00:00When you look at that CapEx fan, and I'm talking specifically for Meta, what do you think Zucker...
00:05...is trying to do with that money?
00:07Well, first of all, last year they made all those investments.
00:10And people, and a lot of us out there that were analyzing it were going, are they over...
00:15...spending?
00:15And so what I think happened, and you heard about MetaCloud, is that Mark Zuckerberg...
00:20...has turned the corner, it was all about models.
00:22So people were getting a little impatient, are they going to release the...
00:25...next model?
00:26Is China getting ahead of them with open source?
00:28Are these frontier model labs beating?
00:30Meta on the sidelines, did they waste all that spend?
00:32And I think they've been in the lab doing things.
00:35And I think what Zuck understands is that compute is actually going to be...
00:40...the metric of whether or not these companies are going to be able to stay flexible enough to be successful.
00:45So, Romain, what I really mean by that is some companies right now...
00:50...I think the model themselves are the differentiator.
00:52We're seeing Anthropic and OpenAI and Gem...
00:55...they're leapfrogging each other quarter after quarter.
00:57Mark is saying, if we have enough compute...
01:00...we already have our advertising business, we're making money in AI because of this moat.
01:05If we have the flexibility to continue to build, to continue to innovate, we can stay ahead.
01:10Continue to deliver AI ROI and that the key to doing this is going to be CapEx and...
01:15...having that massive amount of compute.
01:17He's doing the right thing.
01:18It will be the differentiator.
Comments