Trump ignores biggest reasons his AI data center buildout is failing
Chip Somodevilla / Staff | Getty Images News
Nearly half of the AI data centers planned for this year face the specter of delay or total cancellation, calling the global technological arms race into question. Although the Donald Trump administration has identified the construction of artificial intelligence infrastructure as a national security priority, aggressive tariff policies toward China are proving counterproductive. Key components such as transformers, switchgear, and batteries, which Chinese manufacturers have supplied for decades, are now nearly unobtainable. Delivery lead times have surged from a pre-pandemic two years to a record five, directly impacting the schedules of major tech players. Currently, only one-third of the massive projects planned for 2026 are in the active construction phase. American industry is unable to meet the skyrocketing demand on its own, and developers feel trapped between the necessity of paying high tariffs and the risk of losing their competitive edge over Asia. For users and creators worldwide, this signifies a tangible slowdown in the deployment of new AI models and higher costs for cloud services. Without fluid supply chains, even the most ambitious political declarations remain merely theoretical, as modern data centers cannot function without the physical power infrastructure that cannot be manufactured overnight in local factories.
The race for dominance in the field of artificial intelligence has ceased to be solely the domain of brilliant programmers and algorithms, moving brutally onto the ground of heavy energy infrastructure. Donald Trump, by making the construction of AI data centers a national priority, has hit a wall that cannot be jumped over by decrees alone. Although Washington's rhetoric focuses on overtaking China, it is Beijing that holds the keys to American server rooms, controlling the supply chains of components without which modern computing clusters are merely empty halls.
The scale of the problem is staggering: nearly 50% of data center projects planned for this year in the USA face the specter of delay or total cancellation. It is not a lack of Nvidia chips or a shortage of engineers that is hampering development, but a deficit of the most basic elements of the power grid. Developers are unable to source enough transformers, switchgear, and industrial battery systems that form the backbone of any high-power-density facility.
The Tariff Paradox and Broken Supply Chains
The irony of the current situation is the fact that it is precisely Donald Trump's aggressive tariff policy, aimed at Chinese imports, that is becoming the greatest obstacle to realizing his own vision of technological power. For decades, American manufacturers have relied on China for the supply of critical electrical equipment. Today, as the demand for AI computing power grows exponentially, trade barriers are drastically extending waiting times for equipment. Before 2020, the delivery of key components took between 24 and 30 months. Currently, this time has stretched to an unimaginable five years.
Read also
This five-year lag is of critical strategic importance. It is estimated that China currently remains about five years behind the USA in the race for the most advanced AI models. However, if American investments are frozen due to a lack of transformers, this gap could be closed rapidly. Trump is pushing a model in which the United States produces the necessary equipment itself, but Bloomberg reports are merciless: domestic US production capacity is in no way keeping up with the demand generated by tech giants.
Empty Construction Sites and Investor Desperation
Analyses conducted by the firm Sightline Climate shed light on the real state of investments that were supposed to revolutionize the market in the near future. Only one-third of the largest AI data centers scheduled to launch in 2026 are currently in the active construction phase. The rest are stuck in a planning vacuum, waiting for component deliveries that physically do not exist on the American market. This creates a dangerous precedent where companies feel "paralyzed" by top-down regulations.
Desperation in the sector is so high that many players are deciding to take risky steps. Companies are willing to pay high tariffs and take on alleged national security risks just to obtain goods from China sooner. For an investor building a multi-billion dollar facility, waiting half a decade for a switchgear is a business-unacceptable scenario. They thus choose the lesser evil, which stands in direct contradiction to the narrative of total independence from the Asian rival.
Ignorance of Technical Foundations
Donald Trump, in his speeches, seems to bypass this uncomfortable foundation of the problem. By ordering tech companies to "build, bring, or buy" energy for their centers, he ignores the fact that a source of clean energy is useless if there is nothing to plug it into. The problem lies not in power generation itself, but in the transmission and transformer infrastructure that allows this power to be brought into the server room. Without switchgear and transformers, giant wind farms or modular reactors remain merely expensive monuments to ambition.
This situation exposes a deep gap between political optimism and engineering reality. Building a supply chain for heavy electrical industry is a process that takes decades, not months. The American manufacturing sector currently lacks the scale to replace China as the workshop of the world in the short time required by the generative artificial intelligence revolution. Consequently, a strategy aimed at weakening China paradoxically strengthens its role as the only supplier capable of satisfying Silicon Valley's energy hunger.
In my assessment, without a radical revision of tariff policy or massive, direct state subsidies for the production of grid components, the American AI data center boom could end in a painful landing. AI technology does not exist in a vacuum—it needs copper, silicon, and steel, which the USA is currently unable to provide for itself independently. If Washington does not understand that AI energy security starts with the transformer, not the algorithm, the advantage over China may turn out to be only a temporary anomaly in the history of technology.
