New infrastructure loading-1

“Every great crisis of capital eventually finds a new space for itself.”

- David Harvey

The history of capitalism is the history of companies, markets, technologies, and reconstructed infrastructures. Every major accumulation of capital has risen upon a specific technological and institutional infrastructure that reorganizes the processes of production and circulation. The industrial capitalism of the nineteenth century, thanks to steam power and railways, not only increased production capacity but also accelerated the circulation of capital on a national scale. In the first half of the twentieth century, electrical grids and internal combustion engine technologies formed the foundation of a new regime of production. The Fordist mass production model includes assembly lines, a cheap and continuous supply of electricity, standardized production processes, highway networks, and public investments supporting national markets. In this process, the state emerged as one of the fundamental regulatory actors that built the infrastructure and made industrial capitalism possible.

Neoliberal globalization, which took shape from the 1970s onwards, relied on a different infrastructural transformation. Financial liberalization, container shipping, digital communication networks, and the internet made it possible for production to transcend national borders and reorganize within global value chains. In this period, the primary goal of economic competition was not so much to produce more, but to be able to coordinate production on a global scale. Digital networks that accelerated the flow of information were the new form of organization for economic power.

Today, this infrastructural transformation appears to have entered a new phase. The internet has not disappeared; it has transformed into an invisible infrastructure upon which a new layer of production rises. What is now decisive is the ability to process information with massive computational capacity. For this reason, data centers, advanced semiconductors, high-performance computing systems, energy grids, and the digital networks that connect them constitute the strategic infrastructure of the digital economy. In this new historical phase, in the age of generative artificial intelligence (AI), capital accumulation is increasingly being reorganized around computational capacity. If this observation is correct, reading the current competition in AI technologies merely as a technological race between software companies would be insufficient. The real struggle will take place over who controls the infrastructures, energy resources, semiconductor production, digital networks, and platforms that make computational capacity possible.

In recent years, AI has mostly been discussed through technological innovations—such as more powerful algorithms, larger language models, faster processors, and smarter software—and the socio-economic-ethical impacts and reflections they cause. Yet, this narrative, which is widely accepted by the public, focuses on the visible face of the transformation while pushing the material relations that make it possible into the background. The problem is not just the appreciation and adoption of AI, but that AI is being handled in isolation from its historical context. Evaluating a technology independently of the production relations that make it possible is an approach as incomplete as thinking of the steam engine without coal mines, the automobile without oil, or the internet without fiber optic networks. Technologies do not write or create history on their own; they gain meaning within specific economic, political, and institutional conditions. Therefore, the question of why global capital is turning to data centers, semiconductors, and computational capacity to this extent today is important.

David Harvey argues that one of the historical tendencies of capitalism is crises of over-accumulation of capital. Crises emerge when accumulated capital cannot find sufficiently profitable areas for investment; capital then tries to reproduce itself by turning to new geographies, new sectors, and new infrastructures. Harvey explains this process with the concept of a “spatial fix.” This approach argues that crises are temporarily shifted to other areas. Every major economic crisis brings with it the redirection of capital to new areas of accumulation. This concept offers an important starting point for understanding today's AI economy. When global capital flows are examined over the last decade, a remarkable trend is observed. Infrastructure funds, pension funds, sovereign wealth funds, and technology companies are not only investing in software companies. The real concentration is taking place in hyperscale data centers, power generation facilities, high-voltage transmission lines, submarine fiber optic cables, and semiconductor production capacity. The capital expenditures of Microsoft, Alphabet, Amazon, and Meta should be read in conjunction with data center construction and energy deals. Likewise, it is no coincidence that Blackstone, Brookfield, KKR, and other global infrastructure investors view data centers as the new generation of strategic assets. This shows us that in the AI age, value is concentrated not only in the companies developing the algorithm but also in the capital financing the infrastructure on which the algorithm will run.

Perhaps we should no longer ask how smart AI will be, but why the world's largest capital movements are directed toward data centers, energy infrastructures, and computational capacity. And why is capitalism reorganizing itself through AI, a general-purpose technology, and its infrastructure?

Of course, if we are talking about an infrastructure economy, we must also show the material conditions upon which it is based. AI models cannot work without electricity. Electrical grids cannot function without transmission lines. Data centers cannot be sustained without advanced cooling systems. Advanced chips, meanwhile, depend on extremely complex global supply chains. Therefore, the primary input of the AI economy is not just data; it is energy, water, copper, silicon, capital, and geopolitical stability. Data is important; however, it does not produce economic power on its own. What turns data into a strategic advantage is the computational capacity to process it and the infrastructure that supports this capacity. For this reason, computational capacity is treated as a strategic infrastructure capacity that shapes economic competition and geopolitical power, beyond being a technical performance indicator.

This approach inevitably leads us to the role of the state. Because throughout history, no major infrastructural transformation has occurred solely through market mechanisms. Railways, electrical grids, highways, the internet, and space technologies have been shaped by public investments, military research, and industrial policies. AI infrastructure is no exception to this historical line.

In the coming weeks, developments that appear independent of one another—from NATO's technology policies to the US CHIPS and Science Act, from the European Union's semiconductor strategy to China's technological rise and the data center investments of Gulf countries—will be discussed. Why are actors that appear so different from one another turning to data centers, semiconductors, and computing infrastructures at the same time? Is this really just a technological transformation, or is it a sign of a deeper restructuring in the global economy? If the strategic resource of the new age is not oil or data, but computational capacity, who will be the winners and losers of this new order? Therefore, this series of articles will attempt to understand the new infrastructure that makes AI possible and the political economy shaped by this infrastructure, rather than AI itself.