The artificial intelligence effect is no longer restricted to the displays, codes or VC pitch decks. It has now become a part of the physical geography of the United States. In the American heartland, covering broad swaths of farmland, the dead industrial plants and dusty industrial belts are turning into sprawling data center campuses, which in size, energy consumption and economic impact over the long term rival cities. This transition is the turning point in the age of AI, the era of digital ambition by corporeal constraints, the age of debt, electricity, and land as strategic as code.
The West Texas land has its own story. The dust there is laden with iron, and orange-red, and blown up by the wind and so resting on all it touches. It covers the skin, permeates the air and makes itself felt with every breath. This is the location of the OpenAI chief executive Sam Altman, who is constructing Stargate, an extensive system of hyperscale data centers that are planned to run the next stage of artificial intelligence. This place might look dreadful, yet it is perfection to the AI constructors. The land is cheap, the local governments are accommodative and huge infrastructural developments are not subjected to a lot of regulatory road blocks as on the coasts.
Thousands of vehicles are transported to the Stargate construction site every morning. The employee count is bigger than the overall international workforce of OpenAI. The campus is so expansive that it is not necessarily a facility, but more of a mini-cities in the making. The rain will cause the ground to become very thick mud only to be cracked once again by the sun. The colors of the sky are dark reds and dark oranges at dusk, and this serves as a reminder of how harsh environments can also give rise to a moment of dark beauty. These are the extremes, physical and financial, which are now included in the cost of building artificial intelligence scale.
This is what it has got to bring AI, Altman said at a visit to the site in September. There exists so much infrastructure that is needed as compared to before technological revolution or the internet as we know it. And this is part of it, a little sample.
Anything but modest is that small sample. Truly, each Stargate location is approximated to cost approximately 50 billion dollars and the more radical goals of OpenAI may extend to 850 billion dollars. Such a number in itself is nearly half of the world AI infrastructure investment that is currently estimated by the large financial institutions. What would have been thought as an extreme estimate is now being used as a baseline assumption on the technology and energy market.
There is already one data center in the Abilene campus, with the second one on the verge of completion. OpenAI has a finance leadership that believes that the site might someday surpass one gigawatt of power capability. That electricity would be sufficient to serve approximately 750,000 households, which is similar to the residential power consumption of cities such as Seattle and San Francisco. This magnitude of consumption makes AI facilities one of the biggest single consumers of power within the country that compels the utilities, regulators and policymakers to reconsider grid expansion schedules that were not originally built to support this type of load.
The shovels we are putting in the ground here today, they are actually about compute that goes online in 2026, said the OpenAI chief financial officer Sarah Friar. The initial Nvidia push will be that of Vera Rubins, which are the new frontier accelerator chips. However, then it comes to what is being built on to 27, 28, and 29. The current state is a huge wring of the crunch of computing.
In the background, the financial framework behind this growth is also attracting increased criticism. The current surge in AI infrastructure is being powered by unprecedented amounts of debt as compared to the previous technology booms which depended on significant amounts of equity funding. Bonds and credit default swap spreads by data center developers and other companies have started to expand, and this is an indication of increased concern in credit markets. Other commentators observe that some recent developments mirrored unpleasant echoes of the fiber-optic boom of the early 2000s, when overbuilding and excessive leverage left permanent scars.
At the centre of this money web is OpenAI which has alliances constituting a complex, interdependent ecosystem. Vendors of the chips include Nvidia, cloud and infrastructure services are offered by Oracle and Amazon Web Services, hardware options are offered by AMD and Broadcom, and capital and strategic support are offered by SoftBank. The outcome is what other observers call a circular AI economy, in which the success of every participant is conditional on the rest of them remaining willing to spend, build and grow. The system is valid as long as demand continues to increase. It is the worry as to what will occur in case of a weak link.
We are expanding quicker than any other business I ever heard of before, said Altman. And now we would be much bigger with much more capacity.
That saying is the boldness and the exposed heart of the time. The AI services are still in demand as the enterprise adoption, consumer tools, and government interest keep rising. However, software innovation is no longer the limitation to growth. It is restricted with transformers, transmission lines, water access and availability of land zoned to be used in heavy industrial processes. Electricity that was a back-stage concern in both the boardrooms and statehouses has now taken a front seat issue.
Other technological giants are taking the same routes. Google, Microsoft, Amazon, Elon Musk through his xAI as well as Meta are all installing huge data center campuses in the Midwest and the South. Communities in the rural areas are all at once striking agreements that offer employment and tax but cast doubt on water consumption, grid capacity and sustainability. The short-term economic stimulus is factual, but so are the trade-offs, particularly where one facility is capable of consuming more electricity than a previous county.
The cultural shift is also in progress. Technological progress over the decades has been experienced as one-sided, with it being mostly a software upgrade and cloud abstraction game. Today’s AI boom is different. It is noisy, dirty, and hungry in energy. It transforms skylines and landscapes leaving concrete, steel and transmission towers as their enduring signature in the physical manifestation of a digital age.
