Data Centers: Some Thoughts on Impacts, Benefits, and Backlash

 

       AI is capital-intensive, energy-intensive, water-intensive, emissions-intensive, replaces some humans, and has the potential to increase income inequality as small groups of executives can be the only employees of multi-billion-dollar companies. It is also likely a major factor in increasing electricity prices for everyone. Despite all these downsides, it has the potential to improve many aspects of life and benefit humans, society, and even the environment in a myriad of ways. Technophobia is not the best approach. Neither is unbridled technophilia.

     Agentic AI is designed to mimic human intuition to address its intended functions. Machine learning has the potential to find hidden patterns in data that are too difficult or time-consuming for humans to discover. It can analyze vast amounts of data quickly. It is not going to exceed its capabilities and be out-of-control any time soon. Ideas like AI superintelligence – the point where AI exceeds human cognitive abilities in every way, singularity – the idea that when superintelligence occurs, technological growth becomes uncontrollable and irreversible, and artificial general intelligence – the point where AI matches human cognitive abilities – these ideas are interesting and all, but we are likely nowhere near those points. Right now, AI acts basically as a very good digital assistant. There is still a lot of slop and hallucinations being produced by bad training. It does outperform humans on many tasks, and that makes it likely to expand.

     People should probably be madder at cryptocurrency than AI. It uses about a third of the power that AI uses now, although after expected AI growth, that will fall to about 20-25%. Crypto does not provide benefits to society aside from secure transactions. It requires many big data centers. It is also the choice of criminals and terrorists. However, its decentralized nature means it doesn’t require brick-and-mortar facilities for employees like other payment systems, such as credit card companies. That offsets some of its energy intensity.

     Innovation usually requires some trial-and-error in terms of impacts. Technology by itself is neither good nor bad. I can certainly understand why people don’t want to live near data centers. I have heard there are noise and heat island effects. Water consumption is important, but data centers need not produce wastewater that is highly toxic if wastewater treatment is adequate. Nonetheless, treated wastewater needs to be monitored as well. In some places, water used for data centers could compete with drinking water resources.  

    Tech companies building data centers are also backing off on their emissions trajectories and net-zero pledges since they will likely use natural gas to power at least some of their operations. In a lot of cases, there is simply not enough available land space to build out acres of solar panels just to power a portion of operations. Behind-the-meter power is a popular choice that takes some stress off power grids. However, those solutions also compete with grid power for components like transformers and gas turbines, both of which have been in short supply for a few years. A big question is how much power ratepayers are contributing to data center buildout. I don’t want AI to increase my power bill, and I sure as hell don’t want crypto increasing my power bill.

     There are several legitimate reasons to oppose data centers, but I would add that each project is different in terms of the sensitivity of location and effects on local people and the local environment, and each project is powered differently, which affects its impacts.

     On the other hand, I am seeing some social media posts suggesting that AI is some kind of inherent corporate evil that must be stopped at all costs. That is not sensible at all. The misguided young people who planned to use drones and snipers in a terrorist attack to disrupt Trump’s UFC fight cited data centers as one of their grievances.

     Once seen as left-wing oriented, tech companies like Google, Amazon, Meta, and others are now seen as more right-wing corporations hellbent on making a profit at all costs.

     Anti-capitalists also have some fair arguments that AI companies are investing in technology much more than in employees, and as the technology advances, even fewer employees will be needed. Others say that AI will end up being a bubble that will be hard to make the kind of profit predicted.

     Some states are putting the brakes on, such as New York, banning new data center development. Other states, like Ohio, are drawing back incentives meant to lure in these companies for data center projects.