Jakarta, INTI – The explosion of artificial intelligence (AI) has brought the United States into a new dilemma: facing an energy crisis or bearing the burden of massive infrastructure investments. Electricity demand from data centers powering AI has surged dramatically, challenging the capacity of a national power grid that has long been oriented toward serving households and conventional businesses.
The AI Data Center Demand Boom
According to a report by Reuters, most of the 13 largest electricity providers in the U.S. are now receiving power requests that far exceed their current capacities. These demands are primarily coming from tech giants building large-scale data centers to support AI processing.
In some cases, a single data center can consume more electricity than an entire city. Oncor Electric, for instance, has received requests totaling 119 gigawatts—four times its peak supply. Meanwhile, Allentown has been asked to supply 50 gigawatts, despite only having a capacity of 7.2 gigawatts.
Massive Investments, Double Risks
To meet this surge in demand, utility companies are being forced to ramp up capital spending. In fact, this year’s budget alone has doubled their five-year investment plans.
But it’s a high-stakes gamble. If the projected demand is overestimated or AI projects are canceled due to economic pressures or rising interest rates, utility providers could face significant financial losses. The consequence? Higher electricity rates for households and businesses.
On the other hand, if they hold back on investments and demand does skyrocket, the risk of a power crisis becomes very real—widespread blackouts could occur.
Project Uncertainty and Survival Strategies
Demand from tech giants is also hard to forecast, as projects are often confidential and fragmented. Many utility companies are struggling to make accurate projections because requests are submitted to multiple providers simultaneously, with unclear scope and uncertain timelines.
“We’re seeing a lot of abstract projects that no one really knows the shape of,” said Jon Gordon of Advanced Energy United.
To mitigate risk, Oncor Electric now only agrees to build new infrastructure if tech companies provide a letter of credit or upfront cash as a guarantee.
Hope from More Energy-Efficient AI Technology
Despite the complexity of the situation, there is a glimmer of hope from the technology side. A new AI model from China’s DeepSeek has shown that next-generation AI could operate with significantly lower power consumption. If this trend continues, the pressure on the power grid could ease dramatically in the coming years.
Conclusion
The United States now faces a major dilemma: meet the skyrocketing electricity demand of AI with multibillion-dollar infrastructure projects, or risk blackouts and soaring power rates. The uncertainty of AI projects and the immense cost of investment place utility providers on a razor’s edge. Amid all this, one crucial question remains: who will foot the energy bill for the AI revolution?
Read More : MediaTek Genio 720 & 520: The AI Revolution in Energy-Efficient IoT