Two news items in the last 24 hours underscore the interconnectedness of energy and artificial intelligence (AI). We can’t have one without the other—and we need both.
AI is an unstoppable force. Writing in Axios this morning, the buzzy Mike Allen observes of the billions pouring into AI, “This kind of investment tidal wave has come once every 15 years or so in Silicon Valley since the arrival of personal computing around 1980. These waves have their own booms and busts. But none of them has ever been stymied.”
That’s surely correct. AI is coming. In fact, it’s already here; it’s just coming bigger and bigger and bigger.
Yet there’s a complicating factor: energy. However, it’s a complicating factor that will, in fact, soon enough be simplified.
How so: The power of AI is going to force a solution to AI’s energy needs. And the solution will not be turning off the machine.
AI is going to need a lot of new energy. Rep. Cathy McMorris Rodgers, chair of the House Energy and Commerce Committee, outlined the dimensions of this need at a Congressional hearing yesterday:
In states across the country, utility planners and regulators are confronting the hard truth that they need more reliable power to meet the needs of their communities and the growing demands from our digital economy. Some are projecting a ten-fold increase in the growth rate of new power demand, compared with the past decade.
Just across the river in Northern Virginia, power demand is projected to increase from 2,500 megawatts in 2020 to over 8,000 megawatts by 2028.
In Georgia, utility companies had to quickly update their plans to reflect a jump from 400 megawatts of future demand to 6,600 megawatts.
To put that in perspective, Georgia would need about five more new Vogtle nuclear power plants to meet that level of demand.
Driving this demand, in many cases, are the industries that process digital information, the data centers that process cloud services, AI, and the digital transactions that are increasingly essential to modern life.
These services are critical to advancing our nation’s prosperity and will need more, not less, reliable baseload power—the kind of power that can be generated 24 hours a day, seven days a week, 365 days a year.”
So there you have it: AI needs lots of energy. McMorris added, it’s this tandem, AI and abundant energy, that we need to compete with China.
Where will it get this energy? Most likely, from the “all of the above” approach favored, loudly, by most Republican, and favored, more quietly, by many Democrats. That is, energy from everything, from solar to nuclear.
Yet predominantly, we’re going to be burning a lot of carbon fuels. It’s where the energy is most easily found and distributed, because the infrastructure is in place; it’s been powering the world’s industrial economy for the past two centuries.
But are we running out of what are often called fossil fuels? No. Decidedly not. In the recent words of mining engineer John Lee Pettimore,
In the U.S., the number of technically recoverable oil resources was estimated to be around 143.5 billion barrels of oil at the end of 1990, according to the EIA. That number has more than doubled despite rising production. So are we running out of oil? No we are not.
And he sources his assertion to the U.S. Department of Energy's Energy Information Administration, which tells us that as of 2020, oil resources in the U.S. totaled 373.1 billion barrels. By the way, if we were to assume an oil price of $80 a barrel, these are assets totaling nearly $30 trillion. (Other estimates of oil resources, such as from the Institute for Energy Research, are much higher.)
We can immediately note that this is just oil. Natural gas, coal, and other carbon products are in addition.
We might further note Pettimore undercuts the idea that carbon fuels are fossil fuels, which helps explain why we keep finding more carbon fuel. He cites the work of Robin Menotti, who is among the many (going back, in fact, to the 16th century metallurgist Agricola) who argue that the earth is, in fact, making more carbon energy. That is, through volcanic action, originating in the earth’s molten core. So carbon fuels are not fossil fuels at all—they could be called, in a way, geothermal fuels. But why not keep it simple and just call them carbon fuels?
Pettimore writes, “The truth is oil is actually the second most prevalent liquid on earth next to water, and regenerates within the earth faster than it can be depleted.”
Okay, so even if we have a never-ending supply of carbon fuels, what about carbon dioxide in the atmosphere? What about climate change?
Well, there’s an answer to that: If there’s too much CO2 in the atmosphere, take it out. The science of carbon capture is zooming, although the pioneering work was accomplished eons ago, by the humble tree.
But every day bring news items of new processes for carbon capture, often involving cheap and abundance raw materials. So it’s easy to see a national and international strategy for reducing CO2, as I have been writing about since 2017. And we can point to myriad other ideas in carbon capture; for instance, Aramco is finding new ways to capture capture in cement. In fact, it's obvious that buildings and construction materials could serve as permanent carbon sinks.
This is the Grand Carbon BargAIn, synthesizing energy production, consumption, and carbon capture into a pleasing green circularity. And it means we'll have all the energy we need for AI.
No comments:
Post a Comment