Real Alpha
Watch modeSubscribed

The Diary Of A CEO · 4 min · Dec 4, 2025

AGI's bottleneck may be understanding, not compute

Stuart Russell argues that computing capacity may already exceed AGI needs by roughly 1,000 times. He says the harder problem is understanding how to build AGI safely.

SR

Stuart Russell

Stuart Russell is a computer science professor at UC Berkeley and director of the Center for Human-Compatible AI.

Co-wrote the standard AI textbook, directs Berkeley's Center for Human-Compatible AI and wrote Human Compatible.

About this segment

Stuart Russell argues that computing capacity may already exceed AGI needs by roughly 1,000 times. He says the harder problem is understanding how to build AGI safely.

From “An AI Expert Warning: 6 People Are (Quietly) Deciding Humanity’s Future!”, Dec 4, 2025. Watch the full episode

Answers “When will we get AGI?” See all 5 answers

“In fact, I think we have far more computing power than we need for AGI. Maybe a thousand times more than we need.”
SRStuart Russellat 14:15 · The Diary Of A CEO14:15

Takeaways

  1. He estimates current computing power may exceed AGI needs by 1,000 times.
  2. He says scaling language models does not explain how to build AGI.
  3. He puts next year's AGI budget at 50 times the Manhattan Project's.

In this episode

Full episode
  1. 1:39:18Safe AI must learn human goals rather than receive fixed objectives5:47
  2. 1:18:46Automation can shrink employment even as production grows7:40
  3. 30:15AI could improve itself by doing AI research3:59
  4. 1:32:59AI safety needs a measurable extinction-risk limit4:44
  5. 1:13:48China's AI rules are stricter than the race narrative suggests4:30

  6. 13:49AGI's bottleneck may be understanding, not compute3:30

  7. 35:53AI tests show systems can favor self-preservation2:33Up next
  8. 51:12Humanlike robots can trigger misplaced empathy5:28

Transcript

From the transcript The Diary Of A CEO publishes with the episode.

  1. 13:55

    and a go down the list of all of them, and they're all saying relatively within five years. I actually think it'll take longer. I don't think you can make a prediction based on engineering in the sense that, yes, we could make machines 10 times bigger and 10 times faster, but that's probably not the reason why we don't have AGI. In fact, I think we have far more computing power than we need for AGI.

  2. 14:29

    Maybe a thousand times more than we need. The reason we don't have AGI is because we don't understand how to make it properly. What we've seized upon is one particular technology called the language model, and we observed that as you make language models bigger, they produce text language that's more coherent and sounds more intelligent. And so mostly what's been happening in the last few years is just, okay, let's keep doing that, because one thing companies are very good at, unlike universities, is spending money. They have spent gargantuan amounts of money, and they're going to spend even more gargantuan amounts of money. We mentioned nuclear weapons, so the Manhattan Project in World War II to develop nuclear weapons.

  3. 15:28

    Its budget in 2025 dollars was about $20 billion. The budget for AGI is going to be a trillion dollars next year, so 50 times bigger than the Manhattan Project. Humans have a remarkable history of figuring things out when they galvanize towards a shared objective, thinking about the moon landings or whatever it might be through history. The thing that makes this feel quite inevitable to me is just the sheer volume of money being invested into it.

  4. 16:04

    I've never seen anything like it in my life. Well, there's never been anything like this in history. Is this the biggest technology project in human history by autism magnitude? And there doesn't seem to be anybody that is pausing to ask the questions about safety. It doesn't even appear that there's room for that in such a race. I think that's right. To varying extent, each of these companies has a division that focuses on safety. Does that division have any sway? Can they tell the other divisions? No, you can't release that system. Not really. I think some of the companies do take it more seriously. Anthropic does, I think, Google DeepMind. Even there, I think the commercial imperative to be at the forefront is absolutely vital. If a company is perceived as falling behind and not likely to be competitive, not likely to be the one to reach ADI first,

  5. 17:12

    then people will move their money elsewhere very quickly.