In an A16Z interview, Sam Altman did something that left a deep impression on me: an AI company CEO repeatedly returned to energy issues throughout the conversation.

Not talking about how impressive GPT-5 is, not discussing when AGI will arrive, but talking about electricity, nuclear fusion, and infrastructure.

This reveals something many people haven’t yet realized: The ceiling of the AI revolution isn’t in algorithms—it’s in infrastructure. More precisely, in electric power.

The Flywheel: Reducing the Cost of Intelligence

Altman described OpenAI’s business logic as essentially a three-stage flywheel.

The first stage is the frontier research lab—creating the most advanced AI models. The second stage is large-scale infrastructure—deploying models at a scale that massive numbers of users can access. The third stage is personalized applications—enabling everyone to use AI for their own purposes.

These three stages form a self-reinforcing cycle: research produces better models, models bring more users, users generate revenue, revenue funds larger infrastructure and more research.

What’s the core driving force of this flywheel? Continuously reducing the “cost of intelligence.”

Think about it: ten years ago, to get deep analysis in a specialized field, you needed to pay for consultants. Today, you can use ChatGPT to get at least an eighty-percent answer within minutes. The cost of intelligence is declining at an astonishing rate.

The lower the cost, the more widespread the usage. The more widespread the usage, the greater the returns. The greater the returns, the more infrastructure can be invested. The more infrastructure, the lower the costs continue to fall.

This is the power of the flywheel.

Sora: More Than a Video Generator

Altman’s positioning of Sora in the interview differs from most people’s understanding.

Most people view Sora as an “AI video generation tool”—you input text, it outputs video. But Altman believes Sora’s true significance lies in: it’s the starting point of a world simulator.

Why? Because to generate realistic videos, AI can’t just “draw” images—it must understand the causal laws of the physical world. How does a ball move after being thrown? How does light refract when hitting water? How does a person’s center of gravity shift when walking?

These seemingly simple things mean AI must construct an internal model of the physical world. It’s not “generating videos”—it’s “simulating worlds.”

What’s the extension of this direction? Altman mentioned “AI scientists”—future AI won’t just analyze data, it can participate in the process of scientific discovery: proposing hypotheses, designing experiments, predicting results. If AI can truly achieve this, the pace of scientific progress will undergo qualitative change.

In “AI Always On: The Economic Order Being Restructured,” I discussed how AI is restructuring economic operational logic. But Altman’s vision extends further—he sees not just economic restructuring, but the restructuring of the scientific method itself.

Energy: The Ultimate Bottleneck

This was the most eye-opening part of the entire interview.

AI’s computational demands are growing exponentially. And what’s behind compute power? Electric power.

Training a large language model consumes electricity equivalent to several months’ worth of a small city’s power usage. And as models get larger and users multiply, electricity demand will only continue soaring.

Altman’s energy strategy spans three time scales. Short-term reliance on natural gas—not environmentally friendly, but reliable. Medium-term reliance on solar plus energy storage technology—increasingly cheaper, but with intermittency issues. The long-term ultimate bet is nuclear fusion.

Why nuclear fusion? Because nuclear fusion has the potential to bring order-of-magnitude reductions in energy costs—Helion Energy’s long-term goal is 1 cent per kilowatt-hour, compared to current U.S. average prices of about 12-15 cents, meaning costs could potentially drop to one-tenth of current levels. If nuclear fusion truly materializes, the constraints on AI compute expansion would be completely eliminated.

This is why Altman personally invested in nuclear fusion company Helion Energy. He’s not doing charity work—he’s buying insurance for AI’s future.

In “Comprehensive Analysis of Sovereign AI,” I discussed how autonomous AI development requires technological sovereignty and data sovereignty. But Altman reveals an even more fundamental sovereignty: energy sovereignty. Without stable and abundant electricity supply, all AI dreams remain empty talk.

Sunrise, Not Explosion

At the end of the interview, Altman used a metaphor: AGI’s arrival won’t be an instantaneous explosion, but more like a sunrise.

Light doesn’t illuminate the entire world with a sudden “snap.” It gradually, slowly, bit by bit seeps through from the horizon. You won’t notice a specific moment when “the sun has risen”—you’ll only realize in retrospect that the world has become bright.

This metaphor is important. Because many people’s imagination of AI is either the panic of “it will instantly replace everyone” or the dismissiveness of “it’s just a tool, no big deal.”

The reality might lie between these extremes: AI’s impact is gradual, continuous, and irreversible. Society will have time to adapt, but “having time” doesn’t equal “automatic adaptation.” You must actively adjust your position to align with the rhythm of this transformation.

Alignment, Not Opposition

The greatest takeaway from Altman’s interview for me wasn’t any specific technical detail, but a way of thinking: The AI revolution isn’t an event, but a process. It requires not just technological breakthroughs, but energy, infrastructure, social institutions, talent development—supporting systems across the board.

A thriving AI ecosystem needs stable, predictable, trustworthy platforms. This is an engineering of trust, not a technological arms race.

Before dawn, the most important thing isn’t predicting when the sun will fully rise. It’s adjusting your position—ensuring that when the light comes through, you’re standing on the side that welcomes it, rather than turning your back to it.