Congressional Testimony On Artificial Intelligence Data Center Energy Needs
Ensuring AI & Power Needs Serve the Public Interest
By Tyson Slocum
Today Tyson Slocum testifies before the Subcommittee on Economic Growth, Energy Policy & Regulatory Affairs of the U.S. House Committee on Oversight. The title of today’s hearing is America’s AI Moonshot: The Economics of AI, Data Centers, and Power Consumption. My testimony has the following highlights:
- In recent years, American companies established dominance in artificial intelligence development, model training, and deployment; coupled with hosting the required domestic information technology infrastructure — including data centers — to support large language models (LLMs) and cloud computing.
- Projections of increased power demand for AI are wildly scattered, complicating accurate projections of needed power supply. As with any technological advance, the pace of innovation is staggering, with disruptive breakthroughs altering fundamental assumptions and projections of the microprocessing and power consumption requirements to sustain continued AI dominance. The current expectations for power consumption to meet AI’s software and hardware needs may be fundamentally different 18 months from now.
- Making long term, capital-intensive power infrastructure investments based upon tech industry consultants’ short-term guesstimates of data center demand threatens consumers financially — particularly if President Trump passes these costs on to consumers and taxpayers under emergency authorities. A crucial shortcoming of President Trump’s “energy abundance” narrative is the failure to include energy efficiency and demand management, levers that — when utilized — can save consumers money and reduce the risk of massive overbuilding and stranded assets.
- Using federal emergency or other expedited authorities to overrule state and local ordinances and approve power infrastructure to support AI is inconsistent with the public interest and could backfire, resulting in expensive stranded assets that financially burden ratepayers and taxpayers and harm local communities, especially if emergency presidential powers are hastily invoked.
- Reviving coal to power data centers is nonsensical, polluting, and unrealistic: no one will build a new coal power plant today because they are extremely expensive relative to other options, including solar/ battery systems and gas-fired power plants.
- Pushing natural gas as a primary fuel choice conflicts with President Trump’s ambition to dramatically increase exports of liquified natural gas (LNG). Combined, rising demand from these two sectors will lead to domestic supply shortages and higher domestic price volatility. New nuclear capacity remains largely theoretical and absurdly expensive. Renewables and battery storage can scale and deploy quickly to meet demand.
- While FERC recently has taken steps to ensure consumers are protected from data center buildout, more certainty is needed to ensure ratepayers and reliability are prioritized.
- President Trump’s chaotic use of tariffs is not only inviting an economic recession; his haphazard trade policies are hindering access to the supply chains needed to build out AI infrastructure. Trump’s counterproductive policy is unleashing incompetence, not dominance.
- Corporate consolidation and control of generative artificial intelligence technologies, and the lack of adequate federal guardrails and protections, leaves society vulnerable to significant abuses and unneeded disruptions stemming from AI.
Access the full testimony here tysontestimony0331