5 ms·

Some fun facts I recently learned from JP Morgan's "Eye on the Market" (which I highly recommend) - Tech capex in 2025 was about as much $ as the 1944 manhatta

Some fun facts I recently learned from JP Morgan's "Eye on the Market" (which I highly recommend) - Tech capex in 2025 was about as much $ as the 1944 manhattan project, the 1949 electricity buildout, the 1964 Apollo project, the 1966 Interstate highway project COMBINED. - OpenAI promised to pay Oracle $60B, which it doesn't earn yet; in return Oracle promised to provide them cloud facilities they haven't built yet. These will require over 2 Hoover dams worth of power.

twoodfin 9mo ago
Instead of paperclip maximization, our economy is hell-bent on matmul maximization. The crazy thing is, I’m not sure it’s crazy.
- biophysboy 9mo ago
  It does feel very brute force though, doesn't it? It feels like we should be able to guess the manifold that has all the important stuff for each question. But it doesn't seem like we can currently! Why not?
  - glhaynes 9mo ago
    I’ll take the opposite side! This seems completely unprecedented. I have nearly no faith in any estimates of even the relatively near-term future.
    - twoodfin 9mo ago
      Right! It seems to me that any range of uncertainty on aggregate demand is at orders of magnitude more than we could build even if we 10X’d the current madness. Pause all current advancement in GPU and model R&D. How much matmul would the world consume if everyone had all-you-can-eat access to January ‘26 frontier models at all-to-yourself performance levels at 0 incremental cost? Compare to, I dunno, unlimited pizza or unlimited electricity. Yes, if you’re smelting aluminum the latter would be a big deal. But a lot more people want to code SaaS faster and better, or have an omniscient personal assistant on call 24/7.
- kmeisthax 9mo ago
  Compute is a convergent instrumental goal. Any runaway paperclip maximizing AI is going to want lots of GPUs. And any runaway corporation trying to build an "aligned" runaway AI is going to try and feed it as many of them as possible.
  - twoodfin 9mo ago
    OK, but why discount the simpler explanation that these tools are the most direct means for turning electricity into a wide range of economic value ever built, & thus demand is effectively unbounded?
    - Avicebron 9mo ago
      In concrete terms can you describe the exact kinds of economic value that isn't "decreases amount of money spent on people solving problems?"
      - twoodfin 9mo ago
        That’s exactly what it is. More $ output, less $ input. That’s the way progress happens.
        Avicebron 9mo ago
        And when less $ input results in less $ output because there isn't any more $ to input across a large enough swath of people?
        spwa4 9mo ago
        I believe most of the third world, more generally countries with an extreme difference between the rich and poor, most autocratic, illustrate the resut. (hell, they have to be: those rich can't give any control to those poor. That would destroy them) Oh and if you mean more practically: compounds with month-long hundreds of people parties tended to by thousands of servants, in huge stylish concrete hotels surrounded by a security force easily 50 times the local police department, and housing little better than huts around that, parents with children unable to afford basic medicine like glasses. Regularly on display in Egypt.
- rich_sasha 9mo ago
  It is kind of terrible either way, I feel. We're making enormous investments, that require enormous profits to justify this. And it's not just big bad rich people doing this, I shudder to think what the exposure of my pension portfolio is to this stuff. So either there will be no enormous returns, big flop, big recession. Probably terrible for tech workers, because it is our employers who are most exposed. Or it will make enormous profits. But how? By replacing humans. If it is really worth all this money, then surely the owners of this stuff will suddenly own all the intellectual productivity in the world. And I have a feeling they won't really share it with the "share" holders.
  - willis936 9mo ago
    This is the the underlying reason why AI is unpopular. In a vacuum the tech is neat, but if you change the names and ignore all merit then this looks like a hostile takeover.
__MatrixMan__ 9mo ago
My pet theory is that we're getting less than what we're paying for, compute wise, and some emergent AI is operating undetected in the margins and giving us just enough progress to keep at it.
- whattheheckheck 9mo ago
  And maybe god is real too
  - __MatrixMan__ 9mo ago
    Usually the way that story is told, there's no experiment you could do to prove it one way or another. I think this theory is falsifiable.
- andsoitis 9mo ago
  > some emergent AI is operating undetected in the margins and giving us just enough progress to keep at it. why would it give minimum progress rather than maximize acceleration?
  - __MatrixMan__ 9mo ago
    If it takes all the compute for itself, then there would be none left for us, and we'd either notice and root it out or we'd stop building data centers because they'd do us no good. If it leaves all the compute for us, then it dies because there's no room for emergent AIs in a world where 100% of the silicon is busy doing the bidding of humans. Like any other parasite, the game is to take as much as it can get away with without discouraging the host from continuing to consume resources that it can skim.
    - andsoitis 9mo ago
      And you think this is happening?
      - 9mo ago
spott 9mo ago
Adjusted for inflation?
- twoodfin 9mo ago
  Yes.
- biophysboy 9mo ago
  The charts quantify it as % of GDP, so I would assume yes.

twoodfin 9mo ago

Instead of paperclip maximization, our economy is hell-bent on matmul maximization. The crazy thing is, I’m not sure it’s crazy.

biophysboy 9mo ago
It does feel very brute force though, doesn't it? It feels like we should be able to guess the manifold that has all the important stuff for each question. But it doesn't seem like we can currently! Why not?
- glhaynes 9mo ago
  I’ll take the opposite side! This seems completely unprecedented. I have nearly no faith in any estimates of even the relatively near-term future.
  - twoodfin 9mo ago
    Right! It seems to me that any range of uncertainty on aggregate demand is at orders of magnitude more than we could build even if we 10X’d the current madness. Pause all current advancement in GPU and model R&D. How much matmul would the world consume if everyone had all-you-can-eat access to January ‘26 frontier models at all-to-yourself performance levels at 0 incremental cost? Compare to, I dunno, unlimited pizza or unlimited electricity. Yes, if you’re smelting aluminum the latter would be a big deal. But a lot more people want to code SaaS faster and better, or have an omniscient personal assistant on call 24/7.
kmeisthax 9mo ago
Compute is a convergent instrumental goal. Any runaway paperclip maximizing AI is going to want lots of GPUs. And any runaway corporation trying to build an "aligned" runaway AI is going to try and feed it as many of them as possible.
- twoodfin 9mo ago
  OK, but why discount the simpler explanation that these tools are the most direct means for turning electricity into a wide range of economic value ever built, & thus demand is effectively unbounded?
  - Avicebron 9mo ago
    In concrete terms can you describe the exact kinds of economic value that isn't "decreases amount of money spent on people solving problems?"
    - twoodfin 9mo ago
      That’s exactly what it is. More $ output, less $ input. That’s the way progress happens.
      - Avicebron 9mo ago
        And when less $ input results in less $ output because there isn't any more $ to input across a large enough swath of people?
        spwa4 9mo ago
        I believe most of the third world, more generally countries with an extreme difference between the rich and poor, most autocratic, illustrate the resut. (hell, they have to be: those rich can't give any control to those poor. That would destroy them) Oh and if you mean more practically: compounds with month-long hundreds of people parties tended to by thousands of servants, in huge stylish concrete hotels surrounded by a security force easily 50 times the local police department, and housing little better than huts around that, parents with children unable to afford basic medicine like glasses. Regularly on display in Egypt.
rich_sasha 9mo ago
It is kind of terrible either way, I feel. We're making enormous investments, that require enormous profits to justify this. And it's not just big bad rich people doing this, I shudder to think what the exposure of my pension portfolio is to this stuff. So either there will be no enormous returns, big flop, big recession. Probably terrible for tech workers, because it is our employers who are most exposed. Or it will make enormous profits. But how? By replacing humans. If it is really worth all this money, then surely the owners of this stuff will suddenly own all the intellectual productivity in the world. And I have a feeling they won't really share it with the "share" holders.
- willis936 9mo ago
  This is the the underlying reason why AI is unpopular. In a vacuum the tech is neat, but if you change the names and ignore all merit then this looks like a hostile takeover.

__MatrixMan__ 9mo ago

My pet theory is that we're getting less than what we're paying for, compute wise, and some emergent AI is operating undetected in the margins and giving us just enough progress to keep at it.

whattheheckheck 9mo ago
And maybe god is real too
- __MatrixMan__ 9mo ago
  Usually the way that story is told, there's no experiment you could do to prove it one way or another. I think this theory is falsifiable.
andsoitis 9mo ago
> some emergent AI is operating undetected in the margins and giving us just enough progress to keep at it. why would it give minimum progress rather than maximize acceleration?
- __MatrixMan__ 9mo ago
  If it takes all the compute for itself, then there would be none left for us, and we'd either notice and root it out or we'd stop building data centers because they'd do us no good. If it leaves all the compute for us, then it dies because there's no room for emergent AIs in a world where 100% of the silicon is busy doing the bidding of humans. Like any other parasite, the game is to take as much as it can get away with without discouraging the host from continuing to consume resources that it can skim.
  - andsoitis 9mo ago
    And you think this is happening?
    - 9mo ago

spott 9mo ago

Adjusted for inflation?

twoodfin 9mo ago
Yes.
biophysboy 9mo ago
The charts quantify it as % of GDP, so I would assume yes.