> ## Content Index
> Fetch the complete content index at: https://wire.fourthweb.ai/llms.txt
> Use this file to discover other available public pages before exploring further.

# Howard Marks Says He Can't Value AI Companies
- URL: https://wire.fourthweb.ai/howard-marks-says-he-cant-value-ai-companies/
- Published: 2026-09-20T16:00:53.000Z
- Updated: 2026-09-20T16:00:55.000Z
- Description: The smartest distressed debt investor alive just said he can't value AI — and neither can you. Howard Marks, Oaktree Capital co-founder, told Bloomberg that AI enthusiasm is real but three variables make it unvaluable: profitability timelines, defensible valuations, and job displacement scale
- Author: Travis Wright
- Tags: AI Agent Economy, AI Agents, Funding Rounds

**The smartest distressed debt investor alive just said he can't value AI — and neither can you.**

### The Summary

- [Howard Marks, Oaktree Capital co-founder, told Bloomberg](https://www.bloomberg.com/news/videos/2026-09-20/howard-marks-flags-uncertainty-in-ai-investing-video?ref=wire.fourthweb.ai) that AI enthusiasm is real but three variables make it unvaluable: profitability timelines, defensible valuations, and job displacement scale
- Markets have been optimism-drunk since late 2022, pushing caution to the sidelines — exactly when caution matters most
- When the guy who built a $200B fortune finding value in chaos says he can't price something, that's not bearishness, that's honesty

### The Signal

Howard Marks doesn't do hot takes. He does cold math on assets everyone else has given up on. When he says AI is "impossible to determine whether that exuberance is irrational," he's not hedging — he's naming the actual problem: nobody knows how to value what hasn't been built yet.

[The three uncertainty vectors he flags](https://www.bloomberg.com/news/videos/2026-09-20/howard-marks-flags-uncertainty-in-ai-investing-video?ref=wire.fourthweb.ai) cut to the core of the AI investment thesis. Profitability: when do model costs drop below revenue per user? Valuations: what multiple do you pay for a company burning cash to train models that might be obsolete in 18 months? Job impact: does mass automation create deflationary pressure that kills the consumer spending AI was supposed to unlock?

> "Enthusiasm around AI is undeniable but uncertainty over its profitability, valuations and potential impact on jobs makes it impossible to determine whether that exuberance is irrational."

These aren't edge cases. They're the base case assumptions every AI pitch deck glosses over. Marks has watched optimism cycles run hot before — dot-com, housing, crypto 2021\. His pattern recognition isn't about timing tops. It's about recognizing when the crowd stops pricing risk because the story feels inevitable.

The job displacement question is the one nobody wants to model. If [AI agents](https://wire.fourthweb.ai/tag/ai-agents/) automate 30% of knowledge work over five years, what happens to:

- Consumer spending (the laid-off don't buy SaaS)
- Corporate margins (who pays for agents if customers are broke?)
- Tax revenue (unemployment insurance isn't funded for this scale)

Marks isn't predicting collapse. He's saying the range of outcomes is so wide that traditional valuation tools break. You can't DCF your way through "maybe this changes everything, maybe it changes nothing." That's not a market. That's a casino with really good PowerPoints.

### The Implication

If Marks can't value AI, your job isn't to pretend you can. Your job is to position for the range of outcomes. That means holding companies with revenue today, not revenue "when agents scale." It means watching for deflationary signals in labor markets — wage stagnation in AI-adjacent roles, hiring freezes in back-office functions. And it means keeping dry powder for the repricing that happens when optimism meets margin calls.

The agents are coming. The business models are not.

### Sources

[Bloomberg Tech](https://www.bloomberg.com/news/videos/2026-09-20/howard-marks-flags-uncertainty-in-ai-investing-video?ref=wire.fourthweb.ai)