Hacker News Reader

Comment by summerlight | Hacker News Reader

The only major investment left would be computing infrastructures, but it's severely limited by supply. Even if Google wants to spend more money, simply there's no chips to buy. I don't see any significant future investment opportunities other than Waymo in the foreseeable future, but it seems still far from scaling out.

View on HN

Google is in the fortunate position of having their own in house AI compute modules (TPU's) that are more or less on par with Nvidia chips.

They might have to compete for foundry time, but it's better than having to compete for AI chips.

> Even if Google wants to spend more money, simply there's no chips to buy.

This strongly suggests that there's an opportunity to spend money relaxing this shortage. If your business growth is being throttled by suppliers, then investment either allows you to grow those suppliers, or vertically integrate them out of the picture.

You can't just create new supply. Chip fabs take multiple years to spool up. You also have to get the machines to make the chips and guess what, those are supply constrained too.

It's not a problem you can simply spend money at, it's a global supply chain problem with many disparate companies each with their own constrained supply chains.

Yeah, I think Google should be actively working on addressing shortage. But I don't think this can be solved in the short term even with infinite capitals since ASML is the fundamental bottleneck and everyone is competing for the same limited capacity.

There is a $6B increase in purchase of properties and equipment vs. 23Q1.

Reply on news.ycombinator.com