تولید اجتماعی، مالکیت خصوصی و تمرکز قدرت

فارسی

English

From the Socialization of Production to the Privatization of Infrastructure, Models, and Power

Developments on August 7 and 8 make a qualitative shift increasingly clear: artificial intelligence is moving from a primarily software-based industry toward one dependent on enormous capital investment, data centers, energy, networks, and model ownership. Participation in production may therefore grow broader every day, while entry into the layer where ownership and economic power reside becomes more difficult and restricted.

1. The “AI Factory” Becomes an $80 Billion Asset

The latest example is Switch, a data-center operator that provides the power, cooling, and connectivity required for massive GPU clusters. An August 7 report says that the company has confidentially filed for an initial public offering and that its potential valuation, including debt, could reach approximately $80 billion. The offering could raise as much as $10 billion.

Connection to Social Order in the Age of Consciousness: This development strengthens a hypothesis advanced in previous reports: the data center is taking the place of the industrial factory as a key means of production. Unlike the early internet, where entry costs were relatively low, ownership of these “intelligence factories” requires tens of billions of dollars. The socialization of production therefore does not necessarily lead to the socialization of ownership of the means of production.

2. Even an “Open” Model Can Appropriate Part of the Value Produced by Others

Another important development concerns Alibaba. Despite Qwen3.8-Max being an open-weight model, the company intends to demand a share of revenue from large commercial users. Moonshot’s Kimi K3 model has a similar mechanism for some large businesses.

Connection to the book: Model ownership becomes more complex here. Thousands of developers can build products on top of a model and create new value, while the model owner can claim a portion of the value produced downstream. In political-economic terms, ownership expands from the “sale of a product” into a right to receive a share of the entire productive ecosystem.

3. Millions of Developers, a Few Infrastructure Bottlenecks

Cloudflare attracted two million new developers in the second quarter alone, while growing AI demand is increasing the revenue and value of infrastructure companies. At the same time, Amazon has said that it lacks sufficient cloud capacity to meet all of this year’s demand.

The combination is highly significant: at the lower levels of the network, millions of people and companies produce; at the top, a limited number of companies own the bottlenecks in computing, networking, and cloud infrastructure.

Connection to the book: This may be one of the clearest new forms of the contradiction between social production and private ownership. The number of producers grows, dependence on infrastructure grows, the number of infrastructure owners remains limited, and the power of production bottlenecks increases.

4. Capital Concentration Has Grown So Large That AI Is Being Discussed as “Too Big to Fail”

Some U.S. Federal Reserve officials are examining the speed and scale of AI investment from the perspective of financial stability. The president of the Kansas City Federal Reserve has raised questions about the industry’s complex financing relationships and the risk of contagion, asking whether the AI sector could eventually become “too big to fail.”

Connection to the book: If private ownership of AI infrastructure becomes so concentrated that its failure threatens the entire economy, a new contradiction emerges: profits are private, while potential risk becomes social. Society may be required to guarantee the systemic stability of a structure that remains privately owned.

5. The Physical Costs of Artificial Intelligence Are Also Socialized

Research in 2026 on AI infrastructure shows that data centers are no longer an invisible digital activity. Electricity consumption, transmission grids, land, water, and cooling facilities have become essential components of AI production. A recent study estimates that the electricity consumption of six leading companies could rise from roughly 118 terawatt-hours in 2024 to between 239 and 295 terawatt-hours in 2030, with regions including Virginia facing substantial pressure on the electrical grid.

Connection to the book: The contradiction becomes more tangible here. The company owns the model and data center, but the electrical grid, natural resources, land, public infrastructure, and environmental effects are shared by society. The issue is therefore not only the privatization of profit, but also the socialization of part of the cost of production.

6. Automation: The Central Issue Is Not the Elimination of Work, but Ownership of the Time That Is Freed

Recent reports continue to show companies using AI to make teams smaller, remove layers of management, and increase the amount of work that fewer employees can perform.

For the book’s framework, however, we must avoid an oversimplification: not every layoff can be attributed directly to AI. The more fundamental issue is the expansion of productive capacity with less human labor.

Connection to the book: If society can produce the same wealth with half as many hours of human labor, who owns the half of human life that has been freed? Under existing relations, increased productivity can become lower labor costs and higher returns on capital. In a social order based on consciousness, the same technological achievement could instead reduce working hours and expand free time, education, creativity, and social participation.