RE: Pv power is hard 😪 one gone over the 🌈3 May 2026 15:27
‘The four largest technology companies in the world — Amazon, Google, Meta and Microsoft — are collectively expected to spend $715 billion on capex in 2026.
That is nearly three times what they spent in 2024, and five times the 2023 figure.
Amazon alone is approaching $200 billion. Google and Microsoft are each guiding $190 billion. Meta has raised its own guidance by $10 billion to a range of $125–145 billion.
The rationale is AI infrastructure.
Data centres, chips, fibre and power. The build-out is real, and it is enormous, but cracks are appearing in the underlying business case.
OpenAI — the company that arguably started this arms race — is burning $9 billion in cash against $13 billion in revenue. Their own internal projections show $143 billion in cumulative losses before they reach profitability.
They have missed their own user growth targets.
They have missed their own revenue targets.
Their CFO has reportedly told colleagues she is worried the company cannot afford future computing contracts if revenue doesn’t grow fast enough.
This is like an AIM CEO admitting to a placing coming down the line. We all know how that ends.
Meanwhile, Oracle has committed to building roughly $348 billion in data centres and needs OpenAI to pay it $75 billion a year to cover costs. That is an interesting chain of dependency to have built on a company that is currently selling dollars for seventy cents.
Half of planned US data centre builds have now been delayed or cancelled.
Are we in a bubble?
Just think about this.
Claude has overtaken ChatGPT in usage share.
Nobody seems to have realised this yet, but this means these are commoditised products with no switching costs and no real moat. Users move freely between models based on which one happens to be best in a particular week.
If that’s true — and the data suggests it is — then none of the frontier labs will ever turn a profit at scale. The more users they acquire, the more they lose.
The cascade, if it comes, runs like this:
Frontier labs fail to reach profitability, so GPU cloud intermediaries, who borrowed billions at peak valuations, get crushed, so the hyperscalers cut capex, so Nvidia cycles down.
Remember the reverse flywheel from the BTC treasuries last year?
Yeah.“
More here from Charlie boy
https://open.substack.com/pub/thatstocksguy/p/small-caps-may-review?r=4zgkh3&utm_campaign=post&utm_medium=email
Trek