WASHINGTON, D.C., April 4, 2025 – Artificial intelligence has long ​been announced as the next technological frontier. Smart ​assistants to predictive analysis and autonomous vehicles, AI promises a large-scale transformation. But behind the elegant interfaces and bold visions is a growing economic dilemma that is beginning to shake global markets, ​energy networks and trade policies.

According ​to energy analyst Mark Mills, AI’s energy needs are sufficient to reconfigure national strategies. Speaking at the National Energy ​Analysis Centre, Mills ​offers an amazing analogy: a mid-sized ​AI data centre uses as ​much electricity per day as it takes to ​launch a rocket on the moon. ​It’s not hyperbole. It is ​a cold fact of ​physics, and it is composed of the exponential growth of AI.

Biden’s past attempts to ​regulate ​artificial intelligence have now been replaced by a deregulation position under President Trump to promote what he calls “American innovation in AI.” But with AI’s growing ​resource appetite, the conversation is no longer limited to innovation. These are ​sustainability, sovereignty and survival.

How much energy ​does ​AI actually use?

The IV can be digital, but its footprint is physical and massive. Mills paints ​a living picture: “A billion dollars invested in AI data ​centres requires ten times more electricity than the same amount spent on electric vehicles. This comparison was stunned, given the fanfare around EVs and green technology. Moreover, these data centres are not only present in technology centres. They grow ​in states like Louisiana and Pennsylvania, where companies like ​Meta build ​$10 billion in natural gas-fired gigawatt campuses.

This is not just about ​infrastructure. The energy demand generated by these data centres could exceed the growth of the public utility network for decades. ​According to Mills’ ​testimony ​at Congress, the U.S. electricity grid has not experienced this level of demand for more than ​30 years. This means that the AI revolution could revive industries like ​natural ​gas and even coal, the techniques many assumed were relics of the past.

But AI not only consumes power, it also reshapes geopolitical and economic power.

Why is China’s role so critical in the AI world?

To ​understand ​the global ramifications of AI growth, one must look at minerals. As Mills and Asia Times analysts confirm, China dominates the global capacity to refine critical minerals, ​controlling between 50 and 90 per cent of ​global production by resource. ​This dwarf even the oil production part of OPEC.

Essential ​minerals ​such as lithium, ​cobalt and rare soil elements are not only used in batteries, they ​are the DNA of AI infrastructure. Servers, chips, cooling systems and cloud networks depend on it. ​However, the US ​is ​highly dependent on imports of at least 20 critical materials. Mills points out ​that this dependency is not simply a supervision, it is a ​strategic vulnerability, especially ​in a world where trade ​tensions erupt at the ​fall of a ​hat.

The ​danger? Economic or political disagreement with China could stifle US access to these resources, mainly by curbing the entire AI industrial engine. It’s not a theoretical risk. China has already armed its rare earth exports.

Can tariffs solve the manufacturing crisis?

President Trump’s renewed tariffs are aimed at reducing manufacturing industry and reducing dependency. However, Mills warns that without the accompanying reforms – such ​as tax incentives and regulatory simplification – only tariffs could be ​immersed in a 70-type stagnation spiral: slow growth and rising prices.

Trump’s policies echo the ​strategies of the Reagan era that used tariffs ​as ​a negotiating tool while reducing taxes and deregulation industries. But the current circumstances are ​more complex. AI is not a car or ​toaster, it is ​a ​distributed environmental technology. You can’t do a chatbot. It is not easy to trace where ​a large linguistic model has been formed, especially ​when data, infrastructure and processing ​are distributed on all continents.

This ​has led some experts, ​including researchers from the Forbes and ​Brookings Institute, to raise an eyebrow: are these new tariffs designed, or at least ​modelled, by AI? The ​logic seems shredded, directed at ​both rivals and allies, without a ​coherent long-term strategy. This raises ethical and strategic ​questions: should we let Amnesty ​International report on national policy decisions such as tariffs?

Should government policy influence AI?

Although there is no strong evidence that IA ​dictates behind the scenes policies, the possibility has become less remote. AI models are already used in simulations of predictive economy, logistics and trade. But ​as Conversation points out, IV can be manipulated by partial data ​or manipulated to justify predetermined political objectives on the pretext of objectivity.

Without transparency, these tools become dangerous. ​AI has no empathy, cultural understanding, or geopolitical vision. He ​cannot weigh the ethical ​consequences, and he certainly cannot accept responsibility. If an AI-oriented policy ​causes economic damage, who is responsible, the software developer, policy maker or someone?

In addition, while IA decouples ​national borders, training in one country, deployment ​to another and value production into ​a third traditional economic lever such ​as tariffs are beginning ​to break down. You cannot tax what you cannot follow.

What is the alternative to AI age tariffs?

Instead of relying on outdated economic tools, the experts propose ​a new manual that includes:

• AI Provenance Standards: Traceability ​from data to model to decision.
• Cross-Border Licensing: Agreements for AI ​services used across jurisdictions.
• Data Integrity Regulations: Controls on how ​data is collected, ​cleaned, and used.
• Ethical AI Treaties: International consensus on responsible AI use.

Countries such as Canada are already moving in this direction with ​legislation such as the Artificial Intelligence and Data Act (IAA), to ​regulate transparency and accountability in AI. On the contrary, the United States has been reluctant, as evidenced by its refusal to sign AI’s recent global security statements and Trump’s regression from Biden’s regulations.

This hesitation could go back. As AI is involved in global ​finance, health and defence, the ​lack of ​government opens the door to manipulation, bias and even financial collapse. The 2010 “combat shock” triggered by the cotton trade showed how fast markets can accelerate when left to ​automated systems.

What is the way forward?

Mark Mills believes that the United States still has time to correct the course, but only if ​it acts quickly. It calls for a return ​to the fundamental principles of enabling private companies to lead, but removing unnecessary burdens that ​hinder progress. Destroy if necessary, but ​not blindly. Offer incentives, ​but not brochures. Ensure clarity and coherence of ​policies, ​particularly with regard to energy ​and minerals.

He stressed that the current energy infrastructure was not yet ready to manage the AI boom. Allowing bottlenecks, bureaucratic delays and regulatory uncertainties prevent the rapid deployment ​of ​new power plants. However, companies are not waiting. Projects such as the Homer City ​data centre in Pennsylvania and Meta gas construction in Louisiana show that demand already exceeds supply.

Whether it is refining capacity, access to minerals ​or electricity generation, AI’s ambitions in America require ground work – and fast work. Without this foundation, we ​run the risk of ​becoming a nation of users, not manufacturers, consumers of technology, not creators. As Mills points out, the inability to act could put the United States on a path similar ​to that of ​Germany or the United States – once economic ​giants, who are now struggling with the consequences of deindustrialisation.

The question is no longer whether ​we should accept AI. That train left the station. The real question is whether we will do the hard work we need to manage it responsibly, ​regulate it fairly and ensure ​that it strengthens rather ​than weakens our economy. This ​means that energy, trade and AI ​are interconnected and not ​silent problems.

Because at ​the heart of all this, the decisions ​we make today – on AI’s tariffs, energy and governance – will define whether we remain a world leader or lose ourselves in economic misfit.