Hyperscaler capex spending is now projected to reach $3 trillion on a rolling 12-month forward basis, up from $500 billion in December 2022, representing an unprecedented commitment to long-term AI infrastructure.
The combined capex of Amazon, Google, Meta, Oracle, and Microsoft has grown sixfold in three years, with the hosts noting these companies deserve credit for pivoting from short-term share buybacks to massive long-term investments. ✦ AI generated
Michael Batnick · The Compound · 2026-08-12 · original ↗
starts at this moment · 3:15
Look at this chart from Hyperscaler Farmer uh or Hyperscale Farmer. Bloomberg consensus on a rolling basis now should pass $3 trillion. So, we're looking at the three-year sum on a rolling 12-month forward basis for Amazon, Google, Metal Meta, Oracle, and Microsoft. And in December 2022 when this thing started, it was at it was at $500 billion and and it's gone up and up and up and up and now it's going vertical and we're looking at $3 trillion over the next 3 years I suppose. Is that what we're looking at here? That's amazing. I think one of these things these companies don't get enough credit for is the fact that for years it was kind of like uh all these companies are too short-term in nature and they just care about share buybacks and and and now look at these companies are investing everything they have essentially into back into long-term projects.
verbatim transcript · starts at 3:15
3:15a big part of the story. The fact that the consumer has been able to be resilient through all of this is remarkable, but through the lens of the stock market, what matters obviously is is uh is the AI trade and the continued expansion of CapEx numbers. This is wild. Look at this chart from Hyperscaler Farmer uh or Hyperscale Farmer. Bloomberg consensus on a rolling basis now should
3:40pass $3 trillion. So, we're looking at the three-year sum on a rolling 12-month forward basis for Amazon, Google, Metal Meta, Oracle, and Microsoft. And in December 2022 when this thing started, it was at it was at $500 billion and and it's gone up and up and up and up and now it's going vertical and we're looking at $3 trillion over the next 3 years I suppose. Is that what we're
4:01looking at here? >> That's amazing. I think one of these things these companies don't get enough credit for is the fact that for years it was kind of like uh all these companies are too short-term in nature and they just care about share buybacks and and and now look at these companies are investing everything they have essentially into back into long-term projects. These things are not really paying out
4:26yet. And I think they deserve credit for saying, all right, fine. We're going to we're going to think long think and act for the long term. And people still hate it because people hate data centers and we'll get into that later, but this is this is it just there's never been anything like this before where the biggest companies just said, all right, we're changing everything. We're getting rid of all of
4:45our money. Take it. See you later. Hopefully hopefully this works. >> You mentioned uh the start of the show all the all-time highs that were enjoying I mean I hope you're enjoying them. I sure am. Um and uh it won't last forever as we know. So, what could be on the other side of this? I think this is this is the obvious danger. Is this line is now such a gigantic I
5:08mean it's it's going absolutely vertical and it can't continue to do this forever. It's not going to go from 3 trillion to 4 trillion to 7 trillion. Like at some point there's an upper limit. And uh what's awesome about this cycle is that you're not relying on these fly-by-night companies that have no earnings or anything like that. But the hyperscalers are powering a lot of the earnings growth that we're seeing
- ·Rolling 12-month forward capex now projected at $3T
- ·Up from $500B in December 2022
- ·Sixfold increase in just three years
- ·Amazon, Google, Meta, Oracle, Microsoft lead spending
- ·Pivoted from short-term share buybacks
- ·Unprecedented commitment to AI infrastructure