SIAgentNews.com
← All news

Howard Marks warns AI leaders face risks from competition and costs

The Oaktree co-founder argues that rapid technological progress and massive capital requirements could threaten the current dominance of OpenAI and Anthropic.

· First reported by Crypto Briefing

$1 trillionprojected annual AI investment requirement

The short version

  • Marks argues that rapid progress means today's advantages can be erased by the next major breakthrough.
  • Major technology companies have the resources to compete directly with current frontier labs.
  • AI investment requirements are projected to reach $1 trillion annually.
  • About 42 AI-related companies have accounted for roughly two-thirds of the S&P 500’s price gains since January 2024.
  • Goldman Sachs expects nearly $600 billion in global AI-related debt issuance in 2026.

Why it matters

For investors, this highlights the risk of over-concentration in a few AI winners. It suggests that being an early mover in a high-cost industry does not guarantee long-term profitability or market dominance.

The risk of rapid progress

Howard Marks, the co-founder of Oaktree Capital Management, recently issued a memo titled “A Layman Thinks About AI.” In it, he explores the durability of the current leadership positions held by companies like OpenAI and Anthropic. Marks argues that in a field moving this quickly, any competitive advantage built today can be wiped out by the next major breakthrough.

He emphasizes that the winners of today do not have a guaranteed seat at the table tomorrow. Because the technology is evolving so fast, the barrier to entry for new innovations remains low. This creates a volatile environment where the current pecking order is constantly under threat from the next wave of development.

Marks uses his perspective as a self-described layman to strip away complex engineering debates. By focusing on risk and capital rather than technical specifics, he highlights a fundamental truth about tech cycles. Getting to the frontier first is a significant achievement, but it does not inherently ensure a company will stay there as the field matures.

Big Tech and capital hurdles

A major concern for the current leaders is the sheer scale of resources available to their competitors. Marks points out that major technology companies possess the capital and infrastructure to challenge the existing frontier labs directly. This rivalry suggests that the current dominance of smaller, specialized labs could be short-lived as larger players enter the fray.

The cost of maintaining this lead is becoming a central pillar of the investment conversation. Marks notes that the yearly amount needed to fund AI progress is expected to hit $1 trillion. When spending reaches such massive levels, the primary concern becomes whether the resulting revenue will eventually justify the enormous infrastructure outlays.

This capital intensity puts a heavy burden on profitability and creates a high-stakes environment for the companies involved. Marks notes that debt markets are currently carrying a significant portion of this load. He suggests that the high cost of staying in the game could weigh on the financial health of even the most successful AI pioneers.

Market concentration and debt

The financial landscape of AI is currently characterized by extreme concentration. According to the memo, roughly two-thirds of the price gains for the S&P 500 since January 2024 can be attributed to just 42 AI-linked firms. This means a large portion of the broad index's recent performance is tied to a single technological theme.

Investors who believe they are holding a diversified portfolio may actually have significant exposure to this narrow group. This concentration creates a situation where the success of the entire market is tethered to the performance of a small number of companies. Marks warns that this creates a high level of uncertainty for those looking at the broader economy.

The debt pipeline further illustrates the scale of the ambition behind these projects. Data from Goldman Sachs shows that global debt for AI-related projects is forecasted to reach nearly $600 billion in 2026. Of that total, around $400 billion is expected to come from the US market alone, highlighting the reliance on credit to fuel growth.

Lessons from the past

To frame his warnings, Marks looks back at historical technology shifts to provide context for current trends. He cites Microsoft and Amazon as examples of companies that successfully turned early positions into lasting dominance. These companies were able to navigate the transition from early adoption to sustained market leadership.

Conversely, he points to Napster and Yahoo as cautionary tales of companies that failed to maintain their positions. These examples serve as a reminder that spotting a trend is the easy part of investing. The much harder task is determining which companies will survive the transition and which will be left behind as the technology matures.

For the leaders of OpenAI and Anthropic, Marks' memo serves as a call for a humble and measured posture. He suggests that while the current trajectory is impressive, the landscape could change sharply if new competitors arrive with better models or lower costs. The goal for investors is to remain aware of these risks rather than assuming the current winners are permanent.

What to watch

  • Whether AI revenue growth begins to close the gap with the $1 trillion annual spending requirement.
  • The progression of the $600 billion global AI debt pipeline in 2026.
  • Whether the concentration of the S&P 500 gains in AI stocks begins to unwind.

Try it

Check your investment portfolio to see how much of your exposure to the S&P 500 is tied to the 42 AI-related companies mentioned in the report.

Sources and links

Read the original at Crypto Briefing

First reported by Crypto Briefing. Written by the SI Agent News Desk from the reporting above.Written with AI from the source above. Every number is checked against the source. Spot a mistake? Corrections.

Liked it? Say why

Add a line for other readers.

No account needed. We never show your email. Comments with links, ads or abuse are held and not shown. Your comment shows after a quick check.

More AI news