In early markets, hybrids always prevail because there’s not enough electric infrastructure. As it develops and people can charge their cars more easily, the share of EVs increases.
Excellent synthesis of the solar/EV exponentials. The projection gap between OPEC and reality feels like watching 19th century coal barons forecast the futrue of steam engines. I work in energy infrastructure and the speed at which utilites are scrambling to adapt their grids tells the real story, the centralized model everyone built for is getting disrupted way faster than the 2050 timelines suggest.
I read that new data centers are constrained by how much power (electricity) can be delivered. Are new data centers being powered with solar/battery systems? I read about gas turbines and anticipation of nuclear (fission, but also some hype about fusion). But I don't recall reading about big solar installations to power data center installations. (Except maybe in China.) Have I missed it?
Exponentials are killing us and, as Tomas says, are also providing hope. In an exponential meets exponential world, I think ‘predictions’ are peering into the mists.
Possibly space and it will give great power (excuse the pun) to those with launch capabilities - no longer the US but Musk/Bezos. But what about failure modes and why not just collect sunlight back on earth?
Re fusion, it seems to me that once the engineering challenge of controlling a plasma is overcome, the next one will be how to harness the output. Current fission reactors run water pipes to a steam turbine. Somehow I don’t think that’s a very practical idea for something hotter than the sun?
Excellent summary! I've been writing about this epic renewables revolution for years at The Weekly Anthropocene, and I just got back from in-person reporting across China.
What did all those hyperscalers bought? That’s right gas turbines. No matter how inefficient or old. Why not solar and batteries? Do I say not as I do.
I really respect your opinion but this whole sun and wind thing really sounds far fetched to me. It has reduced co2 intensity somewhat , increased prices a lot and led to de industrialization.
Unfortunately, I cannot link to the exact graphic, only the article. The Energy Bad Boys did a similar analysis for Texas, which is outstanding for wind and solar and the disparities were also huge.
They have $67 for new combined-cycle gas turbines and $600 for 100% new solar + battery.
In the US Southwest, solar makes a lot of sense for generating electricity from late spring to early fall during the daytime, but that is only about 20-25% of the year.
But once you push beyond that, the economics make no sense. Just look at realistic calculations for Texas, which is one of the best regions for renewable electricity in the world.
At some point it will be right, you just need to figure out when.
Note what I’m saying: peak oil in the next few years, it it won’t shrink that fast upfront. But it will be shrinking quite fast in 20 years! The transition will be during those two dates.
Great article, and I really want to share your optimism. My question is about China: Are we green-washing their solar use? Doomberg sites an article from the Financial Times that indicates they're building hundreds of new coal-fired plants at the same time they're pushing solar.
“More than 100 coal-fired power station generator units are set to start supplying electricity worldwide in 2026 as China continues to develop projects using the fossil fuel even as it installs huge amounts of wind and solar power. The 104 projects due to open this year are among 439 under construction, according to an analysis of plans by Global Energy Monitor, highlighting the enduring reliance on coal-fired power stations despite massive progress in lower-carbon technologies.
The vast majority of the coal power projects are in China, which accounts for 212 gigawatts of the 256GW total new capacity under construction. China has 85 coal-fired units scheduled to open in 2026, accounting for 55GW of the global 63GW, according to the GEM data. The final tallies could be higher or lower depending on construction timelines.”
Yes it’s true. The thing is that everything in China is going up and to the right, so you have to compare how fast each curve is going.
The curve of energy demand is going up very fast. Solar is growing extremely fast to cover as much of that as possible, and its share is growing and growing. But it’s starting from a small base, so it’s not enough to cover China’s energy demands quite yet. As a result coal is still growing, but slower than total energy, and much slower than solar.
The result, when you project this into the future, is that coal will likely start shrinking in absolute terms in the next few years
Thank you for this post Tomas! Appreciate a lot of your writing on Substack and subscribed due to the quality of your posts.
I have a question however. I know your essay here is focusing on long-term trends of energy consumption and how it is changing, but there’s one fact in the present day that leaves me skeptical about solar and other renewables replacing fossil fuels. Utility bills and electricity prices have gone up due to governments both local and national mandating a change to renewable energy. If these new forms of energy raise prices, won’t voters and their elected leaders shun them eventually, and keep the more reliable pricing of natural gas and oil?
That’s because solar and wind were more expensive than other sources before! Also because regulatory issues and transmission conversion were problems. Now they’re cheaper, and the regulation will catch up. Fossil fuels will have a role, but more limited I believe
>What will happen then to countries like Russia, Venezuela, or Saudi Arabia?
Countries with very cheap to produce oil or natural gas (Saudi Arabia, USA) will maintain (maybe lower margin) extraction for a long time for all the use cases where electrification doesn't yet have good alternatives (high temperature industrial uses like steel smelting, airplane fuel, various classes of plastics and industrial/agricultural chemicals, and a few others uses). Technology will address those eventually, but it could be many more decades. With decreased demand, they'll keep going as all the high-cost extraction stops pumping. So those countries won't be as impacted, and will have a slower offramp.
But the high-cost extraction countries are going to be wrecked. They already are, and often have wild swings even with the current global energy mix. Venezuela was made 10x worse due to their inability to maintain production. If that is due to market conditions rather than just horrible governance, just fixing their governments won't reorganize their economies - in fact the continued failure of the economies probably reinforces the kleptocarcy's need to hold on onto what dwindling resources are left.
"Electricity went from less than 5% electric..." You sure ? :) Notice that the IEA officially announced that conventional peak oil happened back in 2018. So far, the current plateau relies almost exclusively on US tight oil a few other sources. But US tight oil isn't infinite either...
So obviously the Chinese government saw the data and decided that they'll do what must be done. While Trump pretends the data doesn't exist or something.
It’s funny. If you look at the current consumption of oil globally, it already barely grows as you say. Surprising that ppl haven’t caught on to that: peak oil soon is not that polemic, just look at the current data!
I have just completed my own project as part of a municipal endeavour to determine local GHG emissions.
As you say, exponentials everywhere - CO2 emissions, forcing function, sea-level. I am convinced we will hit 2C in 10 years and will need FF reductions of nominally 10% pa thereafter.
You give me hope it may be possible but we will still have to contend with a diminished natural world and economy. But better than hitting 4C by 2050.
I read your previous blog on a plentiful future and am unconvinced but respectful of the fact you have looked at both sides of things.
I have just completed my own project as part of a municipal endeavour to determine local GHG emissions.
As you say, exponentials everywhere - CO2 emissions, forcing function, sea-level. I am convinced we will hit 2C in 10 years and will need FF reductions of nominally 10% pa thereafter.
You give me hope it may be possible but we will still have to contend with a diminished natural world and economy. But better than hitting 4C by 2050.
I read your previous blog on a plentiful future and am unconvinced but respectful of the fact you have looked at both sides of things.
Many of your solar curves are installed capacity,not actual generation. Capacity is useless more than half of the time because of nighttime and clouds, so you need to look at actual generation or installed battery capacity. And when adding batteries as backup, you now need twice as many solar panels to charge the batteries for overnight USA. And for northern areas, like Europe you need more batteries and solar for charging them to cover the diminished daylight in winter and cloudier weather.
As for electric vehicles, American new car sales last year were less than 8% and are not growing quickly. And transportation also includes heavy trucks, trains, ships and aircraft, none of which, other than passenger trains, are being electrified except for small experimental tests.
Hmm about half of the graphs are capacity and the other half is electricity production.
Their connection is pretty straightforward too, as solar can produce electricity equivalent to 25% of the time.
So you 4x capacity and use batteries, and it’s broadly equivalent (up to ~70-90% of the time, when reliability costs start kicking in)
I don’t see how that changes anything!
The US is a laggard here because, among other things, it’s the world’s top oil producer. Please don’t look at one market and say “and therefore all other markets are like that”, that’s cherry picking!
Trains, trucks, buses and ships are also prone to electrification. The one that isn’t is air. That plus petrochemicals are the two use cases for oil in the medium term.
You are neglecting the massive cost of utility-scale batteries. They are on the scale of magnitude of nuclear power, and they do not even generate electricity.
Even the largest in the world does not have the size of output of a typical fossil fuel plant. And there are tens of thousands of those in the world.
I have admittedly not analyzed batteries independently yet, but I don’t leave them out of the analysis. FWIW their LCOE combined with solar is pretty decent already, and their cost has gone down by 40% in 2y. Their price drop has continued for decades now. Just project into the future!
Not sure where you got those pretty decent LCOE numbers, but they likely do not assume the need for 24/7/365 across all geographies which is what all modern societies require.
The scale of batteries that would need to be deployed for that is breathtaking (even if declining).
It's admittedly tough in Northern cloudy areas to provide 24/7/365 at very high reliability. But part of it is to overbuild the solar, not the batteries, so even in cloudy conditions you can produce meaningful energy. Solar panel ASPs are down to <$0.20/watt globally (in the $0.30s here in the US with tariffs and various other restrictions), and large-scale batteries are <$0.25/watt-hour if purchased at scale. So the question is how much PV and battery do you need to serve load (which is also lowet at night and in the winter anyway, in most of the US).
Not going to work, you just generate more electricity at a time and place where it is not needed. This is a particularly bad solution for winter in temperate latitudes. So this “solution” just dramatically increase the price.
In North America, natural gas is far more cost-effective.
The cost of gridscale batteries is an order of magnitude cheaper than nuclear today. And that is also apples and oranges, since batteries do not produce energy, just move it in time. One needs to compare solar+batteries to new nuclear for an apples-to-apples LCOE. And that leaves out the modularity and related reliability benefits of solar+batteries vs. single-unit reactor.
We should really stop overegulating nuclear and start getting better at building it it would help massively in delivering a cleaner and cheaper grid , just look at france vs germany and denmark and you can see which has spent less and got a cleaner and cheaper grid
Yes, it is apples and oranges, but you can compare cost per MWHr.
The problem is that utility-scale batteries typically last only 4 hours and then needs to be recharged over a longer duration, while nuclear power runs 24/7. So you need far more battery capacity to deliver the same amount of electricity.
When you actually calculate the price for 24/7 electricity, they are relatively close. And as you correctly state, batteries do not produce electricity.
Technology is also improving to the point where we can soon "mine" copper from existing mine tailings and garbage dumps. That alone will contribute plenty of supply in addition to more conventional recycling, plus conventional mining and exploration too.
Just a reminder that exponential growth curves often turn into S-curves at some point. There are only finitely many resources, there are carrying capacities, bottlenecks, and so forth. Without regulation, we aren't simply going to have a "nearly vertical line" going from 0% adoption to 100% adoption. Of course with regulation, that is entirely possible.
In fact, where the inflection point on the S curves happen matters for so many things you discuss, including energy consumption, population growth, crop yields, artificial intelligence capabilities, and so on. The argument that "we won't see an inflection point in the near future" is a fine argument, but presuming anything, including model intelligence growth, is going to continue to grow exponentially is unlikely. I think we simply disagree on this point, but considering LLMs are simply generative prediction models, once we get beyond the ability to predict patterns based on current knowledge via matrix multiplication, LLMs will most likely stop improving a rapid rate. This may well be AGI by whatever definition we settle on, but it won't be super intelligence unless we redefine what that means. Ask them to solve an unsolved problem sometime, or ask them to compare their operation speed to the matrix multiplication of 10,000x10,000 matrices.
At best, we get a series of exponential leaps that, when looking back on them will continue to look like exponential growth. But again, opinions on such topics vary - dismissing them is what will, in my opinion, eventually cause an AI bubble which will collapse.
The thing with EV/ is that the trending is probably going to *accelerate*:
- More sales will improve the tech faster, bringing costs down, esp the key element: batteries
- The bigger the installed based, the more infra for electric cars (esp charging stations), so the better the deal. Conversely, fewer gas pumping stations, and worse to get a gas powered car.
- I said the inflection point for AI is likely going to be within 5 years unless we reach ASÍ in that timeline
I am extremely skeptical that renewables will replace fossil fuels anytime within decades (possibly ever). That is not how energy transitions work. In all previous energy transitions, the new energy source is IN ADDITION TO the previous energy source, not replacing it.
Just look at the first graphic in your article. Increased renewables are in addition to even larger increases in fossil fuels over the last 30 years (and nuclear and hydro).
Wood, for example, is still widely burned for energy, despite an energy transition.
Oil did not replace coal.
Natural gas did not replace oil or coal. Nor did nuclear or hydroelectric.
My guess is that fossil fuels will be the dominant form of energy for the next few decades. The fact that it is over 80% today (plus nuclear and hydro) despite energy use massively increasing and despite decades of government subsidies for renewable energy shows how little actual progress has been made.
I predict that by 2050 we combust more fossil fuels than today (just like it has increased during previous decades).
Excellent synthesis of the solar/EV exponentials. The projection gap between OPEC and reality feels like watching 19th century coal barons forecast the futrue of steam engines. I work in energy infrastructure and the speed at which utilites are scrambling to adapt their grids tells the real story, the centralized model everyone built for is getting disrupted way faster than the 2050 timelines suggest.
Ah, very interesting. A data point from the ground. Thanks for sharing!
I read that new data centers are constrained by how much power (electricity) can be delivered. Are new data centers being powered with solar/battery systems? I read about gas turbines and anticipation of nuclear (fission, but also some hype about fusion). But I don't recall reading about big solar installations to power data center installations. (Except maybe in China.) Have I missed it?
Will cover that very soon! From two angles. Think also space!
Exponentials are killing us and, as Tomas says, are also providing hope. In an exponential meets exponential world, I think ‘predictions’ are peering into the mists.
Possibly space and it will give great power (excuse the pun) to those with launch capabilities - no longer the US but Musk/Bezos. But what about failure modes and why not just collect sunlight back on earth?
Re fusion, it seems to me that once the engineering challenge of controlling a plasma is overcome, the next one will be how to harness the output. Current fission reactors run water pipes to a steam turbine. Somehow I don’t think that’s a very practical idea for something hotter than the sun?
Excellent summary! I've been writing about this epic renewables revolution for years at The Weekly Anthropocene, and I just got back from in-person reporting across China.
What did all those hyperscalers bought? That’s right gas turbines. No matter how inefficient or old. Why not solar and batteries? Do I say not as I do.
They are going to, now that building their own electric power generation is on the table, if the NEPA reviews are not blockers
I really respect your opinion but this whole sun and wind thing really sounds far fetched to me. It has reduced co2 intensity somewhat , increased prices a lot and led to de industrialization.
Guess we’ll see…
Yes, in North America combined-cycle gas turbines are far more cost-effective for 24/7/365 electricity than renewables.
Interesting statement! Please do share sources!
Here is one example from Minnesota.
The LCOE of existing combined-cycle gas turbines is $27.08 per MW versus:
New wind + batteries: $272.41
New solar + batteries: $471.60
https://frompovertytoprogress.substack.com/p/why-solar-cannot-displace-fossil
Unfortunately, I cannot link to the exact graphic, only the article. The Energy Bad Boys did a similar analysis for Texas, which is outstanding for wind and solar and the disparities were also huge.
They have $67 for new combined-cycle gas turbines and $600 for 100% new solar + battery.
https://energybadboys.substack.com/p/rising-natural-gas-costs-make-wind
This is why I am not bullish on solar. Nuclear and geothermal are the future.
This is simply not true 1) at current construction costs, 2) in the southwest US (where there is a lot of land and very strong solar resource)
In the US Southwest, solar makes a lot of sense for generating electricity from late spring to early fall during the daytime, but that is only about 20-25% of the year.
But once you push beyond that, the economics make no sense. Just look at realistic calculations for Texas, which is one of the best regions for renewable electricity in the world.
https://energybadboys.substack.com/p/rising-natural-gas-costs-make-wind
Scroll down and look at the graphic labelled "Always on cost of energy..."
As much as I want to believe this, I’ve been hearing it for well over 30 years now
At some point it will be right, you just need to figure out when.
Note what I’m saying: peak oil in the next few years, it it won’t shrink that fast upfront. But it will be shrinking quite fast in 20 years! The transition will be during those two dates.
Great article, and I really want to share your optimism. My question is about China: Are we green-washing their solar use? Doomberg sites an article from the Financial Times that indicates they're building hundreds of new coal-fired plants at the same time they're pushing solar.
https://newsletter.doomberg.com/p/assisted-thinking?utm_source=post-email-title&publication_id=343139&post_id=185072557&utm_campaign=email-post-title&isFreemail=false&r=254yy&triedRedirect=true&utm_medium=email
“More than 100 coal-fired power station generator units are set to start supplying electricity worldwide in 2026 as China continues to develop projects using the fossil fuel even as it installs huge amounts of wind and solar power. The 104 projects due to open this year are among 439 under construction, according to an analysis of plans by Global Energy Monitor, highlighting the enduring reliance on coal-fired power stations despite massive progress in lower-carbon technologies.
The vast majority of the coal power projects are in China, which accounts for 212 gigawatts of the 256GW total new capacity under construction. China has 85 coal-fired units scheduled to open in 2026, accounting for 55GW of the global 63GW, according to the GEM data. The final tallies could be higher or lower depending on construction timelines.”
Yes it’s true. The thing is that everything in China is going up and to the right, so you have to compare how fast each curve is going.
The curve of energy demand is going up very fast. Solar is growing extremely fast to cover as much of that as possible, and its share is growing and growing. But it’s starting from a small base, so it’s not enough to cover China’s energy demands quite yet. As a result coal is still growing, but slower than total energy, and much slower than solar.
The result, when you project this into the future, is that coal will likely start shrinking in absolute terms in the next few years
Thank you for this post Tomas! Appreciate a lot of your writing on Substack and subscribed due to the quality of your posts.
I have a question however. I know your essay here is focusing on long-term trends of energy consumption and how it is changing, but there’s one fact in the present day that leaves me skeptical about solar and other renewables replacing fossil fuels. Utility bills and electricity prices have gone up due to governments both local and national mandating a change to renewable energy. If these new forms of energy raise prices, won’t voters and their elected leaders shun them eventually, and keep the more reliable pricing of natural gas and oil?
That’s because solar and wind were more expensive than other sources before! Also because regulatory issues and transmission conversion were problems. Now they’re cheaper, and the regulation will catch up. Fossil fuels will have a role, but more limited I believe
>What will happen then to countries like Russia, Venezuela, or Saudi Arabia?
Countries with very cheap to produce oil or natural gas (Saudi Arabia, USA) will maintain (maybe lower margin) extraction for a long time for all the use cases where electrification doesn't yet have good alternatives (high temperature industrial uses like steel smelting, airplane fuel, various classes of plastics and industrial/agricultural chemicals, and a few others uses). Technology will address those eventually, but it could be many more decades. With decreased demand, they'll keep going as all the high-cost extraction stops pumping. So those countries won't be as impacted, and will have a slower offramp.
But the high-cost extraction countries are going to be wrecked. They already are, and often have wild swings even with the current global energy mix. Venezuela was made 10x worse due to their inability to maintain production. If that is due to market conditions rather than just horrible governance, just fixing their governments won't reorganize their economies - in fact the continued failure of the economies probably reinforces the kleptocarcy's need to hold on onto what dwindling resources are left.
Yes!
But the US has costly oil!
"Electricity went from less than 5% electric..." You sure ? :) Notice that the IEA officially announced that conventional peak oil happened back in 2018. So far, the current plateau relies almost exclusively on US tight oil a few other sources. But US tight oil isn't infinite either...
So obviously the Chinese government saw the data and decided that they'll do what must be done. While Trump pretends the data doesn't exist or something.
It’s funny. If you look at the current consumption of oil globally, it already barely grows as you say. Surprising that ppl haven’t caught on to that: peak oil soon is not that polemic, just look at the current data!
Outstanding Tomas,
I have just completed my own project as part of a municipal endeavour to determine local GHG emissions.
As you say, exponentials everywhere - CO2 emissions, forcing function, sea-level. I am convinced we will hit 2C in 10 years and will need FF reductions of nominally 10% pa thereafter.
You give me hope it may be possible but we will still have to contend with a diminished natural world and economy. But better than hitting 4C by 2050.
I read your previous blog on a plentiful future and am unconvinced but respectful of the fact you have looked at both sides of things.
Outstanding Tomas,
I have just completed my own project as part of a municipal endeavour to determine local GHG emissions.
As you say, exponentials everywhere - CO2 emissions, forcing function, sea-level. I am convinced we will hit 2C in 10 years and will need FF reductions of nominally 10% pa thereafter.
You give me hope it may be possible but we will still have to contend with a diminished natural world and economy. But better than hitting 4C by 2050.
I read your previous blog on a plentiful future and am unconvinced but respectful of the fact you have looked at both sides of things.
It will happen indeed!
And soon.
How soon is hard to tell. That’s the thing with exponentials.
What’s hopeful is that now the optimistic scenario is the CURRENT trajectory!
Many of your solar curves are installed capacity,not actual generation. Capacity is useless more than half of the time because of nighttime and clouds, so you need to look at actual generation or installed battery capacity. And when adding batteries as backup, you now need twice as many solar panels to charge the batteries for overnight USA. And for northern areas, like Europe you need more batteries and solar for charging them to cover the diminished daylight in winter and cloudier weather.
As for electric vehicles, American new car sales last year were less than 8% and are not growing quickly. And transportation also includes heavy trucks, trains, ships and aircraft, none of which, other than passenger trains, are being electrified except for small experimental tests.
Hmm about half of the graphs are capacity and the other half is electricity production.
Their connection is pretty straightforward too, as solar can produce electricity equivalent to 25% of the time.
So you 4x capacity and use batteries, and it’s broadly equivalent (up to ~70-90% of the time, when reliability costs start kicking in)
I don’t see how that changes anything!
The US is a laggard here because, among other things, it’s the world’s top oil producer. Please don’t look at one market and say “and therefore all other markets are like that”, that’s cherry picking!
Trains, trucks, buses and ships are also prone to electrification. The one that isn’t is air. That plus petrochemicals are the two use cases for oil in the medium term.
You are neglecting the massive cost of utility-scale batteries. They are on the scale of magnitude of nuclear power, and they do not even generate electricity.
Even the largest in the world does not have the size of output of a typical fossil fuel plant. And there are tens of thousands of those in the world.
I have admittedly not analyzed batteries independently yet, but I don’t leave them out of the analysis. FWIW their LCOE combined with solar is pretty decent already, and their cost has gone down by 40% in 2y. Their price drop has continued for decades now. Just project into the future!
Not sure where you got those pretty decent LCOE numbers, but they likely do not assume the need for 24/7/365 across all geographies which is what all modern societies require.
The scale of batteries that would need to be deployed for that is breathtaking (even if declining).
It's admittedly tough in Northern cloudy areas to provide 24/7/365 at very high reliability. But part of it is to overbuild the solar, not the batteries, so even in cloudy conditions you can produce meaningful energy. Solar panel ASPs are down to <$0.20/watt globally (in the $0.30s here in the US with tariffs and various other restrictions), and large-scale batteries are <$0.25/watt-hour if purchased at scale. So the question is how much PV and battery do you need to serve load (which is also lowet at night and in the winter anyway, in most of the US).
Not going to work, you just generate more electricity at a time and place where it is not needed. This is a particularly bad solution for winter in temperate latitudes. So this “solution” just dramatically increase the price.
In North America, natural gas is far more cost-effective.
The cost of gridscale batteries is an order of magnitude cheaper than nuclear today. And that is also apples and oranges, since batteries do not produce energy, just move it in time. One needs to compare solar+batteries to new nuclear for an apples-to-apples LCOE. And that leaves out the modularity and related reliability benefits of solar+batteries vs. single-unit reactor.
We should really stop overegulating nuclear and start getting better at building it it would help massively in delivering a cleaner and cheaper grid , just look at france vs germany and denmark and you can see which has spent less and got a cleaner and cheaper grid
Yes, it is apples and oranges, but you can compare cost per MWHr.
The problem is that utility-scale batteries typically last only 4 hours and then needs to be recharged over a longer duration, while nuclear power runs 24/7. So you need far more battery capacity to deliver the same amount of electricity.
When you actually calculate the price for 24/7 electricity, they are relatively close. And as you correctly state, batteries do not produce electricity.
https://frompovertytoprogress.substack.com/p/utility-scale-batteries-are-as-expensive.
LCOE is not a valid calculation for renewable energy as it varies so much based on geography and time.
As Arjun Murti has explained at length, peak oil is decades away. One link:
https://arjunmurti.substack.com/p/obliterating-peak-oil-demand-a-progress
Just skimmed it. No mention of the biggest force, electrification/solar/batteries
Also overfitting to the EU EV slump from a few years ago, which as I mention is unlikely to be valid for long
I recently read something to the effect that we are exhausting the supply of recoverable copper and that this will curtail electrification. Thoughts?
Extremely unlikely. We’ve never ran out of any mineral. Copper is common enough that if we look for it we can find it.
Technology is also improving to the point where we can soon "mine" copper from existing mine tailings and garbage dumps. That alone will contribute plenty of supply in addition to more conventional recycling, plus conventional mining and exploration too.
Just a reminder that exponential growth curves often turn into S-curves at some point. There are only finitely many resources, there are carrying capacities, bottlenecks, and so forth. Without regulation, we aren't simply going to have a "nearly vertical line" going from 0% adoption to 100% adoption. Of course with regulation, that is entirely possible.
In fact, where the inflection point on the S curves happen matters for so many things you discuss, including energy consumption, population growth, crop yields, artificial intelligence capabilities, and so on. The argument that "we won't see an inflection point in the near future" is a fine argument, but presuming anything, including model intelligence growth, is going to continue to grow exponentially is unlikely. I think we simply disagree on this point, but considering LLMs are simply generative prediction models, once we get beyond the ability to predict patterns based on current knowledge via matrix multiplication, LLMs will most likely stop improving a rapid rate. This may well be AGI by whatever definition we settle on, but it won't be super intelligence unless we redefine what that means. Ask them to solve an unsolved problem sometime, or ask them to compare their operation speed to the matrix multiplication of 10,000x10,000 matrices.
At best, we get a series of exponential leaps that, when looking back on them will continue to look like exponential growth. But again, opinions on such topics vary - dismissing them is what will, in my opinion, eventually cause an AI bubble which will collapse.
The thing with EV/ is that the trending is probably going to *accelerate*:
- More sales will improve the tech faster, bringing costs down, esp the key element: batteries
- The bigger the installed based, the more infra for electric cars (esp charging stations), so the better the deal. Conversely, fewer gas pumping stations, and worse to get a gas powered car.
- I said the inflection point for AI is likely going to be within 5 years unless we reach ASÍ in that timeline
I am extremely skeptical that renewables will replace fossil fuels anytime within decades (possibly ever). That is not how energy transitions work. In all previous energy transitions, the new energy source is IN ADDITION TO the previous energy source, not replacing it.
Just look at the first graphic in your article. Increased renewables are in addition to even larger increases in fossil fuels over the last 30 years (and nuclear and hydro).
Wood, for example, is still widely burned for energy, despite an energy transition.
Oil did not replace coal.
Natural gas did not replace oil or coal. Nor did nuclear or hydroelectric.
My guess is that fossil fuels will be the dominant form of energy for the next few decades. The fact that it is over 80% today (plus nuclear and hydro) despite energy use massively increasing and despite decades of government subsidies for renewable energy shows how little actual progress has been made.
I predict that by 2050 we combust more fossil fuels than today (just like it has increased during previous decades).