TMTPOST -- Artificial intelligence (AI), perhaps, is simply energy that has found self-awareness, not as a competition over algorithms or computing chips, but as a civilizational transition driven by energy, said Zhang Lei, chairman of Envision Group.
Zhao shared his reflections in a conversation with Jany Hejuan Zhao, the founder and CEO of NextFin.AI and the publisher of Barron's China, during the 2025 T-EDGE conference, which kicked off on December 8, and runs through December 21. The annual event brings togehter top scientists, entrepreneurs and investors to discuss pressing issues of the AI era.
展开剩余91%He argued that the explosive rise of artificial intelligence is forcing humanity to confront a deeper constraint than data, talent or hardware: the physical limits of energy itself. In his view, intelligence — whether biological or artificial — is ultimately an energy phenomenon. And as AI grows more powerful, it will demand a transformation of the global energy system no less profound than the Industrial Revolution.
From the aging power grids of the United States to China’s rapid build-out of renewable energy, from data center bottlenecks to the economics of green hydrogen, Zhang sketched a vision in which energy becomes the decisive factor in the AI race — and perhaps the defining axis of future civilization.
From “Compute Anxiety” to “Energy Anxiety”
Over the past year, the technology industry’s dominant concern has quietly shifted. The earlier panic over shortages of advanced chips has given way to a new worry: electricity.
Executives from Nvidia, OpenAI, Microsoft and Tesla have warned that even if enough chips are available, data centers cannot be deployed fast enough because power infrastructure cannot keep up. In some parts of the United States, AI companies are reportedly waiting up to five years just to connect new facilities to the grid.
Zhang said this is not a temporary glitch, but the result of structural limits.
“The U.S. power grid was largely built in the 1970s and 1980s, when demand growth was only one or two percent a year,” he said. “Now AI is pushing demand growth beyond five percent, and the system simply cannot absorb that shock.”
More critically, he noted, roughly 90 percent of U.S. computing centers rely on electricity generated from natural gas. But multiple studies predict that U.S. gas production will peak around 2035.
“That means today’s AI boom is being built on an energy foundation that is finite, volatile and environmentally unsustainable,” Zhang said. “If energy prices spike or supply tightens, the cost structure of AI collapses.”
In other words, the very infrastructure meant to support a new technological era may already be approaching its limits.
Intelligence as an Energy Phenomenon
Zhang’s argument goes beyond economics or infrastructure. He situates AI within a much larger physical and philosophical framework.
According to the second law of thermodynamics, the universe naturally trends toward disorder. Creating and maintaining any form of order — a living organism, a city, a legal system, or a neural network — requires energy.
“Life is a system that acquires, organizes and cycles energy,” Zhang said. “A rock does not do that, so it is not alive. Civilization is similar. It is a higher-level energy organization system.”
From this perspective, AI is not merely software. It is a new form of ordered structure that consumes enormous energy to exist.
Training large models already requires massive amounts of electricity. Future “physical AI” — systems that simulate climate, oceans, logistics, or planetary processes in real time — would demand orders of magnitude more.
“As intelligence grows, its appetite for energy grows without an obvious upper bound,” Zhang said. “Material needs saturate. Intelligence does not.”
This, he argued, explains the sudden sense of “energy anxiety” in the tech world: AI has revealed a mismatch between human ambition and the energy systems we currently possess.
China-U.S. Energy Gap
Zhang contrasted the energy positions of China and the United States in the emerging AI race.
China, he argued, does not face an absolute shortage of electricity. It has built one of the world’s most interconnected and efficient power grids and now adds more renewable energy capacity each year than any other country.
However, China still needs a new type of energy system tailored specifically to AI — one that is ultra-stable, ultra-cheap, and deeply integrated with computing infrastructure.
The U.S., by contrast, faces a more fundamental constraint.“America’s problem is not optimization,” Zhang said. “It is access.”
Fragmented regional grids, long permitting processes, and heavy dependence on fossil fuels mean that power itself has become a bottleneck for AI deployment.
In a technology race where product cycles last months, not years, a five-year wait for grid access is effectively fatal.
“Energy is becoming a geopolitical variable,” Zhang said. “Not because it is scarce globally, but because systems for producing and delivering it are uneven.”
To illustrate how energy can unlock civilizational leaps, Zhang drew a historical analogy.
For most of human history, population growth was constrained by food production — the so-called Malthusian trap. That changed only when chemical fertilizers allowed farmers to inject fossil-fuel energy into agriculture, dramatically increasing yields.
Zhang called this the “fertilizer moment.”AI, he suggested, now needs its own equivalent.
“For AI to scale, energy costs must fall another 50 percent — maybe 80 percent — from today’s levels,” he said.
Fossil fuels cannot provide that. They are finite, geopolitically fraught, and increasingly expensive. Nuclear fusion remains speculative and costly.
Renewable energy, by contrast, offers abundance. The sun delivers more energy to Earth in one hour than humanity consumes in a year. Over the past three decades, the cost of solar and wind power has fallen by about 99 percent.
The challenge is not generation, but management: intermittency, storage, transmission and real-time balancing.
This is where Zhang sees a self-reinforcing loop between AI and energy.
AI systems can optimize weather forecasting, grid management, storage deployment and energy markets — making renewables cheaper and more reliable. Cheaper energy, in turn, enables larger and more powerful AI systems.
“Energy and intelligence will evolve together,” Zhang said. “Like a double helix.”This logic, Zhang argued, fundamentally changes how data centers should be designed.
Today, data centers are often treated like commercial real estate: locate near cities, secure permits, connect to the grid.
In the AI era, Zhang said, the logic should be reversed.“Energy should come first,” he said. “Compute should go where energy is cheapest, cleanest and most abundant.”
This could mean building major AI centers in deserts, windy plains or remote regions rich in renewable resources, rather than in urban tech hubs.
It also means that future leaders in AI infrastructure may not be traditional cloud companies, but energy-technology firms that can integrate generation, storage, forecasting and computing into a single system.
A Civilization Driven by Energy
As the conversation turned philosophical, Zhang said that if life is an energy-organizing system, and civilization is a higher-order version of life, then AI represents a new evolutionary phase.
“Perhaps artificial intelligence is energy that has found self-awareness,” he said.
Carbon-based life, he suggested, may be only one vessel for intelligence. Silicon-based systems could become another — potentially far more scalable.
In that future, energy is not just a resource. It is the substrate of consciousness, computation and civilization itself.
Whether or not one accepts this view literally, Zhang’s core message is clear: the limits of AI will not be set by code, but by power.
And the societies that master the next generation of energy systems — sustainable, integrated and intelligently managed — will shape not only the future of AI, but the future of civilization itself.
发布于：北京市壹配资网门户提示：文章来自网络，不代表本站观点。
