Three Interconnected Developments in the Political Economy of Artificial Intelligence, August 17–19, 2026
The developments of August 17, 18, and 19, 2026 cannot be understood as three separate and unrelated news stories. Taken together, they offer a clear picture of a new stage in the political economy of artificial intelligence: a stage in which the production of knowledge, data, software, and culture becomes increasingly socialized, while the material and computational infrastructure required to convert this social knowledge into economic and political power remains the property of a small number of large corporations. Under these conditions, artificial intelligence is no longer merely a technology or a collection of computational models; it has become the point at which chips, energy, land, power grids, data centers, financial capital, and political power converge.
Every artificial-intelligence model is the product of an extensive range of social activities. The data used to train these models is formed from writing, images, research, languages, experiences, user behavior, and humanity’s accumulated historical knowledge. Universities, researchers, programmers, writers, artists, social-media users, and millions of direct and indirect workers all contribute to the creation of this capacity. Alongside this flow of knowledge and social activity is a material and infrastructural flow: chips, electricity, water, land, transmission networks, data centers, and financial capital. These two flows come together in AI models and computational capacity; yet the income, ownership, decision-making power, and financial rights resulting from this convergence remain largely in the hands of private corporations and financial institutions.
August 17, 2026: The Integration of Chips, Capital, and Energy
On August 17, a collaboration among NVIDIA, SB Energy, and OpenAI was announced to create one of the largest artificial-intelligence infrastructure complexes in the state of Ohio. Under the agreement, SB Energy will provide the necessary land, energy, and buildings at the PORTS-Pike complex in Pike County; OpenAI will be the principal user and tenant of the capacity; and NVIDIA will be the exclusive provider of the AI computing infrastructure.
The complex is designed for up to eight gigawatts of computing capacity, and its first phase—approximately 800 megawatts—is expected to come online in 2028. NVIDIA also announced that it would invest $1.5 billion in SB Energy and provide up to $105 billion in credit support to guarantee portions of the commitments related to land, electricity, buildings, and long-term leasing. According to the announced plan, supplying electricity to the complex may require the creation of roughly ten gigawatts of new generation capacity and nearly $4.2 billion in investment in the regional power grid. NVIDIA’s official announcement, Reuters report
The significance of this agreement lies not only in the scale of its figures. Here, chips, capital, land, energy, and the market for computing are joined within a single structure. The company that manufactures the chips also becomes a financial guarantor of the project; the company that builds the data center retains ownership of the land and infrastructure; and the company that develops the AI models becomes dependent on this capacity through a long-term lease. Computational capacity thus becomes not only a means of production, but also a financial asset, collateral, a source of future revenue, and a lever of power.
This structure demonstrates that in the AI economy, ownership of the means of production no longer refers only to ownership of factories or traditional machinery. Ownership of land, energy, chips, networks, and computational capacity can determine who is able to train large models, who can enter the market, and who can generate revenue from socially produced knowledge and data. Even if models and software become more open, access to the infrastructure required to use them can remain under the control of a small number of corporations.
August 18, 2026: Under Public Pressure, Society Entered the Equation
One day later, on August 18, another development occurred in Pennsylvania. Governor Josh Shapiro issued a new executive order concerning data-center development. This policy change did not occur in a vacuum. The rapid expansion of data-center projects had encountered growing protests and pressure from residents, environmental advocates, and local communities. Their concerns were not limited to electricity consumption; the large-scale use of water, rising household energy costs, air and water pollution, strain on the power grid, changes in land use, noise, and the secrecy surrounding agreements between the government and developers were all central issues in these protests.
Shapiro, who had previously supported the extensive development of data centers and investment by technology companies in Pennsylvania, was compelled, under public pressure and in the face of environmental and social protests, to impose stricter regulations. When announcing the order, he acknowledged that over the preceding year he had heard directly from Pennsylvanians concerned about the impact of data centers on the environment, utility bills, and the lives of local communities. Official statement from the Pennsylvania governor’s office, Executive Order 2026-05
Under the order, data-center projects must commit to environmental standards, transparency, community participation, and consumer protection. Developers must pay the full cost of generating, transmitting, and distributing the electricity they require so that these costs are not shifted onto households and local businesses. Before receiving state permits, projects must obtain the approval of local authorities and provide clearer information about their consumption of electricity, gas, and water, as well as the measures they will take to prevent pollution.
All data-center projects were also removed from the state’s Fast Track permitting program, and state agencies were prohibited from signing nondisclosure agreements, or NDAs, with project developers. This decision is highly significant because confidential agreements can keep the people of a region uninformed until the final stages about the nature, scale, and consequences of a project that directly affects their lives.
This development shows that society does not enter the equation of technological development automatically. Society emerges as an effective force only when people organize, protest, demand transparency, and compel political power to answer to them. If political power had been left to itself, it would have followed capital and wealth, not the needs of the people or the task of addressing social consequences. The Pennsylvania governor’s change of position was not the spontaneous result of government action; it was the product of social pressure and the rising political cost of continuing the previous policy.
This experience demonstrates that policymaking on artificial intelligence and data centers cannot be confined to negotiations between governments and large corporations. The people who bear the economic and environmental costs of these projects must have a direct role in decisions concerning their location, scale, resource consumption, environmental effects, and social benefits. Technology can serve society only when society participates meaningfully in determining the direction of its development and how it is used.
August 19, 2026: The Material Costs of the AI Economy Became More Visible
On August 19, new reports on the expansion of data centers and their supply chains showed how the growth of artificial intelligence has increased demand for electricity, industrial equipment, generators, cooling systems, cables, steel, and grid infrastructure. This development challenges the common image of AI as a purely digital and immaterial phenomenon. Behind every model and every AI-generated response stands an enormous array of physical infrastructure, natural resources, and human labor.
Data-center industry studies show that pressure on power grids has forced developers to move projects farther away from dense urban centers and saturated networks. In some regions, the distance between data centers and grid connection points has increased sharply, while the share of greenfield projects—facilities built from the ground up on previously undeveloped land—has risen from about eight percent in past years to nearly 39 percent. This shift entails expanded transmission lines, changes in land use, and the transfer of part of the cost of AI development to new communities and environments.
At the same time, factories and industrial-equipment manufacturers are profiting from this wave of demand. Generac has invested approximately $250 million to expand its production of industrial and commercial generators. Its backlog of data-center-related orders has reached about $1.6 billion, and the company plans to hire nearly one thousand additional workers. Manufacturers of cooling systems, cables, prefabricated walls, construction machinery, and industrial components are likewise experiencing increased orders. Reuters report on the data-center manufacturing supply chain
This expansion of industrial production can create employment and income; however, the issue is not merely the number of jobs created. The central questions are who decides the direction of these investments, who owns the infrastructure, who receives the profits, and who pays the social and environmental costs. If the costs of expanding the power grid, consuming water, producing pollution, changing land use, and raising energy prices are shifted onto society while income and ownership rights remain in private hands, then we face a pattern in which costs are socialized and benefits are privatized.
From Social Production to the Concentration of Power
A common contradiction can be seen in all three developments. Knowledge, data, language, software, and culture are produced socially. Material infrastructure, too, cannot be created without the collective labor of engineers, construction workers, power-grid employees, equipment manufacturers, universities, and public investment. Yet the result of this social process—through the private ownership of data centers, chips, networks, contracts, and financial rights—leads to the concentration of economic and political power.
Companies that control computational capacity own more than a collection of servers. They can determine who gains access to advanced artificial intelligence, which research receives funding, which models are developed, which data acquires economic value, and which needs receive priority. For this reason, computational capacity is gradually becoming one of the fundamental forms of power in contemporary society.
This trend also shows that making a model open or publishing its weights does not, by itself, amount to the genuine socialization of artificial intelligence. If running the model still requires scarce chips, abundant energy, enormous data centers, and billions of dollars in capital, the ability to use it effectively remains with the actors who control this infrastructure. Under these conditions, knowledge may appear to become more open, while the means of converting it into economic power remain private and concentrated.
Conclusion: The Future of Artificial Intelligence Is a Social Choice
The developments of August 17–19 show that the central issue of artificial intelligence is no longer merely the speed of models or the processing power of chips. The question is what kind of economic and social order will take shape around this technology. Will socially produced knowledge and public resources strengthen monopolies and concentrated power, or will society be able to participate in determining the development, ownership, oversight, and distribution of the benefits of this technology?
The Pennsylvania experience shows that, in the absence of social pressure, economic and political power does not naturally place people’s needs first. Society’s entry into the equation requires awareness, organization, transparency, protest, and direct participation. The future of artificial intelligence is not predetermined; it will be shaped through the struggle between social production and private ownership, between public needs and the accumulation of wealth, and between social participation and the concentration of power.
Leave a comment