many variables remain open: the pace of technical progress, the behavior of competitors, the cost of building and running models, and the willingness of customers to pay. When so many factors are in flux at once, confident predictions about who will lead in five or ten years carry little weight.

For an investor with Marks's outlook, low predictability is itself a key piece of information. It argues for sizing positions with care, demanding a margin of safety and avoiding the assumption that current market leaders are locked in.

5. Lessons From Microsoft, Amazon, Napster and Yahoo

Marks grounds his argument in the history of earlier technology waves. He points to Microsoft and Amazon as companies that turned early positions into lasting success, building businesses that remain dominant decades later.

He sets those examples against Napster and Yahoo, which serve as cautionary tales. Napster pioneered digital music sharing but did not survive to profit from the industry it helped transform. Yahoo was once one of the most recognized names on the internet yet lost its leading position as competitors moved ahead.

The period Marks draws on is the turn of the century, when investors correctly identified the internet as a transformative technology. Being right about the technology did not protect many of them from heavy losses, because picking the right technology and picking the right companies turned out to be very different tasks. Some of the businesses that looked strongest at the peak faded, while others that endured went on to define the following two decades of the digital economy.

The lesson is that a new technology can be both real and transformative while still destroying many of the companies that first championed it. Early leadership is one ingredient of long-term success, but it is not sufficient on its own. The memo implicitly asks investors to consider which of today's AI leaders resemble the first group and which may turn out to resemble the second.

6. How Much Capital Does AI Need?

Marks projects that AI will require an average of about $1 trillion in new investment every year in the period ahead. That figure forms the second pillar of his argument, alongside the question of competitive durability.

Spending on that scale only makes sense if revenue eventually justifies it. Marks is skeptical that the infrastructure outlays now under way will generate adequate returns, and he warns that such heavy capital needs could weigh on profitability for the companies making them. Data centers, specialized chips and the electricity to power them all require money up front, while the revenue that might repay those costs arrives later and with far less certainty.

This is a classic concern in capital-intensive industries. When many companies race to build capacity at the same time, the combined investment can outrun demand, compressing returns for everyone involved. Marks does not claim that outcome is certain for AI, but he treats it as a risk investors should weigh seriously.

7. Debt Is Funding a Growing Share of the Build-Out

Much of the AI build-out is being financed with borrowed money. Goldman Sachs data cited in the memo puts global AI-related debt issuance at nearly $600 billion in 2026, with roughly $400 billion of that coming from the US market.

Debt changes the risk profile of an investment boom. Equity investors can absorb losses when projects disappoint, but lenders expect to be repaid on schedule. If the revenue from AI infrastructure arrives more slowly than expected, companies carrying large debt loads could face pressure that equity-funded rivals would not.

For credit investors, a field Marks knows intimately through Oaktree's history in distressed and high-yield debt, the scale of this borrowing is a central consideration. A rising pile of AI-related bonds and loans ties the fortunes of lenders to the same uncertain revenue streams that equity holders are betting on.

8. Concentration Risk Inside the S&P 500

Marks also highlights how much of the stock market's recent performance depends on a single theme. About 42 AI-related companies accounted for roughly two-thirds of the S&P 500's price gains since January 2024, a figure cited in the memo.

That concentration has consequences for everyday investors. Many people hold broad index funds precisely because they want diversification. Yet if a large share of an index's gains comes from one group of related companies, those investors may carry far more exposure to AI than they realize.

Should sentiment toward AI shift, the effect would not be confined to dedicated technology funds. It would ripple through the broad benchmarks that underpin retirement accounts and passive portfolios, which is why Marks treats concentration as a risk worth naming explicitly.

9. What the Memo Means for Investors and AI Leaders

For OpenAI and Anthropic, the memo reads less as a forecast of failure than as a reminder that their positions can be contested. Marks describes a fast-moving technology, well-resourced rivals and extremely high capital requirements, a combination that leaves room for the landscape to shift sharply if new competitors emerge.

His advice to investors concerns posture rather than timing. He does not call a top or recommend selling. Instead, he urges a humble and measured approach in an environment where uncertainty is high and the cost of staying invested keeps climbing.

Three indicators stand out as signals to monitor. The first is whether AI revenue begins to close the gap with AI spending. The second is how the 2026 debt pipeline holds up as more borrowing comes due and new issuance continues. The third is whether the concentration of S&P 500 gains in roughly 42 AI-related companies deepens further or begins to unwind.

Marks is not dismissing AI as a mirage. His own historical examples show that some companies captured enormous value from earlier technology shifts. The harder task, as his memo frames it, is telling which companies will become the next Amazon and which the next Yahoo while both are still spending heavily and the outcome remains far from clear.

Related Articles

NEWSLETTERS

Don't miss another story.

Subscribe to the MINRK Newsletter today.

Tick the newsletters you want. No selection? You'll get MINRK Headlines.

By signing up, you will receive emails about MINRK products and you agree to our terms of use and privacy policy.

MINRK Daybook

Market analysis for crypto traders and investors.

EVERY WEEKDAY

MINRK for Advisors

Defining crypto, digital assets and the future of finance for financial advisors.

EVERY THURSDAY

MINRK Protocol

Exploring the tech behind crypto one block at a time.

WEEKLY

MINRK Long & Short

A must read for institutions. Insights, news and analysis delivered weekly.

EVERY WEDNESDAY

MINRK Headlines

The biggest crypto news and ideas of the day.

EVERY WEEKDAY

MINRK Policy

Examining the intersection of cryptocurrency and government.

WEEKLY

MINRK Research

Join thousands of readers who rely on MINRK for data-driven insights on the latest digital asset trends.

MONTHLY