And they can look forward too, their order book has tens of billions in there for the coming years. And that's orders, on top of that comes the maintenance and support for all the machines in operation - in 2023 they delivered 449 machines (not just their top of the line stuff), which means that there's thousands of machines in operation requiring regular maintenance etc.
ASML's bet paid off and for now at least their business is very sustainable.
Concretely, when you use Django, underneath you have CPython, then C, then assembly, and finally machine code. I believe LLMs have been much better trained on each layer than going end-to-end.
This looks more like an attempt of gaining scarce electricity.
If a country/state has to choice of giving power to data center A or B, it makes sense for Satya to make statements about how only Microsoft provides the most AI value
Well, even though electricity is a commodity it still needs to be bought. My point is that people funding this will run out of patience paying for the electricity long before the public/regulators will need to step in a decided how much of it you can buy.
I guess you could always just use a fraction of the billions in investments and whip up a few new power plants. [1]
reply