Meta becoming a cloud - Excess capacity coming ?

Meta announced it is looking to get into the cloud business and rent out its computing power.

The semi stocks (MU, SNDK, AMAT) promptly dropped a fair bit (say 10% or so)... The AI/HPC plays like Nebius, CoreWeave (dropped say around the teens). These moves are easily explainable i.e. many people already have exposure in semi. The Neocloud companies now have more competition.

Or another take is excess capacity i.e. Meta can't use up all the computing power it has built up.

My take is it is a bit of everything. Meta's valuation is lowish vs. others (19X PE 26). Now contrast the valuation of Nebius or CoreWeave.... So if I were Meta, if I could rent out a bit of my capacity, build a new business, grow a bit faster (supposedly we have a computing shortage so rental for GPUs should be decent), and attract a higher valuation, it makes sense.

So I am leaning to the optimistic side.... It is not quite a ring the bell moment that AI trade is "done". After 2027, I don't know... BUT my base case is capX in 2027 should remain OK.

Nic Baratte of Techstock01 made a most interesting observation i.e. AI capX yield is improving sharply.... Remaining Performance Obligations of hyperscalers are accelerating.

His post is worth a read.

https://techstock01.substack.com/p/ai-is-a-fake-ai-is-a-bubble-again

Quoting from his article:

>>>>>>>>>>>

This means that:

• $1 in Capex (1 million, 1 billion, as you wish) used to trigger $0.5 in committed future revenue (or ½ million, ½ billion, as you wish)

• Now, $1 in capex triggers $3 in committed future revenues.

Maybe that’s a naive illustration. Maybe there are “long and variable” lags between Capex and future revenues. Maybe the increase in RPO is linked to new product launches, not directly Capex. But still, committed future revenues are increasing a lot faster than Capex.

>>>>>>>>>>

In short, if I were Meta, if I can extract extra yield from my GPUs, it makes sense to do so.

In addition, I have not seen anything that happened in the past month to suggest that ALL the semi channel checks in the world are going to be wrong....

Thus, my base case is AI capX should hold up in 2H 26 and 2027. Beyond that, I don't have a strong view.