By Greg Walcher | Guest Commentary, GregWalcher.com
In planning the nation’s 1976 bicentennial celebration, Congress made one of its dumbest-ever boondoggle decisions. Recognizing the near death of railroad passenger service since the 1950’s, Congress decided to spend millions turning the aging and crumbling Union Station into the National Visitor Center. But they missed the obvious red flag – the millions of visitors to the nation’s capital during 1976 would not be coming by train.
The ugly-carpeted National Visitor Center sat mostly empty that year, after which the old depot was boarded up, its roof caving in by 1981. Still ignoring reality, Congress spent millions more on several studies of what to do with the building. Each study concluded that the highest and best use would be as a train station – what a revelation. $181 million was spent restoring it, and over $8 billion more since then. Today, the magnificent Beaux-Arts structure functions much as it did when first built by the railroads in 1907.
There is a lesson in such waste that should be atop the planning agenda at the President’s Council of Science and Technology Advisors that will oversee his artificial intelligence (AI) initiatives, along with David Sacks, the new “AI and Crypto Czar.” The President is intent on shifting the focus of federal AI policy, from the technology’s potential risks to its potential economic benefits, possibly in the trillions. He wants the U.S. to dominate the field, and has touted AI investments of $500 billion from Oracle, OpenAI, SoftBank, MGX and others.
One of the greatest obstacles to the new data centers is their massive need for electric power, in many cases beyond the capacity of the existing grid. Trump has endorsed the idea of building new power plants right next door to the data centers, even fast-tracking their permits, rather than connecting them to an already over-burdened electric grid.
So far, no one has been talking about using power generation facilities the U.S. already has, rather than building new ones. In the 1970’s Congress created Amtrak and almost authorized it to build a massive new train depot in Washington, D.C. – before some genius figured out they already had one of the most beautiful train stations in the world.
Similarly, utilities across the U.S. have been shutting down state-of-the-art power plants, and closing highly productive coal mines, for a decade or more. But most haven’t been torn down yet. So, maybe instead of the expense of building new power plants next to the new data centers, we should be unusually thoughtful about where to build the data centers.
Many utilities have committed to the goal of carbon-free electricity and to phasing out coal-fired power plants. That goal is not consistent with new Administration priorities, but there is no need to second guess corporate decisions based on their internal economics. There is, however, good reason to second guess the wisdom of shutting down electric generating capacity the country still needs.
Consider locating one of the new AI data centers in northwest Colorado, for example. Two of the state’s most efficient and productive coal mines are within a stone’s throw of two of its most efficient power plants. The Trapper and Colowyo mines together produce over 3 million tons of coal per year, their primary customers being the nearby Craig and Hayden power plants. All of that infrastructure is in the slow and agonizing process of shutting down, with an estimated $320 million economic impact that includes losing 2,800 high-paying jobs and 43 percent of local property tax revenues.
The State of Colorado, and power plant owners Tri-State and Excel, have spent millions to help the communities “transition” to some other economic base, mostly with public money. When the U.S. Department of Agriculture announced $970 million in new clean energy grants, Tri-State was first in line, and has now committed $22 million to northwest Colorado’s “transition.” The State’s Office of Just Transition has doled out millions more – none of which can offset the economic assassination of those communities.
Most of us don’t blame private companies for making business decisions. But these are not decisions based on a free market. They are driven by government policy – which is now pursuing a dire need for more electric power, not less. If leaders want to dominate the global AI industry, they ought to build data centers where that infrastructure already exists.
Or, they could study the problem for a few years, and then decide to “transition” power plants back into power plants – like turning the train station back into a train station.
READ MORE OF GREG’S COMMENTARY AT GREGWALCHER.COM
Editor’s note: Opinions expressed in commentary pieces are those of the author and do not necessarily reflect the opinions of the management of the Rocky Mountain Voice, but even so we support the constitutional right of the author to express those opinions.