What you need to know

• Accelerating LLM capability, innovations in AI agents, and the ability of AI to turn text into actions will usher in profound changes to the world in the next two years, at a scale that almost no one understands.
• According to Eric Schmidt, the impact will be "Much bigger than the horrific impact we've had from social media."
• Frontier models like OpenAI and Anthropic look like they are pulling away from the pack, Schmidt says, a change on his view from just six months ago and something he cheerfully acknowledges despite the money he bet on that former perspective.
• But longer term there are two huge potential drags on the growth in AI - money and power.
• Schmidt says even if investment money becomes available - OpenAI CEO Sam Altman, told him the company needs US$300bn - there's not enough energy in the US to meet the demands of Altman and his industry peers.

Large language models (LLM) have reached the stage where they can interpret a million-word prompt and are on their way to 10 million, according to Eric Schmidt, Google CEO for a decade, then Google and Alphabet Chairman for almost a decade more.

This rapid capability development, along with innovations in AI agents, developments in text to action, and Nvidia's unassailable position in GPUs – the hardware driving AI in the data centre (plus a decade's worth of specific software libraries) – are set to have a profound effect on the world, Schmidt told attendees at an event at Stanford University recently.

“So what does all this mean? In the next year, you're going to see very large context Windows, agents, and text action that, when they are delivered at scale, are going to have an impact on the world at a scale no one understands," he said. “I'm quite convinced it's the union of those three things that will happen in the next wave.”

It's an impact that will be “Much bigger than the horrific impact we've had from social media", he cautioned.

Here’s why Schmidt believes this.

The first is developers can use context windows as short-term memory. "The interesting thing about short-term memory is that when you ask it a question like ‘read 20 books’ and you give it the text of the books in the query, then you say, ‘tell me what they say’, it forgets the middle. It's exactly how human brains work too. That's where we are," he explained.

With respect to agents, Schmidt noted people are now building LLM agents that can read something like chemistry, discover the principles of chemistry, test it, and then add that back into their understanding. Again, that's an extremely powerful capability now emerging, he said.

To illustrate the potential power of text to action, Schmidt gave an example drawn from the current debate in the US about banning TikTok. “If Tiktok is banned, here's what I propose: Each and every one of you say to your LLM the following: Make me a copy of Tiktok. Steal all the users, steal all the music, put my preferences in it, produce this program in the next 30 seconds, release it, and in one hour, if it's not viral, do something different along the same lines.

“That's the command, boom, boom, boom, boom. You understand how powerful that is.”

Such a shift from arbitrary language to arbitrary digital commands will upend the programming landscape. "Imagine that each and every human on the planet has their own programmer that does what they want," Schmidt continued.

“Imagine a non-arrogant programmer that actually does what you want, and you don't have to pay all that money to and there's infinite supply of these programs, and this is all within the next year or two.”

But before we get too far ahead of ourselves, there’s the small matter of funding it and powering it, Schmidt said.

Money and Power

Six months ago, Schmidt believed the gap between the frontier models (basically a state-of-the-art machine learning model that represents the pinnacle of current capabilities) of which he says now number three, and the plethora of newcomers was closing. Now it appears to be getting larger.

“Six months ago, I was convinced the gap was getting smaller, so I invested lots of money in the little companies. Now I'm not so sure. I'm talking to the big companies, and the big companies are telling me that they need $10 billion, $20 billion, $50 billion, $100 billion," he said.

For instance, Sam Altman at OpenAI, who Schmidt described as a close friend, told him the company needs $300bn and maybe more. But the sheer power required to deliver such AI power is beyond the reach of current power grids, Schmidt said.

“I told him I’d done the calculation about the amount of energy required," he said. “Then in the spirit of full disclosure, I went to the White House on Friday and told them that we need to become best friends with Canada, because Canada has really nice people, helped invent AI, and has lots of hydropower. The alternative is to have the Arabs fund it. I like the Arabs personally, spent lots of time there, but they're not going to adhere to our national security rules, whereas Canada and the US are part of a trumpet where we all agree."

Go fast or go under

Schmidt’s speech initially attracted a degree of controversy because he attributed Google's underwhelming performance in AI (relative to OpenAI and Anthropic) to its culture, specifically its work-from-home culture. He called out the need for speed in decision-making at a time of upheaval and disruption in the tech sector.

Explaining the context for the comments during his presentation, Schmidt said, “The reason I'm being so harsh about work is that these are systems which have network effects. Time matters a lot.”

That not true in most other businesses, he argued. “You have lots of time. You know, Coke and Pepsi will still be around, and the fight between Coke and Pepsi will continue to go along, and it's all glacial, right.”

Yet Schmidt noted when he worked with telco deals, it commonly took 18 months to sign, often for no good reason. “We are in a period of maximum growth, maximum gain."

He also flagged the importance of risk-taking. “It takes crazy ideas, like when Microsoft did the OpenAI deal. I thought that was the stupidest idea I'd ever heard, outsourcing essentially your AI leadership to OpenAI and Sam and his team. That’s insane. Nobody would do that at Microsoft or anywhere else," Schmidt said. "Yet today, they're on their way to being the most valuable company. They're certainly head-to-head with Apple, and Apple does not have a good AI solution.

“It looks like they [Microsoft] made it work.”