And yet, we're not completely stuck. It is absolutely clear that not enough has been done to reduce our carbon emissions, and we're on a bad path on track to ~2.5°C warming in the next century. However, something has been done, and if nothing had been done we could easily be on track to >4°C global warming. That would be much worse.
So, how did we achieve what little we have? Well, because many people have cared, and have made the right decisions. Not enough people, or maybe not good-enough decisions, but some people, and somewhat good decisions.
So, what were the decisions which brought us down from an apocalyptic +4°C to a very bad 2.5°C path? Was it enough consumers making environmentally conscious choices, even if they were less convenient or more expensive. No. It was enough voters wanting their leaders to do something, even if it wasn't quite enough, but it was something. And something isn't nothing.
We will never have enough people voting with their wallet to fight climate change, because our rational understanding of the big-picture cannot overpower our intuitive day-to-day choices. However, we may have enough people voting with their ballot to fight climate change, because the rational big-picture can, sometimes, decide whom we vote for.
Considering that current energy usage of AI is very likely to plummet in the coming years due to efficiency gains (which we've seem massive improvements on in the last couple years) and most of these large companies building datacenters are looking for and investing in specifically clean energy sources (nuclear) I don't think AI is a meaningful contributor compared to all the other high causes of global emissions.
Sure, but I meant consumption on a per-task basis. I would also add that at some point building new software is going to become rather valueless as society trends towards building effectively any variant of software anyone could want relatively instantly and cheaply.
There is also a finite number of consumers of AI content on the planet - sure maybe everyone want 4k-per eye live feed video of anything they want 24/7, but once you accomplish that for all 8 billion humans, there is no further demand for it.
Generation costs are very likely to continue to trends rapidly downward and there is a likely final wall of saturation of demand.
I don't expect fusion to make a difference in either direction.
For big reactors: At current aluminium prices, the bill of materials for a global power grid with 1 Ω resistance the long way round is only about 10x the cost ITER (the organisation) expects ITER (the reactor) to be. Plug in your own numbers for target resistance as desired, halve resistance needs double the material.
With such a grid, you can put the PV in Angola and still get useful output in mid-winter nighttime in Anchorage.
For potential small reactors like Helion's "shipping container" target size, I won't say it can't work (I don't know enough to be confident), but I will say that we immediately find we have bigger problems because any hostile actor can simply choose to run them in neutron-source mode and turn everyday cheap depleted uranium* into weaponizable plutonium.
* I note that eBay still clearly has a dictionary merge on all nouns, given my search results came up with this:
Get the best deals for Depleted Uranium Metal at eBay.com. We have a great online selection at the lowest prices with Fast & Free shipping on many items!
Also be aware that US restrictions on sales aren't particularly relevant to this, as the moment this Columbus' Egg** gets solved it rapidly becomes a global problem.
Completely agree with the outlook ("fusion power irrelevant for climate change in every realistic scenario").
But where did you take those grid cost numbers from? Iter costs are <100bn AFAIC; and Germany alone (!!) projects more than that (top end) for grid expansion/operation within 2040 (mainly north/south and offshore connectivity).
Putting this note first, because it's probably the main point of confusion/surprise: I did say "bill of materials"; the estimated full cost for European and US grid upgrades that we need anyway for other reasons, with far less material, is order-of a trillion or so for US, half a trillion for EU.
For scale, this is about what China makes in 2 years; if this is rolled out over 30, which would be optimistic but plausible, it's within the realm of just how much China increased production between 2023 and 2025, being spent every year.
To get to 10x ITER's own estimate for ITER, the wikipedia page says the organisation estimates the reactor will cost about €18-22bn: https://en.wikipedia.org/wiki/ITER
There are a lot of reasons not to do this as a single big 1m^2 "wire", amongst them being that the surface magnetic field is strong enough to be dangerous to approach with ferrous materials.
The "square meter of aluminium" is an interesting take. Not sure how much power you'd get over that thing; extrapolating from existing HVDC systems (Inga-Shaba is 1GW over 2x520mm²), I'd expect around 1TW, so twice the US demand?
But because Nimbys have no appreciation for beautiful pylons, projects in that direction are doomed for now anyway and everything needs to be buried underground at extra cost :(
Losses are I^2 R, which has the annoying consequence that % loss depends on how much juice you put through it, it isn't a constant percentage.
Pylons… eh. Doing this realistically rather than my napkin-maths, it would be a mix of many different solutions in different parts of the world, from competing environmental issues. Some would be pylons, some underground cables. Is the Sahara dry enough to run it on the ground, or in a concrete trench? I have no idea.
As a side note, every so often I keep being surprised on here by Americans who can't rely on the grid in winter because snow disables it, and some Californian forest fires are attributed to unmaintained pylons failing, dropping live wires onto the forest where they spark and light up the dry wood. These could both be resolved by burying more cables. Likewise within urban areas: here in Europe it's rather rare to see overhead lines in urban or suburban areas, unless they're over a tram/railway line.
IMO the real killer of any project like this, is geopolitics, not local politics. EU doesn't trust China, the US, or Russia; the current US administration doesn't trust or doesn't like basically everyone; Russia kinda gets along with China but few else; China would like to sell stuff to everyone but also have border disputes and other friction with many of their neighbours.
I do agree that politics is a big enemy here (with nations historically being most willing to spend big on energy when reducing interdependence, like the french Messmer plan, instead of the reverse).
Honestly the "1m² around the world" is probably a pretty good proxy for what we would need to solve intermittency problems exclusively by boosting grid connectivity (instead of storage), rescaling this to 4m² cumulative cross-section could probably transport the total global electrical energy consumption over ~10000km (but the losses would get uncomfortably high from an economical point of view after crossing the 1000-2000km distance threshold, so you might want even more aluminium when you desire connections that long).
Btw: Underground cables instead of pylons are absolutely a nimby thing. Not only is it much more expensive, because you pay for the earthwork and additional insulator, but it also limits your voltage (to avoid overpaying for insulation even more).
Return current is typically free, either from balancing 3phase AC or because decent electrodes get you <1Ohm over any distance thanks to math for the DC case. Just talking high voltage here; for short range, lower power residential connections the situation is different.
So, how did we achieve what little we have? Well, because many people have cared, and have made the right decisions. Not enough people, or maybe not good-enough decisions, but some people, and somewhat good decisions.
So, what were the decisions which brought us down from an apocalyptic +4°C to a very bad 2.5°C path? Was it enough consumers making environmentally conscious choices, even if they were less convenient or more expensive. No. It was enough voters wanting their leaders to do something, even if it wasn't quite enough, but it was something. And something isn't nothing.
We will never have enough people voting with their wallet to fight climate change, because our rational understanding of the big-picture cannot overpower our intuitive day-to-day choices. However, we may have enough people voting with their ballot to fight climate change, because the rational big-picture can, sometimes, decide whom we vote for.