The Information is 100% correct: OpenAI and Anthropic are money pits

On Tuesday, The Information reported that individual investors now have a way to invest directly in privately held, pure-play large language model companies such as OpenAI and Anthropic.

Until now, public market exposure to LLMs came only through diversified companies with profitable core businesses — Google through advertising, Apple through hardware and services, Meta through advertising, Microsoft through enterprise software.

Direct access to pure-play LLMs changes that equation.
It gives investors 100% exposure to LLMs — both the upside and the risk.

The Information’s warning is therefore well placed: “AI could be a money pit for individual investors.”

What the article does not explain is why.

History is clear: retail investors tend to rush into capital-intensive businesses precisely when structural risks are becoming visible and institutional investors are already wary.

Here is the missing analysis:

Adoption failure

• Consumer LLM adoption plateaued in 2024.
• Model improvements in 2025 did not accelerate adoption.
• Adoption of Anthropic’s Claude remains flat and far below ChatGPT and Gemini.

If adoption is not growing, scale does not compound — it stagnates.

Monetization failure

• OpenAI has not converted its massive audience into paying users.
• Roughly 95% of ChatGPT users do not pay.
• Advertising is not a viable fallback due to global user mix outside high-value North American markets.

Audience size without monetization is not leverage. It is a cost compounded by scale.

Enterprise reality

• Approximately 95% of enterprise LLM deployments have failed to reach durable production.
• Enterprises remain unwilling to absorb hallucination, security and cost risk at scale.
• Anthropic’s enterprise-first strategy makes deployment failure rates existential, not incidental.

If enterprise adoption was supposed to solve the revenue problem, this data breaks that assumption.

Economic impossibility

• Tokens are not a proxy for compute or operating cost.
• Subscription pricing cannot cover inference, safety layers, agents, and capex.
• As features expand, compute costs rise rather than fall.

Even perfect adoption would not fix broken unit economics.

Valuation disconnect

• Anthropic’s valuation is disconnected from adoption, revenue and margins.
• OpenAI’s valuation assumes monetization that has not materialized.
• Retail access introduces price discovery into companies that have never faced it.

Direct exposure does not increase upside. It removes buffers.

Conclusion

The Information is correct.

Pure-play access to OpenAI and Anthropic does not offer a cleaner way to invest in AI. It offers unfiltered exposure to stalled adoption, failed enterprise deployment and economic models that do not close.

That is not opportunity.
That is a money pit.

– Published Wednesday, January 7, 2026