The Captains of CapEx

Six months ago, I laid out the current state of CapEx spend amongst the players in Big Tech. Back then, for their fiscal 2025 years, they looked like this:

• Meta: $65B
• Google: $75B
• Microsoft: $80B
• Amazon: $100B

Amazon was clearly an outlier in that group, but that also makes some sense given that their capital expenditures extend far beyond data center build outs – aside from their actual warehouses, they also remain the largest overall cloud player by a pretty good margin (though CEO Andy Jassy had said that the "vast majority" of that CapEx spend would be going towards AI spend). Two names not on this list? NVIDIA and Apple.

NVIDIA is straightforward. They're the beneficiaries of a huge portion of that CapEx spend by the players above, as everyone buy sup their chips as fast as they can produce them. That, in turn, has made NVIDIA by far the largest company in the world right now in terms of market cap. At the highest level, it makes some sense: four of the other largest companies in the world are spending record amounts of money, and a lot of that is directly tied to, if not directly going to, NVIDIA.

But one company remains an outlier: Apple. Back then, Apple was still the most valuable company in the world at $3.5T. Now they've been lapped by not just NVIDIA, but also Microsoft. Why? AI, of course. In fact, NVIDIA has been well over $1T ahead of Apple for quite some time now.

Still, Apple remains the third most valuable company in the world. And thanks to some recent kissing of Donald Trump's ring, they're nearly back to that $3.5T mark after a bad year stock performance-wise. Their CapEx spend? Even after their own debacles with AI have now been exposed? Largely unchanged.

As I wrote six moths ago:

The other is the king of the hill: Apple. At $3.5T, if they were more in line with their peers, you'd think they'd be spending well north of $100B in CapEx this year. Instead, they're likely to be closer to $12B. That's not a typo. They've gone from $9B two years ago, to something like $11B last year. And extrapolating out from the Q1 numbers this year, you get to around $12B.

We all know that Apple isn't playing the same AI game as the rest, choosing instead to offload a lot of work and capabilities to partners, but it's staggering just how much less money they're spending on it relative to their peers. Perhaps this is prudent! And Wall Street certainly rewarded them post-DeepSeek, while dinging basically everyone else. But it may also be foolish. I'm not saying they have to spend $120B on AI, but you know, maybe more than a tenth of that (much of which likely isn't even related to AI, but their other CapEx needs)?

Well, two quarterly reports later, and Apple's CapEx is... projected to be somewhere between $11B - $14B for the year. Basically unchanged. Meanwhile, their Big Tech peers as of the most recent quarterly earnings:

• Meta: $70B
• Google: $85B
• Microsoft: $88B
• Amazon: $118B¹

Everyone revised CapEx spend upwards – and fairly significantly.² How significantly? Well, just the increases – I repeat, just the increases – are close to or above Apple's overall CapEx spend for the year.

Amazon spent over $30B on CapEx last quarter, which implies a spent rate over $120B at this point. Meta, for their part, already said they expect to be above the $100B mark in the next year (which puts their wild 'Godfather' talent offers into some context). Microsoft? After Satya Nadella ended up indeed "good" for his $80B – again, coming in above that spend for the year – he's now expecting to be spending around $120B next year on CapEx. Google has only said they expect the CapEx number to go up again next year, but didn't give a specific range. So at the very least, over the next year, all of this spend looks like:

• Google: $85B+
• Meta: $100B+
• Microsoft: $120B+
• Amazon: $120B+

We're clearly well past $400B in CapEx spend next year for those four companies alone. And it might end up past $500B. A half-trillion dollars. Almost all on AI.

Apple? Per their comments in their earnings call, they're committed to ramping up such spend – but not "exponentially". I'd be pretty surprised if it passes $20B for the year. It might not pass $15B.

That's not really a commentary on if it's the right or wrong strategy – it literally could still go either way at this point – it's just rather insane that their spending levels are 1/5th to 1/10th that of their immediate peer group. Yes, Apple has a different strategy – a "hybrid" approach, as they like to say. Which is just a fancy way of saying that they offload a bunch of that spend (and thus, compute) to partners – including many of these competing companies. But, I mean, they really have to have a lot of conviction in their own game plan not to be going after owning AI in the same way that those competitors are right now.

Again, they might be right! The others might spend themselves into oblivion. Probably not, but they could very well prove to be misguided in spending this much to build out capabilities that become more commoditized over time. It hasn't played out that way to date, but that doesn't mean it won't at some point. At the very least, Wall Street might turn on them and start hitting them hard on that spend, just as they did with the streaming services post-pandemic... And if that happens, Apple could be sitting pretty – having perhaps spent less in a decade on CapEx than their peers/rivals will have spent in a year at that point.

Of course, there's a flipside to such equations too. If AI ends up as the most transformative technology of all time and Apple is beholden to those peers/rivals to get access to such tech, that would be... not great for them! And that will be exacerbated if that tech leads to the creation of new devices that push the world beyond the smartphone. I'm not saying that's particularly likely at this point, but it's the risk Apple is taking here.

Of course, they would say – and, in fact, they are saying – that they often come late to markets and technology, focusing on the best product. I'm a dead horse on this point, but I just think AI will play out differently than they think it will. Yes, if the tech is eventually commoditized, that could work. But if not, Apple is always going to find themselves behind the eight ball. And more importantly, if they don't have the right talent in-house to harness such tech, the problems are going to be exacerbated. And well, they keep losing such talent.

Anyway, I'm still not saying Apple has to spend $100B+ a year on CapEx to train frontier AI models – it may end up being the smart play not doing that! But I am saying that it's eye-opening and seemingly risky that Apple clearly isn't truly trying to compete here – as the CapEx numbers illuminate. And the outcome seems very binary. They're either very right. Or very wrong.

¹ While Amazon didn't technically change their guidance, during their recent Q2 2025 earnings call, CFO Brian Olsavsky, said the $31.4B spent on CapEx for the quarter is "reasonably representative" of the CapEx rate for the rest of the year, which would put Amazon around $118B.

² It's more complicated than it seems because each of these companies are on slightly different fiscal calendars.