Artificial intelligence has been formally included in the national strategic assets.
But in all public discussions, a more fundamental and fatal constraint has long been underestimated—the electricity and energy system itself.
Energy is no longer just an industrial support, nor is it just a transformation issue.
In the AI era, it is becoming an implicit boundary of national power that influences technological competition, military capabilities, industrial resilience, and institutional efficiency.
Over the past decade, the innovation narratives of the United States and other major global economies have focused on chips, algorithms, and capital market allocations.
But a reality that has been repeatedly ignored is emerging:
(Image caption) The super-large data center and its surrounding power transmission and transformation facilities are becoming the most real energy pressure scene in the AI era—computing power can expand rapidly, but electricity and power grids cannot be replicated in sync, forming an underestimated bottleneck of national strength.
Computing power can expand exponentially.
However, the power system cannot be replicated at the same speed.
From large-scale model training and dense deployment of data centers to the real-time computing needs of defense, intelligence, and critical infrastructure,
Electricity has shifted from a "background assumption" to a "hard bottleneck."
This is not a problem of a single company, nor is it a short-term supply and demand fluctuation.
It is a structural risk that has not yet been fully priced in by the market or fully absorbed by policy.
GFM's focus on AI energy issues is not based on technological optimism or industry promotion.
Rather, it is based on a more serious judgment:
If the energy system cannot support the actual operational needs of AI,
Therefore, any discussion about computing power, sovereignty, and technological leadership is incomplete.
Why should we start by discussing "Why is the world stuck on energy?"
Before delving into AI energy, nuclear energy resurgence, and engineering solutions,
We must first answer a more fundamental, and more uncomfortable, question:
(Image caption) The real-world scenario of fossil fuels, renewable energy, and the power grid coexisting reveals that energy transition is not about "knowing whether a transition is necessary," but rather about being prepared to bear the real costs of institutional changes, costs, and infrastructure capacity.
The world is not unaware that it needs to transform.
Rather, they are not yet prepared to bear the institutional, cost, and responsibility associated with the transformation.
The growth in renewable energy generation cannot mask the stagnation of the primary energy structure;
Progress in installed capacity data cannot fill the gaps in grid, energy storage, and baseload capacity.
If we don't first understand this reality,
AI energy will only be misinterpreted as an "industry bottleneck".
It's not a matter of carrying capacity at a civilizational level.
Therefore, this special issue is based on a flagship research paper.
Why is the world stuck on energy?
This serves as the structural starting point for the entire discussion on AI energy. A structural analysis of the AI energy issue will unfold sequentially:
First, it reveals how power shortages have become a hidden constraint on U.S. national security and technological competition;
Next, we will focus on the "AI power crisis" behind core technology companies such as Nvidia, which has come into the view of decision-makers.
Finally, returning to the engineering site, through the practical experience of Mr. Cheng Maiyue's team, an expert in power, we will explore who truly possesses the expertise in engineering...
The ability to solve this problem at the institutional and capital levels.
⸻
The three core questions of this topic
The GFM "AI Energy" special report will unfold gradually across three levels:
First layer: Structural reality. Why does global energy still heavily rely on fossil fuels under the narrative of energy transformation?
Why have power grids, energy storage, and regulations become the real bottlenecks?
Second layer: AI stress testing
How can AI and data centers bring forward the energy problem by a decade?
Why has this pressure come to the attention of US policymakers?
The third layer: Engineering and institutional solutions. Why has nuclear energy returned to the forefront?
Who has the ability to "outrun time" in engineering, institutional and capital allocation?
⸻
GFM's position
GFM has never been concerned with just the technology itself.
Rather, it is the institutional capacity behind the technology.
In the AI era, energy not only determines efficiency,
It further determines the boundaries.
Who can establish a feasible structure between project implementation, institutional coordination, and capital allocation?
Whoever does so may reshape the map of national power in the next stage.
This is precisely the core question that the "AI Energy" special feature attempts to answer.