00:00These are a couple of high-profile departures after, you know, recently, I believe four weeks
00:08back, there were a couple of other names that had departed. One went to OpenAI. So the fact that
00:15Google is losing talent, I mean, these are people who have been there for very long.
00:21It's definitely not a good sign, if you ask me. And plus, their model release has been delayed.
00:27So the Gemini 3.5 version model, Pro 1, has been delayed. And from what we have gathered,
00:35the Gemini 4 is most likely going to get delayed. So this is a race about the models. And if
00:42you're
00:42not releasing a frontier model, and Anthropic and OpenAI keep creating all the news about how their
00:50models can hack, you know, any system out there, that's definitely feels like, you know, Google
00:58is missing at least the model lead right now. And it could change. But to my mind, this is not
01:05a positive. And I think the stock reflected that sentiment.
01:10So we hear that they're going to really centralize their leadership to California. That kind of is
01:18the seat of everything, even though we know that, is it Larry Page or Sergey Brin, who's been moving
01:23to outside of California? Florida? I don't know. Nevada. Nevada. Oh, that's right. Incline Village.
01:29Yeah, yeah, exactly. Incline Village, which is basically California, but just across the border.
01:33What does that say about California as the center of tech talent and how whatever efforts you want
01:42to make to try to decentralize that to spread it out to other places, maybe lower cost places? It's
01:47kind of a fool's errand, isn't it? Yeah. And look, right now, it feels like Anthropic and OpenAI are
01:54getting the big names when it comes to, you know, the folks who are actually training these models and,
02:01you know, the scaling laws that are driving all the development. So from that perspective,
02:08I mean, Google is still a pretty large firm and they've got, you know, folks all over.
02:13But there was a time when Google and Meta used to be the magnet for the top talent. It feels
02:19like
02:20right now it's Anthropic and OpenAI. And the departures kind of suggest that, I mean, even though Google
02:28has a lot going for them in terms of infrastructure, the TPUs, and all those are positives. But when it
02:35comes to the model race, I mean, that's why Mark Zuckerberg spent millions to form that super
02:42intelligence unit with 50 people, but 50 of the smartest people. And still, Meta doesn't have a
02:49model that is comparable to the Frontier Labs. So it is a very hard problem to solve in terms of
02:55catching up to the Frontier. And I feel once you slip, it will get hard.
03:00So just a follow up on that, if you don't come out with the Frontier model and you have this
03:04talent,
03:05is it just a matter of time before that talent then goes to OpenAI Anthropic out of frustration
03:09that they're not making more concrete progress?
03:11I mean, it sounds like, especially with Jeff Dean and a couple of guys that left, you know,
03:17this week, there was an element of frustration. And that's been the case with Google since the
03:24ChatGPT days. They invented the transformer. Why weren't they the first ones to release a product? So
03:30it keeps coming back with this management team. It's like, you're not releasing new products fast
03:36enough. And yes, they have to defend that search business, which is the cash cow, which is what feeds
03:42everything. So I can see both sides. But when it comes to talent, they'll go to the place where
03:49they see, you know, a lot happening in terms of new products, model development. And so maybe Google
03:54is being conservative on the model side for a good reason. They have to defend search.
04:00I think what we're all coming to appreciate more is this stuff's expensive to do this stuff. And now
04:04and Google's out today with a $25 billion bond offer. They can raise capital at the snap of the
04:10fingers. Yeah. How about like the Anthropics and the open edge? I mean, at some point,
04:15they got to start hitting the capital markets, I would think. I mean, the interesting anecdote here
04:21is a lot of Google's cloud revenue comes from Anthropic running their workloads on Google. In fact,
04:29Anthropics models are even trained on Google TPU. So Google is the infrastructure provider for
04:35Anthropic. But at the same time, they are competing at the model there. So
Comments