We're Seemingly Still in the "Throw Money At It" AI Era...
On the surface, two of the biggest recent stories in AI seem unrelated. The first is Meta’s massive hiring spree to put together their new “Superintelligence” team. The second is xAI releasing Grok 4. But actually, these two may be very related at the highest level. That is, they showcase that we’re still very much in the “throw money at the problem” part of the AI cycle.
This is important because it means that any company with the will and resources can seemingly still get back into the race. Or at least, that’s the hope. A week ago, I would have said we might be beyond that point, which is why I was skeptical that Meta’s mega deals for talent will work. (To be clear, I’m still fairly skeptical there, but not because they’re too late, but rather because of other intangible elements of building in this space right now – team cohesion, Meta’s own focus, etc). But I’m getting less skeptical on the news about Grok 4. And specifically, the fact that it seems to perform — and specifically outperform – the other cutting edge models on the market right now. If nothing else, this shows that Elon Musk’s bet on his “Colossus” data center cluster may be paying off. Perhaps not literally yet, but certainly in these performance metrics.
That’s remarkable because xAI more or less started from zero just a couple years ago. While OpenAI and Google have been in the model-building business for years and years, Elon just… walked in and said, I can do this too. And then, remarkably, did it.
This also seemed to be the case with Grok 3 released earlier this year. But that model, while seemingly up to par with the others at the time, clearly has had other, tangential um, issues. But Grok 4 seems to resolve this debate. Elon was able to get xAI into the race simply by throwing money and talent at the problem.
Of course, I say “simply” and presumably there was nothing “simple” about it, including having to tell Oracle to pound sand at one point to go it alone in building their own supercluster data center — which, incidentally, seemed to open the door for OpenAI to launch the “Stargate Project” (which, to be clear, is not yet operational). But at the highest level, his bet seemed to have worked.
And it may pay off in the form of xAI’s valuation.
Meta is a different story and strategy. But those are also seemingly converging. Mark Zuckerberg is betting at least as much money – if not more – than Elon with compute, that talent can get Meta back into the AI game. This kicked off with the massive $15B Scale AI deal, and it has continued with talent pay packages that may or may not (read: may) be worth hundreds of millions of dollars. Those deals will clearly continue until morale improves — well, performance, at least.
Again, it’s the same basic bet: that it’s both not too late in the AI race and that money can get you back into it, fast. Of course, while Elon’s bet now seems to have been proven out, it’s way to early for Zuck’s. But Zuck also isn’t sitting around to let the talent do their thing and prove him right or wrong, he’s also making the data center bet too.
News today has Meta announcing their “Prometheus” — sound familiar? — data center cluster. In fact, by some estimates – gigawatts! – it will be the largest yet, putting Colossus to shame.
Not to be put to shame, Elon says Colossus will grow! Not to be put to shame not to be put to shame, Meta says Prometheus is merely the first such "titan" data center cluster! And again, Stargate! Of which there will apparently be many!
It’s a good time to be in the data center business. It’s a better time to be NVIDIA, as their $4T market cap makes very clear.
I would just note that Meta has been making such general promises with Llama for a few years now and… well, they’re having to now pay billions to restart their AI work. So again, nothing is certain here. But it does feel like we’re still in the place where… money wins.
And that’s potentially great news for… Apple. While Alex Kantrowitz and I started talking about this general topic on his Big Technology Podcast last week, his discussion with Box CEO Aaron Levie expands on this notion of throwing cash at the problem and how that could be a very good thing for Apple.
It’s obviously simplifying it, but if all you basically need to do in order to catch up in AI is to throw money at the problem — in the form of compute and talent and data — well, Apple has just as much money as anyone. They can do this too.
They also have insane advantages when it comes to the scale at which their devices are distributed into the literal hands of consumers. That feels like something they could leverage! The problem, to date, has been their internal mentality about all this. Which is to say, they were skeptical of “silly” chatbots, didn’t want to pay for NVIDIA chips, and thought Siri would be good enough to answer the AI call.
Wrong, wrong, and wrong.
In order for Apple to actually catch up here, they need a major mentality shift internally, clearly. The shake up of the AI org is a good first step, but there are many more needed. As I’ve long said, it’s one of the reasons why I think they need to do a fairly major M&A deal here. Not necessarily for the tech, but for the talent. And not just for the talent to do their thing with AI — though that’s obviously a huge part of it — but also to shake up the internal mentality around all of this. Apple is clearly, currently, being run by a group of people that just don’t get it.
And they’re only now starting to get it because Wall Street has kicked them in the ass and embarrassed them with their old rivals NVIDIA and Microsoft far surpassing their stock performance.
That’s a good lesson, but I still suspect it’s not enough. Because it doesn't change any of the facts about how they got into the situation in which they find themselves. Elon gets it. Zuck gets it. Tim Cook? I mean, I wouldn’t bet on it, would you?
I’m not saying he has to go. But I’m saying he’s close enough to him going on his own — see also: his #2 retiring last week — that this is a major issue. Who steps up for Apple here? They have what they need because what you need is seemingly money. But they need to know how to use it!
Step one might be building their own actual massive data center cluster. Ideally with a name that sets up a literal clash of the titan-named data centers...
M.G. Siegler
M.G. Siegler
M.G. Siegler