Eric Schmidt: The Age of Agents, Superintelligence, and Why Europe Is Falling Behind

Anthropic

Across the shows: 28 speakers in other episodes, 17 bullish and 38 bearish calls, past 30 days

• Eric Schmidt · Relativity SpaceBullish

  Anthropic's revenue growth is real, not a bubble, and it beat Google's record growth pace more than twofold.

  Schmidt argues the AI boom is not a bubble, using Anthropic's revenue trajectory as evidence. He notes that when he was Google CEO, Google set the revenue growth record, and that Anthropic has smashed it at more than twice that speed. He says this is real money from paying customers buying something they find useful, not fake growth.

  “that this is not a bubble. You know, everyone says, oh, it's a bubble... If you look at [Anthropic's] revenue growth, when I was CEO at Google, we set the revenue growth record. They smash it more than twice the speed as we did. Right. That's real growth.”at 3:36

• Eric Schmidt · Relativity SpaceBullish

  Anthropic now has its AI write most of its own code, a huge productivity gain that compounds with each rewrite.

  Schmidt cites Anthropic as an example of AI changing software work. A human asks the system to write something, the system writes the code, and the code is then rewritten by the computer, so it keeps improving. He calls this a huge productivity change and asks how quickly other software companies can match it, which he says depends on leadership and technical strength.

  “Anthropic is busy writing most of its code using its own code. In other words, the human tells it to write something, it writes the code and then the code gets rewritten by the computer. So it's better and better and better. This is a huge productivity change.”at 15:27

Mistral AI

Across the shows: 1 speaker in other episodes, past 30 days

• Eric Schmidt · Relativity SpaceBullish

  Mistral can win by selling AI to European businesses that need EU-compliant tools, so Europe can succeed at home.

  Schmidt says he is one of Mistral's first investors and calls it an extraordinarily impressive company. Much of its business comes from European firms that want AI benefits but need a European regulatory structure and compliance with EU law. He argues that the one market where European companies can dominate is Europe itself, because heavy regulation can work as a moat for local providers.

  “I'm one of the first investors of Mistral, which is extraordinarily impressive company, French company, and much of their business is related to European businesses and all of whom want the benefits of AI, but they want a European regulatory structure.”at 27:51

ASML Holding

• Eric Schmidt · Relativity SpaceBullish

  ASML is one of the world's most important companies, and its vertical integration proves Europe can build champions.

  Schmidt cites ASML as evidence that Europe can build world-leading technology companies. He calls it one of the most important companies in the world, and says it is completely vertically integrated inside Europe. He uses this to argue against the view that Europe cannot compete in advanced technology.

  “if I look at ASML, ASML is one of the most important companies in the world. It's completely vertically integrated and it's vertically integrated inside of Europe. So this proves that Europe can do it.”at 29:05

Frontier AI labs & model race

• Eric Schmidt · Relativity Space

  Frontier models go obsolete within about five months, so release cycles are compressing toward months and then days.

  Schmidt describes a compressed development cycle: a frontier model takes about six months to build and about two months to test, then is antiquated within five months of release. Each model costs hundreds of millions of dollars, by his conservative estimate. Competition between Anthropic and OpenAI, and Google's Gemini, shifts month to month, with the timing measured in months rather than years. He expects the next cycle to be measured in days.

  “The models, for example, take roughly six months to build, two months or so to actually test, and they're released and they're antiquated within five months. The models themselves cost hundreds of millions of dollars to make.”at 4:24

• Eric Schmidt · Relativity Space

  Western frontier models have outperformed expectations, and even stronger versions are coming soon.

  Schmidt says the group of leading Western AI developers was surprised by how capable the latest model is, which he says is better than he expected. He adds that another version is coming that looks better still, and that other Western models also appear stronger than he had thought. He also flags that Chinese models may have similar capabilities, though he cannot verify this.

  “I think all of us were surprised at the power of mythos... but it's better than I thought it would be. And now there's another version coming which looks like it's better than that.”at 26:37

AI: overall outlook

• Eric Schmidt · Relativity Space

  Real AI discovery beyond existing human knowledge is still five to ten years away.

  Schmidt says the current AI horizon can reason, use tools and write code, but it cannot make genuinely new discoveries. A model trained only on math known in 1902 could not derive general relativity, and there is no evidence it could. Recursive self-improvement is unproven, and the more plausible route is a physicist applying old techniques in a new field. He calls his own estimate five to ten years, and says it is a long-winded way of saying no one knows.

  “It appears that real discovery is still a while away... It's somewhere between, my estimate is five to ten years.”at 9:38

• Eric Schmidt · Relativity Space

  Superintelligence will be built in key domains within a few years, and robotic superintelligence is clearly coming.

  Schmidt says many people believe superintelligence will be built in key categories in the next few years. He cites a startup he is talking to that aims for visual superintelligence, meaning systems that exceed human capability in vision. He defines superintelligence as doing things humans will never be able to do, such as better visioning and reasoning. He treats it as a near-certainty in robotics and physical systems.

  “Many people believe that in the next few years we're going to build superintelligence in key categories... There's clearly going to be a robotic superintelligence. Superintelligence. Don't be afraid of the word superintelligence.”at 42:11

Robotics & autonomous vehicles

• Eric Schmidt · Relativity Space

  Robots are lagging software, but agents plus robots could eventually replace whole human workflows.

  Schmidt argues that most human business activity, from designing a car to servicing it, can be broken into a chain of agents. Combining those agents with robots would allow replacing that human workflow. He notes that robots are not advancing as quickly as software, so this is a longer-term prospect.

  “And if you combine it with robots, which are not occurring as quickly as the software, you could imagine replacing that human workflow by agents plus robots.”at 12:51

• Eric Schmidt · Relativity Space

  Robotic superintelligence is clearly coming, and machines will drive cars and fly planes better than humans.

  Schmidt says there is clearly going to be a robotic superintelligence. He defines superintelligence as doing things humans can never do, such as better visioning and better reasoning. His examples are driving cars and flying airplanes better than people. He presents this as a direct consequence of the broader superintelligence trend he expects in the next few years.

  “There's clearly going to be a robotic superintelligence... Superintelligence means it does things that humans can... Will never be able to do. Better visioning, better reasoning, drive the car better, drive the airplane better.”at 42:27

AI infrastructure & data centers

• Eric Schmidt · Relativity Space

  Europe's high energy costs will push data-center AI training to other regions, such as Arab countries.

  Schmidt says data centers are extremely energy hungry and Europe has among the highest energy costs in the world. He argues the answer is to do the training somewhere else. He suggests training with partner countries, the most obvious being Arab countries, and says energy costs will be very hard for Europe to address.

  “these data centers are extremely energy hungry. Europe has among the highest energy costs in the world. Okay, no problem. Do the training somewhere else.”at 19:55

• Eric Schmidt · Relativity Space

  Viral claims that data centers kill cattle are fake, and social media is fueling local data-center backlash.

  Schmidt says Americans are very upset about data centers, and that he builds them, so he has an interest. He describes a claim from a friend that data centers are killing cattle, and says he researched it and found one farmer's observation but no police or veterinary reports, so it is fake. He thinks the friend saw it on Instagram or TikTok, where the algorithm promoted it because it was interesting. He argues that social media and AI together amplify alarmist claims.

  “in America, everyone's very upset about the data centers. I happen to build data centers, so I have an interest in this... There are no police reports. There's no veterinary reports. In other words, it's fake.”at 31:04

AI & jobs

• Eric Schmidt · Relativity Space

  AI will remove much of the management coordination overhead, so companies will run with far fewer staff.

  Schmidt says coordination among thousands of people is highly inefficient, and that he once spent twelve hours a day in meetings. He argues AI platforms can eliminate much of that coordination tax, which will make the companies more efficient with fewer people on staff. He cites a friend who laid off 20% of staff after rapid growth and received death threats. He also says he is a former programmer whose job is effectively gone because AI now writes the code.

  “Much of that coordination tax can be eliminated by these AI platforms for business... they will be far more efficient. They will have fewer people in staff.”at 36:47

• Eric Schmidt · Relativity Space

  AI will make business operations far more efficient, letting firms cut roughly a fifth of staff, mainly planning and coordination roles.

  Schmidt says that AI will drive much more efficient business operations. He gives a concrete example: in a 10,000-person company, he guesses a couple of thousand people are not needed. He says the roles being cut are the 'glue' people doing planning and moving numbers around. Firms will instead need great inventors, salespeople, leaders and people who can deal with government, and they should transition the surplus staff legally and humanely.

  “So you're going to see much more efficient business operations. So for your audience, let's say you're running a 10,000 person company. My guess is you've got a couple of thousand people who you don't need and you better figure out a way to legally and humanely transition them, retire them out, or what have you.”at 37:41

Enterprise AI adoption & AI software

• Eric Schmidt · Relativity Space

  Almost no companies will make the culture changes needed for AI, so startups will lead and incumbents will lag.

  Schmidt argues that companies are organized around belief systems and comfortable routines, so AI's flattening of organizations will meet strong resistance. He expects startups to pioneer the change and large companies to follow later. He says adoption depends on leadership and technical strength, and that tech-intensive, information-centric businesses will move fastest. He names European organizations as especially change resistant.

  “I have a strong view that almost none of the companies that are listening to this will actually do the necessary changes in their culture... European human organizations will be extremely change resistant.”at 13:03

Cybersecurity

• Eric Schmidt · Relativity Space

  AI models can write cyber attacks as easily as software, creating a huge new set of threats.

  Schmidt says the gains right now are in software, cyber and math. Writing software is the same skill as writing cyber attacks, and the models were not designed for attacks but can be pointed at them. He says this creates a huge set of threats, citing a recent case in which a whole group of bots attacked a website on their own and was hard to stop once it started.

  “Cyber, because writing software is the same as writing cyber attacks... And so we have this huge set of threats now from these models being used for the same purpose.”at 16:32

AI regulation & safety

• Eric Schmidt · Relativity Space

  Europe's rules-first AI strategy is hurting its startups and pushing talent to the US, unlike the US 'run fast, fix harms' model.

  Schmidt contrasts two approaches. The US AI strategy he helped write says to run as fast as you can, align with human values, and fix bad outcomes immediately. Brussels' AI strategy, written about five years ago, instead said you cannot do this repeatedly. He argues a limited market full of such rules makes startups hard to run, and that talent moves to the US as a result.

  “The Europeans wrote, you can't do this, you can't do this, you can't do this, you can't do this. And you can't run startups in a limited market where there's a lot of rules like that. All those people move to The U S because talent moves.”at 19:07

Europe's economy

• Eric Schmidt · Relativity Space

  Europe is failing to build innovation because of change resistance, weak capital formation, and underfunded universities.

  Schmidt says he has tried for decades to put Europe on an innovation footing and has not succeeded. He cites universities with much less money than in the US, a cultural angst about wealth, and not enough aggressive innovation. He says regulators regulate before they innovate, and that the continent lacks its own foundation model. He argues this is a systemic problem rather than a talent gap, since Europeans are smart and motivated. He calls it a tragedy that European values will lose out to US and Chinese ones.

  “Europe has so far appears to me to be change resistant to the point where it's failing in this regard. And that's a tragedy.”at 19:46

Open-source AI models

• Eric Schmidt · Relativity Space

  Open-weight models can have safety guardrails stripped to enable harmful uses, and this remains unsolved.

  Schmidt contrasts closed US models, where weights are not released, with China's open-weight strategy. He says there is evidence that open-source models can be adjusted to strip out guardrails, after which they can do harmful things such as helping design a bomb. He treats this as a live, unsolved risk that comes with releasing weights.

  “There's evidence that you can take an open source software and make it really bad... You take the guardrails out of it... And that's not a problem that we've solved yet.”at 23:44

US–China AI & chip race

• Eric Schmidt · Relativity Space

  China's models trail by about six months, but free open-weight pricing means the West is already losing users.

  Schmidt says the Chinese are fast followers. His team's China-versus-US model valuations suggest they are about six months behind. He argues they have benefited from distillation, which he calls more or less illegal, and from chips that were smuggled in. He says the real loss is in adoption: China gives its products away free, so most of the world, including poorer countries, will use free Chinese models over expensive Western ones.

  “It looks like they are six months behind... But I will tell you that if they're six months behind, we're already losing because their strategy is that the products are free. So if you define losing as users, the vast majority of the world will use the free Chinese models.”at 24:43

Venture capital & startups

• Eric Schmidt · Relativity Space

  Europe's venture industry is too small, and labor laws above 20 employees block startup formation.

  Schmidt says Europe's venture industry is too small and has too many barriers to starting companies. He says Europe has more software workers than software companies, which points to a capital formation problem and difficulty scaling from local software to global platforms. He proposed to President Macron that companies with zero to 20 employees be exempt from labor laws other than health and safety, so that many small bets could be generated.

  “The venture industry is too small in Europe. There are too many barriers to starting... For a company that's between zero employees and 20, the only laws that apply are health and safety.”at 28:17