# Are we underestimating AI revenue by looking at software spending instead of labor costs? — retained record

> Retained reddit record from reddit:singularity.

Canonical: https://highsignal.app/data/records/dc358ac04c339b3e

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:singularity)  published 2026-11-01 00:49 ingested 2026-11-02 02:39

[original source ↗](https://www.reddit.com/r/singularity/comments/1xkq2cu/are_we_underestimating_ai_revenue_by_looking_at/)

## retained text



```
Lately, there has been a lot of discussion about the AI bubble, particularly whether OpenAI, Anthropic, and other companies can generate enough revenue to justify the enormous investments being made.

But let's look at it this way.

Apparently, there are roughly 1 billion white-collar workers worldwide. If companies eventually spend an average of $200/month per employee on AI, that's $2.4 trillion in potential annual B2B spending. This sounds like a ridiculous amount of revenue, but consider this.

I think an average US white-collar employee costs a company around $10,000/month, including benefits. If AI allows a company to reduce its workforce by just 2%, the savings alone could theoretically fund $200/month in AI spending for everyone else. And that's without considering other productivity gains.

And if you are a company, would you rather have 100% of employees with $0 spent on LLMs or 98% of employees with each employee having a $200/month subscription to Claude/GPT/etc.?

Now, obviously, wages are much lower in many other countries. There are also other factors such as cheaper open-weight models, self-hosting, and competition that could dramatically reduce how much revenue frontier-model companies actually capture. But $200/month is already an established price point for high-end AI subscriptions, and it doesn't seem ridiculous that companies would be willing to pay this per employee for better models and the convenience of sticking with one provider.

So at least from this point of view, the numbers look pretty reasonable to me. I don't think it's that far-fetched that AI could become a trillion-dollar industry within the next few years especially with models keep on improving its performance. 

What am I missing here?
```



document fetched 2026-11-02 00:28 · canonical https://reddit.com/r/singularity/comments/1xkq2cu/are_we_underestimating_ai_revenue_by_looking_at
