Stephen Mayne

Well, you know my view, Alan.

What do you think?

Apparently there's 4,100 data centers in America already, but CNBC had a big interview yesterday in Michigan, the biggest one they're building.

They interviewed Sam Altman there.

And it's going to cost $45 billion U.S.

dollars, $16 billion to build it, and then the cost of all the chips is

And it's like 1.4 gig.

So it's absolutely enormous, the amount of compute.

And they've got their own cement plant, their own steel-making plant as part of this whole sort of campus, they're calling it, with three different towers.

So the scale of the thing is just extraordinary.

And how you can invest US$45 billion on one site in Michigan...

and get a return.

That's where a lot of people are saying, I don't think the economics are going to add up.

But then you've got Anthropic, where people say, well, they're now doing annualized revenues of $47 billion, and they're going to be cash flow positive in the next quarter.

So their business is looking real.

But SpaceX and OpenAI, trillion-dollar valuations,

I struggle to see how it stacks up.

I have the same approach but I'm very nervous about the index ETF side of things because it is a phenomenal change in the way markets work and if you suddenly get a rails run into these indexes and all the index funds are forced to buy you, that can lead to quite irrational share prices because you've got BlackRock and State Street and Vanguard just buying with their ears pinned back because they have to under their passive investment mandate.

So...

If they can establish a valuation of a trillion and get that backed up by index weightings, then that just creates a wall of buyers that potentially perpetuates the distortion of the valuation until ultimately the cash flows, profits and dividends decrease.