Lately I have been listening to a lot of Volts, a (mostly) good-news podcast about energy transitions. Volts is helpful because it dives deep into specifics; the guests largely work in the energy sector, and talk in great detail about the work they are doing to generate and distribute clean energy, as opposed to the pie-in-the-sky generalizations politicians make about being "energy superpowers".
The sense that I get from this community is that the energy transition is real and underway. A common theme is that there is now a huge additional demand for electricity, which means there are lots of ways for the energy sector to grow. Some of this is because we are transitioning away from fossil fuels towards electricity for things like electric vehicles and home heating, but another huge demand comes from Silicon Valley, which has big bags of money to throw around. Silicon Valley is fully into generative AI, but training (and running!) generative AI models takes lots of datacenters, which need lots of energy. For a while Microsoft was planning to restart Three Mile Island. There are other initiatives underway to power datacentres using geothermal energy and other mechanisms.
Environmentalists have long lamented that AI is an environmental disaster because it requires so much energy. The counterargument is that Silicon Valley is now focusing attention on generating energy in the same way it focused on hardware improvements traditionally, and thus energy will become cheaper and more efficient to generate. Computers are cheap and plentiful because Silicon Valley companies innovated and competed. In fact, one might argue that computer gamers are the reason we have AI now: the gamers demanded better graphics, so companies like NVIDIA rearchitected computer systems (GPUs) that drew a lot of triangles efficiently, and it turns out that same architecture is useful for training neural networks. As the gamers demanded better and better graphics, GPUs got more powerful, and at some point they were powerful enough to fuel the AI boom we have now.
Could the same happen in the energy space? I am not sure. On the one hand, if Silicon Valley behemoths with bags of money to throw around invest heavily in electricity generation, there is some reason to believe we will find ways to generate (and distribute!) energy very cheaply with a smaller environmental footprint per watt. But then Jevons Paradox bites us: as resources become cheaper and more efficient, we use more of them because they are cheaper. As a result we end up using more of the resource than we would have otherwise.
If this new energy generation is low-carbon but not no carbon, we could still be in a lot of trouble with respect to climate change. Generating a huge amount of low-carbon energy still produces a lot of carbon. If this energy generation is no-carbon, then maybe we are okay, and maybe we are in trouble because some other existential issue raises its head (perhaps the laws of thermodynamics? All the heat we generate along with that energy has to go somewhere).
Aesthetically, it is disappointing to think that we might get abundant clean energy because AI datacentres want to suck up all that energy to generate deepfakes and hallucinated undergrad essays. But it seems to me that this has been the way we have proceeded with technology throughout history. Factories are gross and dirty but they were a great leap forward materially. Cars are gross but they solved a problem of geography, which is why we will never get rid of them. I don't think we get an aesthetically pleasing decentralized clean energy transition, because people do not change unless they are forced. Crisis is one forcing function. Capitalist competitiveness is another. Doing good because it is good does not seem to be one.
People on Volts seem to be very optimistic that the future will be clean energy. I would like to agree with that, but maybe Pierre Poilievre and Danielle Smith are correct and the real future (at least in the medium term) is in the tarsands. We have lots of infrastructure available to process hydrocarbons, and we are still building out clean(er) energy infrastructure. If we need a lot of additional energy RIGHT AWAY to power chatbots, then maybe fossil fuels win.