AI or AC?

This image is a free use photo by antonioespana.

 

AI is on everyone’s minds and lips these days.  Wall Street sees it as the only thing that will grow our economy and national output.  Every CEO must mention AI a dozen or more times in his earnings call to not be considered backward and out of step.  I discussed AI and commercial real estate last year in a three-part series AI-Past and Present, AI & CRE, the Future is Now, AI Needs New Data Centers.

 

What are the downsides to AI?

 

But there may be a downside or two to AI.  We have all heard the congressional interest in addressing personal privacy issues due to the vacuum-like data collection of your personal data from social media sites.  Recently at the Met Gala, there was a beautiful photo of Katy Perry on the red carpet, but – problem – she wasn’t there; it was a deepfake.  She was actually recording a new album and the pic was total AI. You may have heard that AI can even capture your voice off any phone and internet recordings, then evil people can duplicate it, for example by sending your phony voice to a relative, saying perhaps, you are a hostage and to send money right away.  (Suggest you have a family code word for authentication).

But did you know that AI has an impact on the environment, too?  AI lives in newly built datacenters.  The old ones can’t support AI because they are becoming obsolete.  The new AI chips are much larger, far more complex and they use more than 5 times the power of an old chip which means we must build many, many new, expensive datacenters all over the U.S. to support AI. There are at least 35 datacenters now in S.A. with several under construction that will be AI supportive, 151 in Dallas with 16 AI under construction, and over 250 total now in northern Virginia, the largest cluster in the world.  These all require massive amounts of new electrical capacity and energy producers must seriously add to their output volumes by building substantially more power stations.

 

What can support our increasing energy demands?

 

The increase in electric demand to support AI is dramatic. Wells Fargo says that U.S. total power demand is now about 4,000 tera watt hours (TWh).  All AI currently uses about 8 TWh but they project that demand will accelerate to 652 TWh by 2030, which is 16% of all electricity the U.S. now uses. It is estimated that AI alone could use as much energy as all Sweden by 2027.  When we combine AI growth with normal population growth, increased industrialization, more electric cars and rapid electrification of the developing world, then, “We’re gonna need a bigger power plant.” Lots of them!

These new power stations will probably be powered by liquid natural gas or LNG, which is largest source of electricity generation in the U.S. BP, a major oil producer, said that we will need every bit of renewables and natural gas we possibly can in order to meet our near future energy needs.  Investors are even lifting the stock values of utilities and natural-gas pipeline operators on the expectation of higher demand for electricity to supply AI datacenters.

 

Are Renewable Energy Solutions the answer?

 

Renewables alone can’t make up the demand for additional electricity in the years ahead.  In Texas we saw how wind turbines and solar panels can be unreliable in a crisis, like Winter Storm Uri, aka snowmageddon. But even in normal times turbines only produce when the wind is over 9 mph, and must be shut down when the wind velocity is over 50 mph.  They also kill hundreds of thousands of birds each year, use a lot of oil for lubrication, and pollute the land and the oceans.  Tripling and quadrupling the number of windmills will not be the solution.

Solar panels only generate electricity for 5 to 6 hours a day, on sunny days, and have a plethora of negative environmental impacts in production of the panels and then on the disposal side, too.  Wind and solar require massive batteries to be effective and these batteries have not really been perfected yet and are only now being put into the massive use required to run the electric grid.  And then, these batteries require massive mining operations that pollute the environment and atmosphere.  So, the renewables are actually the messy unreliables and not an answer for a safe and clean energy expansion into the future.

 

Coal or Natural Gas?

 

We can agree that coal is not the answer either, so this brings us back to liquid natural gas. Unfortunately, the current administration is very negative toward fossil fuels and are committed to a transition from them to clean energy, so they oppose expanding our electricity production by LNG powered plants.  They are blocking LNG to other countries, too, through a “pause” on exports.  These developing countries are consequently forced to then rely on dirty production from coal plants to meet their growing energy needs.  This is especially true for China and India.

However, the FEDS are committed to this transition to counter Global Warming and yet it is difficult to see how solar and wind can reliably supply our energy needs for the future.   Nothing is going to stop the AI boom, but if we don’t amp up our production of electricity, will we face the battle of the Chatbot VS. Air-conditioning in the future?  No one wants that!

Next month we will look at our local provider, CPS, and how they are addressing the energy expansion issue.