The scale of money riding on AI

From Financial Times:

Instead, Nvidia’s earnings show the strength of spending on infrastructure by companies and countries determined not to be left behind while AI is still in development. If you listened to other tech earnings you would already know this. Amazon, Alphabet and Microsoft, all Nvidia customers, said capex would rise. Meta’s Mark Zuckerberg went further. In January, he declared that his company would have 350,000 of Nvidia’s H100 AI chips by the end of the year. Nvidia does not specify their price but it is estimated that they are up to $40,000 apiece. Even if Meta paid half that, it would mean $7bn. This sort of demand makes up for the fact that US government intervention has curbed Nvidia’s sales to China.

Because Nvidia designs chips instead of making them, gross margins are high at 76 per cent. While infrastructure spending at other companies is booming, Nvidia’s own spending increase is modest next to revenue growth. Hence cash flow from operating activities for the fourth quarter of $11.5bn, up from just over $2bn a year ago. Nvidia may be thinking about ways to put that to work via mergers and acquisitions.

Source: Good luck catching up to Nvidia  by @FinancialTimes

That’s the amount! It is hard to figure out the financials (because the individual break up is not available) but that explains why Facebook (aka Meta) is shifting towards “infrastructure as a service model”. “Nation-states” would replicate this to provide the digital infrastructure for the public.

Leave a comment

This site uses Akismet to reduce spam. Learn how your comment data is processed.