The market seems to think everyone in AI will be a long-term winner. I don’t see how that’s possible

I see the AI companies as being in three distinct groups.

Group A: AI infrastructure – Nvidia, AMD, Micron, Intel etc.

Group B: Hyperscalers – Microsoft, Amazon, Meta, Google (although Google kind of fits into all three groups).

Group C: Pure-play application layer – OpenAI, Anthropic, and others.

People have discussed Group C to death. The entire debate so far is whether these companies can eventually generate enough profits to justify the insane spending happening across the ecosystem. I’ve written before about how a potential OpenAI IPO could actually mark the end of the euphoria. But for the sake of argument, let’s assume they figure it out. Let’s assume OpenAI, Anthropic, and the rest of Group C eventually build massively profitable businesses that justify all of this spending.

What I want to focus on is Group A and Group B, because that’s where the stock gains are today and it still doesn't make sense

Right now, everyone is winning or at least most companies in both groups A & B are winning while the rest are at worst flat. And mathematically, that makes no sense.

At the end of the day, a company’s equity value is based on its expected future free cash flows. Because of that, I don’t see how it’s possible for both Group B and Group C to be long term winners.

I’ll listen to analysts talk about how Amazon is a phenomenal investment because they’re pouring essentially all of their operating cash flow into AI infrastructure that will supposedly generate enormous returns in the future. This means that spending must normalize, capex comes down, and Amazon begins harvesting massive free cash flow from the infrastructure they built.

Then, literally seconds later, the same analyst will say AMD or Micron looks incredibly cheap on a 2027 forward P/E basis because AI demand is exploding.

That makes no sense.

You’re advising me to buy Micron because their free cash flow from the AI infrastructure buildout will look incredible from 2026–2029. Then you’re telling me Amazon’s free cash flow is going to explode starting in 2030 once all this infrastructure is finally built out and capex declines. Those two things directly conflict with each other.

Either:

1. The hyperscalers eventually slow spending, finish the buildout, and reap the rewards, which means the infrastructure party ends, hurting Group A.

Or:

1. These insane levels of spending have to continue indefinitely to stay competitive which means the hyperscalers are being fleeced and their return on investment are nowhere near as attractive as bulls claim.

Both sides cannot win long term. Yet the market is pricing many of them like they will.