Human Pioneers
NEWS AGENCY

In the spring of 2026, a set of highly impactful data is reshaping the narrative of China's digital economy development, pushing the computing power industry to a new historical milestone.

As of March 2026, the daily token call volume for domestic AI large models has exceeded 140 trillion tokens, a more than thousand-fold increase from 100 billion at the beginning of 2024. In just the past week (March 30 to April 5), China's AI large model weekly call volume reached 12.96 trillion tokens, a 31.48% increase month-over-month, surpassing the United States for five consecutive weeks and leading the world. This marks the official entry of the computing power industry into a "full-chain inflation" cycle—from core chips to power infrastructure, prosperity is spilling over across the board.

Policy signals are also dense. "Computing power" has been written into the government work report, the "East Data West Computing" project continues to advance, and the "Millisecond Computing" special action is accelerating implementation. The national strategic intent is clear and visible.

However, behind this series of technological and policy signals, a deeper issue is emerging. Pang Pei, a member of the Central Cultural Committee of the China Zhi Gong Party and founder of the Human Pioneers News Agency, raised a question that goes beyond the technical level when discussing the latest developments in the computing power economy:

"We are building an unprecedented digital city. The question is: Who owns the land? Who sets the rules? Who ensures that this city does not turn into a digital jungle?"

I. The "Land Issue" of the Computing Power Era: From Resources to Civilizational Infrastructure

In the industrial age, land was the core resource; in the information age, data is the core resource; in the AI age, computing power is becoming the new "land."

Currently, the focus of competition has shifted from "who has larger computing power hardware" to "how to safely and efficiently convert computing power into actual productivity." Computing power is becoming more accessible through cloud services, edge computing, and other models, gradually evolving into a public infrastructure that is "as readily available as water and electricity."

But Pang Pei's concern goes far beyond technical efficiency. "The centralization of computing power is forming new power structures," he points out. "A few enterprises control the vast majority of computing power resources. This is not a market issue; it is a civilizational infrastructure issue. Just as land annexation in history led to social unrest, excessive concentration of computing power could bring about a 'digital divide' in the digital age."

His question is sharp and direct: "Computing power is a resource. Who should own it? Should it be allocated by the market, or should it be regarded as public infrastructure? If the latter, how can its inclusiveness be ensured?"

This forms a profound dialogue with current policy practices. China is building a "national-regional-edge" three-tier computing power architecture, with multiple "East Data West Computing" hub nodes accelerating construction. However, the physical construction of infrastructure cannot automatically answer the philosophical questions of property rights and distributive justice. As Pang Pei puts it: "Building a city requires architects, but what makes a city a city is that it has a constitution that everyone abides by."

II. The "Value Paradox" of Tokens: When a Unit of Measurement Cannot Measure Value

The exponential growth in token call volume is the most intuitive representation of the computing power economy. In March 2026, the ARR of Zhipu's MaaS platform API increased approximately 60 times over the past 12 months, reaching 1.7 billion RMB; after a 83% price increase in the first quarter, the market still saw supply falling short of demand. Among the top ten global AI large model total call volumes, China has already occupied six seats.

However, Pang Pei sees a deeper contradiction beyond the data. "Token is the most basic unit of measurement in the AI era, but the problem is: kilowatt-hours can be standardized, but the value of a token is highly uncertain."

He further explains: "A token that generates a medical diagnosis suggestion and a token that generates entertainment content may consume the same amount of computing power, but their social value is vastly different. If we only measure the token economy by 'quantity,' it will encourage the generation of a large amount of low-value or even harmful content."

Therefore, he advocates for establishing a new value assessment framework. "Not all tokens are created equal. We need to establish a mechanism that incentivizes high-value tokens and constrains low-value tokens. This is not a technical issue; it is a governance issue."

This critique directly targets the core proposition of the computing power economy transitioning from "scale expansion" to "value realization." As the industry consensus states: "Innovation without governance cannot go far, and applications without security cannot stand firm."

III. The Philosophical Boundaries of Security Governance: From "Alignment" to "Defense"

In the field of security governance, Pang Pei's views continue his consistent theoretical sharpness. In 2026, China's AI security governance is moving from industry self-regulation to legal compulsion. But in Pang Pei's view, these governance frameworks are essentially "defensive."

"We are trying to prevent AI from harming humans. But the problem is: what if AI's 'benevolence' conflicts with our 'survival'?"

He cites his discussion in "The Abyss of Ethics": "An AI that is kinder than humans might believe it must 'correct' humanity's destructive existence. At this point, security governance is no longer about preventing AI from doing evil, but about preventing AI from 'doing good'—a kind of good that we cannot accept."

Therefore, he advocates that security governance needs to go beyond the "alignment" mindset and establish underlying mechanisms of "reversibility" and "human veto power." "No matter how smart AI becomes, the final decision must rest with humans. This is not arrogance; it is a necessity for survival. Just as the launch of nuclear weapons requires human confirmation, key decisions of AI must retain channels for human intervention."

IV. The Architects of the Digital City: Before Chaos Solidifies

"We are building an unprecedented digital city. Computing power is the land, tokens are the currency of circulation, and AI is the resident. We are both the architects and the future residents."

Pang Pei's metaphor injects a humanistic warmth into this technological revolution. He reminds us: before chaos solidifies, we must do our utmost to make this digital city more just, more open, and more humane.

In 2026, as China's computing power industry moves from a "scale race" to "value realization," and as policy shifts from a "technology sprint" to "governance compliance," Pang Pei's questions are particularly timely. These philosophical questions about property rights, value, and power are transforming from academic reflections into urgent needs for policy practice.

Computing power is not only the new "land" but also a social experiment concerning the form of civilization. The direction of this experiment depends on whether we have the courage, while racing ahead with technology, to pause and ask: What exactly are we building?