Coal demand hit record high in 2024.

[SpareChange]

https://www.powerengineeringint.com/coal-fired/strategic-development-coal-fired/global-coal-demand-set-to-remain-at-record-high-until-2027-finds-iea/

I don't pretend to know a ton about the energy transition, but are there any educated opinions about whether global energy demand growth is simply going to outstrip our ability to install cleaner energy sources?

[GuitarStv]

https://www.powerengineeringint.com/coal-fired/strategic-development-coal-fired/global-coal-demand-set-to-remain-at-record-high-until-2027-finds-iea/

I don't pretend to know a ton about the energy transition, but are there any educated opinions about whether global energy demand growth is simply going to outstrip our ability to install cleaner energy sources?

https://ourworldindata.org/grapher/global-energy-substitution

There has been a lot of growth in renewable energy . . . but altogether none of it even approaches the use of GHG causing energy.  We do seem to be making more renewable energy than yearly increases in energy demand, which is good.  Of course, energy demand is likely to get much higher as the climate warms and life becomes impossible without A/C . . .

[jrhampt]

Could part of this be AI/data processing energy demand?

[GuitarStv]

Could part of this be AI/data processing energy demand?

Yep, that and crypto.  Both of which are expected to continue to skyrocket.

[SpareChange]

Quite the growth story...

The report finds that data centers consumed about 4.4% of total U.S. electricity in 2023 and are expected to consume approximately 6.7 to 12% of total U.S. electricity by 2028. The report indicates that total data center electricity usage climbed from 58 TWh in 2014 to 176 TWh in 2023 and estimates an increase between 325 to 580 TWh by 2028. 

https://www.energy.gov/articles/doe-releases-new-report-evaluating-increase-electricity-demand-data-centers

[SpareChange]

https://ourworldindata.org/grapher/global-energy-substitution

There has been a lot of growth in renewable energy . . . but altogether none of it even approaches the use of GHG causing energy.  We do seem to be making more renewable energy than yearly increases in energy demand, which is good.  Of course, energy demand is likely to get much higher as the climate warms and life becomes impossible without A/C . . .

Man, on that chart solar and wind barely register, hydro and nuclear aren't moving much, and the fossils are at all time highs. I know solar especially is taking off, but that is quite a hill to climb.

[41_swish]

I wonder how much of this has to do with the raise in AI. To the best of my knowledge, AI uses a TON of electricity. In order to maintain a stable grid and keep the loss of load probability as low as possible, power companies will use fossil fuels.

Renewables have certainly made great strides over the last 10 years and will continue to do so, but maybe all of this AI stuff comes at a cost. I am sure all the AC that was used this year for how dang hot it was does not help either.

[GuitarStv]

I wonder how much of this has to do with the raise in AI. To the best of my knowledge, AI uses a TON of electricity. In order to maintain a stable grid and keep the loss of load probability as low as possible, power companies will use fossil fuels.

Renewables have certainly made great strides over the last 10 years and will continue to do so, but maybe all of this AI stuff comes at a cost. I am sure all the AC that was used this year for how dang hot it was does not help either.

Spoke to a Microsoft engineer on the GPT-6 training cluster project. He kvetched about the pain they're having provisioning infiniband-class links between GPUs in different regions.

Me: "why not just colocate the cluster in one region?"
Him: "Oh yeah we tried that first. We can't put more than 100K H100s in a single state without bringing down the power grid."

Yikes.  And that's just one company, running one set of AIs.

[Ron Scott]

The real drive in energy usage has been the underdeveloped countries being lifted out of dire poverty into what we’d consider lower middle class. Renewables are not making much of a dent, but are worth investing in for the future.

[NorCal]

It's worth reading the actual report (or at least the executive summary) found at the link below.  The narratives mentioned in this thread are largely unrelated to the actual report. 

https://www.iea.org/reports/coal-2024

This is a global story, not a US story.  While US electricity growth is continuing rapidly due to things like AI and electrification, coal use in the US is expected to continue to fall over the next few years.  No one is opening new coal plants in the US (although their utilization might go up), and many are still scheduled to close.

The plateau story largely has to do with places like China, India, Indonesia, etc.  These areas are expected to increase usage over the next few years while it declines in the US and EU. 

[Ron Scott]

It's worth reading the actual report (or at least the executive summary) found at the link below.  The narratives mentioned in this thread are largely unrelated to the actual report. 

https://www.iea.org/reports/coal-2024

This is a global story, not a US story.  While US electricity growth is continuing rapidly due to things like AI and electrification, coal use in the US is expected to continue to fall over the next few years.  No one is opening new coal plants in the US (although their utilization might go up), and many are still scheduled to close.

The plateau story largely has to do with places like China, India, Indonesia, etc.  These areas are expected to increase usage over the next few years while it declines in the US and EU.

There is an odd tendency for some groups in the US to believe that the fate of the world’s energy mix and resultant affects of global climate change can be controlled by Congress.

[GuitarStv]

It's worth reading the actual report (or at least the executive summary) found at the link below.  The narratives mentioned in this thread are largely unrelated to the actual report. 

https://www.iea.org/reports/coal-2024

This is a global story, not a US story.  While US electricity growth is continuing rapidly due to things like AI and electrification, coal use in the US is expected to continue to fall over the next few years.  No one is opening new coal plants in the US (although their utilization might go up), and many are still scheduled to close.

The plateau story largely has to do with places like China, India, Indonesia, etc.  These areas are expected to increase usage over the next few years while it declines in the US and EU.

There is an odd tendency for some groups in the US to believe that the fate of the world’s energy mix and resultant affects of global climate change can be controlled by Congress.

Better than believing that the world's energy use has no climate impact at all.

[GilesMM]

Global energy consumption will continue to rise rapidly for decades and all sources will be required to keep up with demand.  Climate considerations won't really be taken into factor by most people until the situation is dire for them personally.

[Fireball]

I wonder how much of this has to do with the raise in AI. To the best of my knowledge, AI uses a TON of electricity. In order to maintain a stable grid and keep the loss of load probability as low as possible, power companies will use fossil fuels.

Renewables have certainly made great strides over the last 10 years and will continue to do so, but maybe all of this AI stuff comes at a cost. I am sure all the AC that was used this year for how dang hot it was does not help either.

Spoke to a Microsoft engineer on the GPT-6 training cluster project. He kvetched about the pain they're having provisioning infiniband-class links between GPUs in different regions.

Me: "why not just colocate the cluster in one region?"
Him: "Oh yeah we tried that first. We can't put more than 100K H100s in a single state without bringing down the power grid."

Yikes.  And that's just one company, running one set of AIs.

And that one company is literally bringing a nuclear reactor online to help cope with AI energy requirements. Wild stuff for sure.

www.npr.org/2024/09/20/nx-s1-5120581/three-mile-island-nuclear-power-plant-microsoft-ai

[GuitarStv]

I wonder how much of this has to do with the raise in AI. To the best of my knowledge, AI uses a TON of electricity. In order to maintain a stable grid and keep the loss of load probability as low as possible, power companies will use fossil fuels.

Renewables have certainly made great strides over the last 10 years and will continue to do so, but maybe all of this AI stuff comes at a cost. I am sure all the AC that was used this year for how dang hot it was does not help either.

Spoke to a Microsoft engineer on the GPT-6 training cluster project. He kvetched about the pain they're having provisioning infiniband-class links between GPUs in different regions.

Me: "why not just colocate the cluster in one region?"
Him: "Oh yeah we tried that first. We can't put more than 100K H100s in a single state without bringing down the power grid."

Yikes.  And that's just one company, running one set of AIs.

And that one company is literally bringing a nuclear reactor online to help cope with AI energy requirements. Wild stuff for sure.

www.npr.org/2024/09/20/nx-s1-5120581/three-mile-island-nuclear-power-plant-microsoft-ai

Imagine if we could harness all that nuclear power for good instead of misinformation, photographs of people with too many fingers, and high school kids cheating on essays.

[41_swish]

I wonder how much of this has to do with the raise in AI. To the best of my knowledge, AI uses a TON of electricity. In order to maintain a stable grid and keep the loss of load probability as low as possible, power companies will use fossil fuels.

Renewables have certainly made great strides over the last 10 years and will continue to do so, but maybe all of this AI stuff comes at a cost. I am sure all the AC that was used this year for how dang hot it was does not help either.

Spoke to a Microsoft engineer on the GPT-6 training cluster project. He kvetched about the pain they're having provisioning infiniband-class links between GPUs in different regions.

Me: "why not just colocate the cluster in one region?"
Him: "Oh yeah we tried that first. We can't put more than 100K H100s in a single state without bringing down the power grid."

Yikes.  And that's just one company, running one set of AIs.

And that one company is literally bringing a nuclear reactor online to help cope with AI energy requirements. Wild stuff for sure.

www.npr.org/2024/09/20/nx-s1-5120581/three-mile-island-nuclear-power-plant-microsoft-ai

Imagine if we could harness all that nuclear power for good instead of misinformation, photographs of people with too many fingers, and high school kids cheating on essays.

We are basically cooking the environment just to train the AIs on skibidi Biden memes.... I am sure we will find a good use for it someday, but right now it seems fickle.

[sonofsven]

https://ourworldindata.org/grapher/global-energy-substitution

There has been a lot of growth in renewable energy . . . but altogether none of it even approaches the use of GHG causing energy.  We do seem to be making more renewable energy than yearly increases in energy demand, which is good.  Of course, energy demand is likely to get much higher as the climate warms and life becomes impossible without A/C . . .

Man, on that chart solar and wind barely register, hydro and nuclear aren't moving much, and the fossils are at all time highs. I know solar especially is taking off, but that is quite a hill to climb.

Speaking of hydro, dams are being removed across the western US, starting with the Elwha in Washington, currently the Klamath in OR/CA, next possibly the three lower Snake River dams.
This is due to the ruinous effects on the salmon runs.
Here's a video about the rewilding of the Elwha post dam removal: https://youtu.be/TZWmy_H9pxY?si=4x6pDgBdxFCSLZQF

[GilesMM]

https://ourworldindata.org/grapher/global-energy-substitution

There has been a lot of growth in renewable energy . . . but altogether none of it even approaches the use of GHG causing energy.  We do seem to be making more renewable energy than yearly increases in energy demand, which is good.  Of course, energy demand is likely to get much higher as the climate warms and life becomes impossible without A/C . . .

Man, on that chart solar and wind barely register, hydro and nuclear aren't moving much, and the fossils are at all time highs. I know solar especially is taking off, but that is quite a hill to climb.

Speaking of hydro, dams are being removed across the western US, starting with the Elwha in Washington, currently the Klamath in OR/CA, next possibly the three lower Snake River dams.
This is due to the ruinous effects on the salmon runs.
Here's a video about the rewilding of the Elwha post dam removal: https://youtu.be/TZWmy_H9pxY?si=4x6pDgBdxFCSLZQF

There are over 100,000 dams in the US and most are old and not worth the cost required to maintain and operate them.  The net savings to the city/county/state operators is a huge driver for shutting them down.  The ecological effects are often a fringe benefit.

[Wintergreen78]

This lets me share some of my favorite charts. Lawrence Livermore National Labs puts out Sankey charts, which show sources for power, how much is used, and how much is lost as waste heat. These are done for the US. I’d be very interested in seeing something similar for the whole world.

https://flowcharts.llnl.gov/commodities/energy

In 2023 the US consumed 93.6 quadrillion Btus, of which 8.2 were from coal, 35.4 from petroleum, 33.4 from natural gas, 2.4 from solar and wind.

In 2013 the numbers were: 97.4 quads total, 18 from coal, 35.1 from petroleum, 26.6 from natural gas, 1.9 from solar and wind.

So, for all the talk about renewables, they are still a tiny portion of the overall picture. The bigger story is that US energy use is actually down over the last 10 years, despite more people, bigger gdp, etc. Energy efficiency works!

Here’s a link where you can see energy/gdp data for other countries. In most “rich” countries, the same pattern is developing.

https://ourworldindata.org/energy-gdp-decoupling

[NorCal]

This lets me share some of my favorite charts. Lawrence Livermore National Labs puts out Sankey charts, which show sources for power, how much is used, and how much is lost as waste heat. These are done for the US. I’d be very interested in seeing something similar for the whole world.

https://flowcharts.llnl.gov/commodities/energy

In 2023 the US consumed 93.6 quadrillion Btus, of which 8.2 were from coal, 35.4 from petroleum, 33.4 from natural gas, 2.4 from solar and wind.

In 2013 the numbers were: 97.4 quads total, 18 from coal, 35.1 from petroleum, 26.6 from natural gas, 1.9 from solar and wind.

So, for all the talk about renewables, they are still a tiny portion of the overall picture. The bigger story is that US energy use is actually down over the last 10 years, despite more people, bigger gdp, etc. Energy efficiency works!

Here’s a link where you can see energy/gdp data for other countries. In most “rich” countries, the same pattern is developing.

https://ourworldindata.org/energy-gdp-decoupling

The Stanley diagrams are great, but the visualization does make renewables look smaller than they are.

The energy efficiency of most steam-turbine energy sources are in the 30%-40% range IIRC. It can be higher for combined cycle plants, but those are not particularly common.  Then you add on about 10% grid transmission losses. This is why the “rejected energy” category is so large.  Gas engines on vehicles are similar, with efficiency in the 35%ish range.

Renewables don’t have this problem. You subtract out the 10% transmission losses and the rest is usable energy.

One of my favorite energy commentators (Michael Liebrich) delves into this a lot as the primary energy fallacy. Many who fall into the camp of “the energy transition is too hard” focus on replacing the inputs on the left hand side of that chart.

In reality, we need to replace the energy services on the right hand side of the chart. Most transition technologies produce negligible rejected energy.

I’ll also reiterate my point that the story of global coal use has very little to do with the US.

[Wintergreen78]

This lets me share some of my favorite charts. Lawrence Livermore National Labs puts out Sankey charts, which show sources for power, how much is used, and how much is lost as waste heat. These are done for the US. I’d be very interested in seeing something similar for the whole world.

https://flowcharts.llnl.gov/commodities/energy

In 2023 the US consumed 93.6 quadrillion Btus, of which 8.2 were from coal, 35.4 from petroleum, 33.4 from natural gas, 2.4 from solar and wind.

In 2013 the numbers were: 97.4 quads total, 18 from coal, 35.1 from petroleum, 26.6 from natural gas, 1.9 from solar and wind.

So, for all the talk about renewables, they are still a tiny portion of the overall picture. The bigger story is that US energy use is actually down over the last 10 years, despite more people, bigger gdp, etc. Energy efficiency works!

Here’s a link where you can see energy/gdp data for other countries. In most “rich” countries, the same pattern is developing.

https://ourworldindata.org/energy-gdp-decoupling

The Stanley diagrams are great, but the visualization does make renewables look smaller than they are.

The energy efficiency of most steam-turbine energy sources are in the 30%-40% range IIRC. It can be higher for combined cycle plants, but those are not particularly common.  Then you add on about 10% grid transmission losses. This is why the “rejected energy” category is so large.  Gas engines on vehicles are similar, with efficiency in the 35%ish range.

Renewables don’t have this problem. You subtract out the 10% transmission losses and the rest is usable energy.

One of my favorite energy commentators (Michael Liebrich) delves into this a lot as the primary energy fallacy. Many who fall into the camp of “the energy transition is too hard” focus on replacing the inputs on the left hand side of that chart.

In reality, we need to replace the energy services on the right hand side of the chart. Most transition technologies produce negligible rejected energy.

I’ll also reiterate my point that the story of global coal use has very little to do with the US.

Agreed - you have to put over 200 megawatts worth of natural gas into a combined cycle natural gas turbine power plant to get 100 megawatts of power into the grid. For a simple cycle power plant it is likely closer to 300 megawatts in to get 100 megawatts out.

[NorCal]

This lets me share some of my favorite charts. Lawrence Livermore National Labs puts out Sankey charts, which show sources for power, how much is used, and how much is lost as waste heat. These are done for the US. I’d be very interested in seeing something similar for the whole world.

https://flowcharts.llnl.gov/commodities/energy

In 2023 the US consumed 93.6 quadrillion Btus, of which 8.2 were from coal, 35.4 from petroleum, 33.4 from natural gas, 2.4 from solar and wind.

In 2013 the numbers were: 97.4 quads total, 18 from coal, 35.1 from petroleum, 26.6 from natural gas, 1.9 from solar and wind.

So, for all the talk about renewables, they are still a tiny portion of the overall picture. The bigger story is that US energy use is actually down over the last 10 years, despite more people, bigger gdp, etc. Energy efficiency works!

Here’s a link where you can see energy/gdp data for other countries. In most “rich” countries, the same pattern is developing.

https://ourworldindata.org/energy-gdp-decoupling

The Stanley diagrams are great, but the visualization does make renewables look smaller than they are.

The energy efficiency of most steam-turbine energy sources are in the 30%-40% range IIRC. It can be higher for combined cycle plants, but those are not particularly common.  Then you add on about 10% grid transmission losses. This is why the “rejected energy” category is so large.  Gas engines on vehicles are similar, with efficiency in the 35%ish range.

Renewables don’t have this problem. You subtract out the 10% transmission losses and the rest is usable energy.

One of my favorite energy commentators (Michael Liebrich) delves into this a lot as the primary energy fallacy. Many who fall into the camp of “the energy transition is too hard” focus on replacing the inputs on the left hand side of that chart.

In reality, we need to replace the energy services on the right hand side of the chart. Most transition technologies produce negligible rejected energy.

I’ll also reiterate my point that the story of global coal use has very little to do with the US.

Agreed - you have to put over 200 megawatts worth of natural gas into a combined cycle natural gas turbine power plant to get 100 megawatts of power into the grid. For a simple cycle power plant it is likely closer to 300 megawatts in to get 100 megawatts out.

Let's use this Sankey diagram to demonstrate the power of electrification and renewables.

The US uses 24.8 Quads of petroleum inputs in Transportation to produce 5.87 quads of energy services and 21.7 quads of rejected energy.

In a hypothetical world where we power all transportation by electricity, we would only need the electricity to deliver 5.87 quads of usable energy for transportation.  At the current mix of electricity generation shown, (13.3 quads of energy services for electricity for 32 quads of inputs), we would only need about 14 quads of inputs to power electrified transports.  That's a 60% reduction in power inputs for the same energy services.

As an extreme example, let's say we are able to power that transportation with mostly renewables, where ~90% of the energy is usable.  We would only need about 6.5 quads of renewables to power transportation in the US.  That's a 74% reduction in energy inputs to power transportation.

While we won't get all of the way there in either category (electric aviation isn't happening), getting to ~80% renewables is very believable.  Getting mostly electric transport is believable, but more a question of politics than than technical capabilities. 

[NorCal]

https://www.powerengineeringint.com/coal-fired/strategic-development-coal-fired/global-coal-demand-set-to-remain-at-record-high-until-2027-finds-iea/

I don't pretend to know a ton about the energy transition, but are there any educated opinions about whether global energy demand growth is simply going to outstrip our ability to install cleaner energy sources?

I've read a decent amount about it.  I have a lot of concerns about AI and other sources of net-new demand.

Outside of that, I am largely optimistic.  On a global level, sales of EV's are up 25% YoY, even if they're only up 9% YoY in the US (a 9% growth rate for a product in a stagnant market is still a major shift).  Other electrification technologies are taking off. 

Total electricity demand growth will likely be higher than our capacity to install renewables.  But this is like the problem of paying too much taxes.  It's only a problem because a lot of other things are going right. 

An electric car requires roughly 2/3 less energy input than a gas car.  So electrifying it is still a climate win, even if it takes extra time to bring renewables online.  A heat pump with a COP of 3 powered by a gas power plant will still use about 25-30% less natural gas than a typical gas furnace.  Electrification is fundamentally harder than switching electricity fuel sources, and equally important.

The majority of new generation capacity coming online is renewable.  But the pace of electricity demand growth is such that we will probably be adding and extending the life of fossil fueled power generation.  It's not ideal, but we should in no way slow down electrification efforts for it.

We just need to not give a free pass to AI and other data-hungry computing operations. 

[GuitarStv]

We just need to not give a free pass to AI and other data-hungry computing operations.

Yeah, don't forget crypto!

AI wasted 147 TWh in 2023 (https://www.mckinsey.com/featured-insights/sustainable-inclusive-growth/charts/ais-power-binge).  Bitcoin alone wasted 131.95 TWh in 2023 (https://ccaf.io/cbnsi/cbeci/comparisons).  And there are hundreds of other types of crypto currency.  This doesn't even discuss NFTs, which are also massively wasteful of resources.

[Wintergreen78]

https://www.powerengineeringint.com/coal-fired/strategic-development-coal-fired/global-coal-demand-set-to-remain-at-record-high-until-2027-finds-iea/

I don't pretend to know a ton about the energy transition, but are there any educated opinions about whether global energy demand growth is simply going to outstrip our ability to install cleaner energy sources?

I've read a decent amount about it.  I have a lot of concerns about AI and other sources of net-new demand.

Outside of that, I am largely optimistic.  On a global level, sales of EV's are up 25% YoY, even if they're only up 9% YoY in the US (a 9% growth rate for a product in a stagnant market is still a major shift).  Other electrification technologies are taking off. 

Total electricity demand growth will likely be higher than our capacity to install renewables.  But this is like the problem of paying too much taxes.  It's only a problem because a lot of other things are going right. 

An electric car requires roughly 2/3 less energy input than a gas car.  So electrifying it is still a climate win, even if it takes extra time to bring renewables online.  A heat pump with a COP of 3 powered by a gas power plant will still use about 25-30% less natural gas than a typical gas furnace.  Electrification is fundamentally harder than switching electricity fuel sources, and equally important.

The majority of new generation capacity coming online is renewable.  But the pace of electricity demand growth is such that we will probably be adding and extending the life of fossil fueled power generation.  It's not ideal, but we should in no way slow down electrification efforts for it.

We just need to not give a free pass to AI and other data-hungry computing operations.

In the US in 2013 there were 38.2 Qudrillion Btu’s of energy used to produce 12.4 quads of useful electricity. In 2014 32 quads of input were used to produce 13.3 quads of useful electricity.

So, in the US electricity generation is getting more efficient faster than electricity demand is growing.

I agree that AI/crypto/etc. seem like wasteful uses of electricity, and they make transitioning away from fossil fuels harder by increasing total energy demand. But, power generation in the US is a success story. There is still a long way to go, but the US is transitioning to more efficient and lower emitting sources of power at a decent rate.

[NorCal]

https://www.powerengineeringint.com/coal-fired/strategic-development-coal-fired/global-coal-demand-set-to-remain-at-record-high-until-2027-finds-iea/

I don't pretend to know a ton about the energy transition, but are there any educated opinions about whether global energy demand growth is simply going to outstrip our ability to install cleaner energy sources?

I've read a decent amount about it.  I have a lot of concerns about AI and other sources of net-new demand.

Outside of that, I am largely optimistic.  On a global level, sales of EV's are up 25% YoY, even if they're only up 9% YoY in the US (a 9% growth rate for a product in a stagnant market is still a major shift).  Other electrification technologies are taking off. 

Total electricity demand growth will likely be higher than our capacity to install renewables.  But this is like the problem of paying too much taxes.  It's only a problem because a lot of other things are going right. 

An electric car requires roughly 2/3 less energy input than a gas car.  So electrifying it is still a climate win, even if it takes extra time to bring renewables online.  A heat pump with a COP of 3 powered by a gas power plant will still use about 25-30% less natural gas than a typical gas furnace.  Electrification is fundamentally harder than switching electricity fuel sources, and equally important.

The majority of new generation capacity coming online is renewable.  But the pace of electricity demand growth is such that we will probably be adding and extending the life of fossil fueled power generation.  It's not ideal, but we should in no way slow down electrification efforts for it.

We just need to not give a free pass to AI and other data-hungry computing operations.

In the US in 2013 there were 38.2 Qudrillion Btu’s of energy used to produce 12.4 quads of useful electricity. In 2014 32 quads of input were used to produce 13.3 quads of useful electricity.

So, in the US electricity generation is getting more efficient faster than electricity demand is growing.

I agree that AI/crypto/etc. seem like wasteful uses of electricity, and they make transitioning away from fossil fuels harder by increasing total energy demand. But, power generation in the US is a success story. There is still a long way to go, but the US is transitioning to more efficient and lower emitting sources of power at a decent rate.

Those in the utility industry are saying this is rapidly changing.  Overall electricity demand has been stagnant in the US for a long time.  Electricity usage has gotten more efficient as the population has grown, largely offsetting each other.  People managing utilities laughed at the projections about massive electricity growth because they'd heard those projections for decades without seeing evidence of it. 

The realities of growing demand seemed to hit all at once over the last year.  AI is a true massive demand.  And electrification of cars in particular is now at a scale that it's moving the needle.  There are roughly 3.3M EV's on US roads today.  About 1.3M of those were added in the last year.  At current "no one wants EV's" growth rates of ~9%, that 3.3M EV's will still roughly double around 2026 or 2027.  Adding a couple hundred thousand EV's a year to electricity demand was a rounding error.  Adding over a million a year is a structural change to the electricity market. 

[Financial.Velociraptor]

Does anyone know if any of our three letter agencies tracks source of power by fossil/non for crypto and AI users?

I see lots of press about about various players securing long term contracts to renewable energy or doing things like restarting 3 Mile Island nuclear plant.   I'm curious if it is possible those are actually drivers of progress as they are increasing more than proportionally demand for renewables, attracting new investment, and giving the providers economies of scale to make them more competitive?

There is a lot of greenwashing by the oil majors (and a little bit of genuine progress from the European based majors).  Some of them are also clearly fullashit.   I suspect signal to noise ratio is pretty low on these matters.

[NorCal]

Does anyone know if any of our three letter agencies tracks source of power by fossil/non for crypto and AI users?

I see lots of press about about various players securing long term contracts to renewable energy or doing things like restarting 3 Mile Island nuclear plant.   I'm curious if it is possible those are actually drivers of progress as they are increasing more than proportionally demand for renewables, attracting new investment, and giving the providers economies of scale to make them more competitive?

There is a lot of greenwashing by the oil majors (and a little bit of genuine progress from the European based majors).  Some of them are also clearly fullashit.   I suspect signal to noise ratio is pretty low on these matters.

There's lots of data on both electricity consumption and electricity production, but blending them together is mostly considered a useless exercise.  I have no way of knowing whether the electrons powering my computer right now are coming from a wind farm, coal plant, or gas plant.  But we know how much power the computer uses and the energy mix of overall production. 

Good data sources:
-The EIA has LOTS of data.  Probably not exactly what you're looking for though.  https://www.eia.gov/analysis/
-My favorite source for electricity production and emissions is the EPA.  Looking at power generation and emissions in your state or by facility will give you a lot of context about where your electricity and emissions are coming from.  https://www.epa.gov/egrid/data-explorer
-The IEA is probably the best source for where the world's energy sources are headed.  Reading through some of their reports will help you understand the world better.  Both the current state, and the likely direction of progress on emissions reductions.  https://www.iea.org/

[Financial.Velociraptor]

Does anyone know if any of our three letter agencies tracks source of power by fossil/non for crypto and AI users?

I see lots of press about about various players securing long term contracts to renewable energy or doing things like restarting 3 Mile Island nuclear plant.   I'm curious if it is possible those are actually drivers of progress as they are increasing more than proportionally demand for renewables, attracting new investment, and giving the providers economies of scale to make them more competitive?

There is a lot of greenwashing by the oil majors (and a little bit of genuine progress from the European based majors).  Some of them are also clearly fullashit.   I suspect signal to noise ratio is pretty low on these matters.

There's lots of data on both electricity consumption and electricity production, but blending them together is mostly considered a useless exercise.  I have no way of knowing whether the electrons powering my computer right now are coming from a wind farm, coal plant, or gas plant.  But we know how much power the computer uses and the energy mix of overall production. 

Good data sources:
-The EIA has LOTS of data.  Probably not exactly what you're looking for though.  https://www.eia.gov/analysis/
-My favorite source for electricity production and emissions is the EPA.  Looking at power generation and emissions in your state or by facility will give you a lot of context about where your electricity and emissions are coming from.  https://www.epa.gov/egrid/data-explorer
-The IEA is probably the best source for where the world's energy sources are headed.  Reading through some of their reports will help you understand the world better.  Both the current state, and the likely direction of progress on emissions reductions.  https://www.iea.org/

Thanks.

The mix is TX is not so much a concern for me personally as the grid is deregulated for most customers here.  I chose a 100% renewables plan on a three year term.  Price for kwh was only a tenth cent higher.

[NorCal]

Does anyone know if any of our three letter agencies tracks source of power by fossil/non for crypto and AI users?

I see lots of press about about various players securing long term contracts to renewable energy or doing things like restarting 3 Mile Island nuclear plant.   I'm curious if it is possible those are actually drivers of progress as they are increasing more than proportionally demand for renewables, attracting new investment, and giving the providers economies of scale to make them more competitive?

There is a lot of greenwashing by the oil majors (and a little bit of genuine progress from the European based majors).  Some of them are also clearly fullashit.   I suspect signal to noise ratio is pretty low on these matters.

There's lots of data on both electricity consumption and electricity production, but blending them together is mostly considered a useless exercise.  I have no way of knowing whether the electrons powering my computer right now are coming from a wind farm, coal plant, or gas plant.  But we know how much power the computer uses and the energy mix of overall production. 

Good data sources:
-The EIA has LOTS of data.  Probably not exactly what you're looking for though.  https://www.eia.gov/analysis/
-My favorite source for electricity production and emissions is the EPA.  Looking at power generation and emissions in your state or by facility will give you a lot of context about where your electricity and emissions are coming from.  https://www.epa.gov/egrid/data-explorer
-The IEA is probably the best source for where the world's energy sources are headed.  Reading through some of their reports will help you understand the world better.  Both the current state, and the likely direction of progress on emissions reductions.  https://www.iea.org/

Thanks.

The mix is TX is not so much a concern for me personally as the grid is deregulated for most customers here.  I chose a 100% renewables plan on a three year term.  Price for kwh was only a tenth cent higher.

Texas is interesting to look at in that energy explorer.  Texas's electricity generation is the highest in the country by a factor of 2x. 

My quick math shows roughly a third of TX electricity related emissions come from coal, even though coal only makes up 13% of generation. 

Texas's overall emissions/kWh is right around the US average.

Simply shutting WA Parish, Sam Semour, Oak Grove, Limestone, and Market Lake would shave a huge amount off of TX emissions, even if the capacity was replaced with gas. 

[big_owl]

This lets me share some of my favorite charts. Lawrence Livermore National Labs puts out Sankey charts, which show sources for power, how much is used, and how much is lost as waste heat. These are done for the US. I’d be very interested in seeing something similar for the whole world.

https://flowcharts.llnl.gov/commodities/energy

In 2023 the US consumed 93.6 quadrillion Btus, of which 8.2 were from coal, 35.4 from petroleum, 33.4 from natural gas, 2.4 from solar and wind.

In 2013 the numbers were: 97.4 quads total, 18 from coal, 35.1 from petroleum, 26.6 from natural gas, 1.9 from solar and wind.

So, for all the talk about renewables, they are still a tiny portion of the overall picture. The bigger story is that US energy use is actually down over the last 10 years, despite more people, bigger gdp, etc. Energy efficiency works!

Here’s a link where you can see energy/gdp data for other countries. In most “rich” countries, the same pattern is developing.

https://ourworldindata.org/energy-gdp-decoupling

The Stanley diagrams are great, but the visualization does make renewables look smaller than they are.

The energy efficiency of most steam-turbine energy sources are in the 30%-40% range IIRC. It can be higher for combined cycle plants, but those are not particularly common.  Then you add on about 10% grid transmission losses. This is why the “rejected energy” category is so large.  Gas engines on vehicles are similar, with efficiency in the 35%ish range.

Renewables don’t have this problem. You subtract out the 10% transmission losses and the rest is usable energy.

One of my favorite energy commentators (Michael Liebrich) delves into this a lot as the primary energy fallacy. Many who fall into the camp of “the energy transition is too hard” focus on replacing the inputs on the left hand side of that chart.

In reality, we need to replace the energy services on the right hand side of the chart. Most transition technologies produce negligible rejected energy.

I’ll also reiterate my point that the story of global coal use has very little to do with the US.

You are not recalling correctly, the majority of natural gas fired powerplants in the US being installed are combined cycle.  And CCGTs have been achieving >60% efficiency for close to ten years now. 

[alsoknownasDean]

I think it'll be interesting to see what happens in China. They're rolling out solar panels in massive numbers, and are producing most of the world's batteries.

[NorCal]

I think it'll be interesting to see what happens in China. They're rolling out solar panels in massive numbers, and are producing most of the world's batteries.

For the most part China is the energy transition story.  If you look at that IEA paper I posted above, you can see that China consumes 56% of the world's coal.

The scale of what China is doing is mind boggling.  For the major energy transition technologies (solar, onshore wind, offshore wind, EV's, & other electrification) they are the predominant buyer.  I haven't looked at the most recent numbers, but last I looked China makes up greater than 50% of the market for most of those technologies.  In 2023, China installed more offshore wind capacity than already existed in the entire rest of the world.

On the flip side, China is also deploying more coal power plants than anywhere else.  They use coal similar to how the US uses natural gas.  They also have extremely limited transmission between provinces, making each province kind of an electricity island.  It's not particularly efficient.  However, approvals for new coal power plants seems to be declining, and the IEA is predicting roughly flat coal demand in China through 2027. 

Demand continues to fall in the US and EU, although they only make up ~7% of the global market.  While shutting down coal in the US and EU is important for climate, it's small compared to what the rest of the world is doing.  China, India, and other Asian countries are expected to make up 75% of world coal demand by.

Most growth in coal usage through 2027 is in India and the rest of Asia.   

 

[SpareChange]

Texas is interesting to look at in that energy explorer.  Texas's electricity generation is the highest in the country by a factor of 2x. 

My quick math shows roughly a third of TX electricity related emissions come from coal, even though coal only makes up 13% of generation. 

Texas's overall emissions/kWh is right around the US average.

Simply shutting WA Parish, Sam Semour, Oak Grove, Limestone, and Market Lake would shave a huge amount off of TX emissions, even if the capacity was replaced with gas.

I was reading somewhere that about a third of Tx coal plants will be shut down by 2030. A few weeks ago one recieved funding to switch to a solar+battery plant. Tx is ridiculously blessed with solar and wind potential, so the transition will be easier here...I think I read that last fall it overtook CA in utility scale solar to now lead the nation (it already leads in wind by a mile).

 

[GilesMM]

Energy transitions tend to be slow. Coal, as noted, is at record consumption levels despite oil and gas production occurring for over a century. The transition to renewables need to be faster than the transition off coal.

[YttriumNitrate]

I think it'll be interesting to see what happens in China. They're rolling out solar panels in massive numbers, and are producing most of the world's batteries.

From what I've read, the new coal power generation brought online by China just in the last year is equivalent to a quarter of the entirety of the USA's coal power generation.

[SpareChange]

Energy transitions tend to be slow. Coal, as noted, is at record consumption levels despite oil and gas production occurring for over a century. The transition to renewables need to be faster than the transition off coal.

For sure. I'll have to look up estimates of the carbon budget for 2 and 3 degrees C of warming. Don't know how much leeway we have to keep going like this.

[rocketpj]

Well, to use a painfully current analogy, I'll bet demand for energy is pretty low in certain parts of Los Angeles right now.

Humans, or more accurately the billionaires who recently bought the United States and a variety of other governments have decided that the way to reduce energy consumption in the future is to maximize their profits now, until so much damage is done that millions or billions die and stop using energy.

How they imagine the survivors won't be hunting billionaires for sport is beyond me, but that's where we stand right now.  Louis XVI also imagined that there would be zero chance of the peasants pushing back on him either.

[SpareChange]

Saw a related headline on Fortune saying that emissions due to ChatGPT are about the same as 260 flights from NYC to London every month.

[Ron Scott]

Saw a related headline on Fortune saying that emissions due to ChatGPT are about the same as 260 flights from NYC to London every month.

ChatGPT Responds…abbreviated:

ME: Saw a headline in Fortune magazine saying that emissions due to ChatGPT are about the same as 260 flights from NYC to London every month. What would be the related comparison to Google searches?

ChatGPT: ChatGPT’s emissions are roughly equivalent to 520 million Google searches per month. Google processes ~8.5 billion searches per day, so ChatGPT’s monthly emissions are about 2% of Google’s daily search emissions.

While the Fortune magazine headline may seem significant, the scale of Google’s operations is much larger.

[Ron Scott]

I think it'll be interesting to see what happens in China. They're rolling out solar panels in massive numbers, and are producing most of the world's batteries.

For the most part China is the energy transition story.  If you look at that IEA paper I posted above, you can see that China consumes 56% of the world's coal.

The scale of what China is doing is mind boggling.  For the major energy transition technologies (solar, onshore wind, offshore wind, EV's, & other electrification) they are the predominant buyer.  I haven't looked at the most recent numbers, but last I looked China makes up greater than 50% of the market for most of those technologies.  In 2023, China installed more offshore wind capacity than already existed in the entire rest of the world.

On the flip side, China is also deploying more coal power plants than anywhere else.  They use coal similar to how the US uses natural gas.  They also have extremely limited transmission between provinces, making each province kind of an electricity island.  It's not particularly efficient.  However, approvals for new coal power plants seems to be declining, and the IEA is predicting roughly flat coal demand in China through 2027. 

Demand continues to fall in the US and EU, although they only make up ~7% of the global market.  While shutting down coal in the US and EU is important for climate, it's small compared to what the rest of the world is doing.  China, India, and other Asian countries are expected to make up 75% of world coal demand by.

Most growth in coal usage through 2027 is in India and the rest of Asia.

If that trend in China continues and if China’s economy continues to slow and degrade as it moves more communist/autocratic and further away from rule-of-law and market economics, the environment will be different indeed!

[bacchi]

ChatGPT: ChatGPT’s emissions are roughly equivalent to 520 million Google searches per month. Google processes ~8.5 billion searches per day, so ChatGPT’s monthly emissions are about 2% of Google’s daily search emissions.

Something's off with that math.

[GuitarStv]

ChatGPT: ChatGPT’s emissions are roughly equivalent to 520 million Google searches per month. Google processes ~8.5 billion searches per day, so ChatGPT’s monthly emissions are about 2% of Google’s daily search emissions.

Something's off with that math.

Yeah.  52/8,50 = 6.12%

Does the fact that all that ChatGPT energy is going to churning out bad/wrong answers to questions that then need to be checked with Google to find the right response factor in at all?  :P

[Ron Scott]

ChatGPT: ChatGPT’s emissions are roughly equivalent to 520 million Google searches per month. Google processes ~8.5 billion searches per day, so ChatGPT’s monthly emissions are about 2% of Google’s daily search emissions.

Something's off with that math.

Yeah.  52/8,50 = 6.12%

Does the fact that all that ChatGPT energy is going to churning out bad/wrong answers to questions that then need to be checked with Google to find the right response factor in at all?  :P

You guys usually seem to be brighter than me but I did the math again and got ~0.2%—which is far less than either you or ChatGPT.

If anybody knows how to do arithmetic, please respond. My calc screenprint is attached.

[GuitarStv]

ChatGPT wrote that it compared it's monthly emissions to Google’s daily emissions:

ChatGPT’s emissions are roughly equivalent to 520 million Google searches per month. Google processes ~8.5 billion searches per day, so ChatGPT’s monthly emissions are about 2% of Google’s daily search emissions.

- ChatGPT’s monthly emissions are 520 million Google searches.
- Google's daily emissions are ~8.5 billion Google searches.

52/850 = 6ish percent for a month/day comparison as stated (which granted, is a dumb comparison . . . but it's what ChatGPT said it was doing).



I think that you're trying to compare months to months (which isn't what ChatGPT said it was doing but makes a lot more sense).  Hence your different answer.  So let me try doing a variation of the apples to apples comparison with a yearly approach:

520,000,000 per month = 6,240,000,000 searches per year
8,500,000,000 per day = 3,102,500,000,000 searches per year

divide both sides by a million to get
6,240/3,102,500

or about 0.2%, which matches up pretty closely with your 30 day per month approximation and makes sense to me.


ChatGPT said it was doing an apples to oranges comparison (monthly to daily usage) then tried to do a like for like comparison in calculation and was off by a magnitude of ten.  Lotta energy for a pretty shit answer.  :P

[BC_Goldman]

I wonder if chatgpt can estimate how much energy it used to answer that question?

[SpareChange]

ChatGPT said it was doing an apples to oranges comparison (monthly to daily usage) then tried to do a like for like comparison in calculation and was off by a magnitude of ten.  Lotta energy for a pretty shit answer.  :P

Whooooooshhh. Another flight from LaGuardia to Heathrow with no passengers. 

[Ron Scott]

520,000,000 per month = 6,240,000,000 searches per year
8,500,000,000 per day = 3,102,500,000,000 searches per year

divide both sides by a million to get
6,240/3,102,500

or about 0.2%, which matches up pretty closely with your 30 day per month approximation and makes sense to me.

OK.

So—bottom line:

1. Google Search is currently using 500 times as much electricity as ChatGPT. (Check my math…)
2. Both products get it wrong too much, but Google Search makes you jump over a page worth of ads before you figure out you there’s nothing there
3. Google Search is a monopoly that’s been allowed to exist as one for decades

Well Ollie, that’s a fine mess you’ve gotten us into!

[GuitarStv]

520,000,000 per month = 6,240,000,000 searches per year
8,500,000,000 per day = 3,102,500,000,000 searches per year

divide both sides by a million to get
6,240/3,102,500

or about 0.2%, which matches up pretty closely with your 30 day per month approximation and makes sense to me.

OK.

So—bottom line:

1. Google Search is currently using 500 times as much electricity as ChatGPT. (Check my math…)
2. Both products get it wrong too much, but Google Search makes you jump over a page worth of ads before you figure out you there’s nothing there
3. Google Search is a monopoly that’s been allowed to exist as one for decades

Well Ollie, that’s a fine mess you’ve gotten us into!

Well, you're comparing asking an AI a question, with asking an AI a question AND getting internet search results.

Today, every Google search makes use of Google's own AI to generate responses that are then put to the top of the list.  I'd be interested to see what the energy costs looked like if you removed all that AI energy waste from the equation (like a comparison with duck duck go's responses for example).

[Just Joe]

s it possible that search engine cache results so they don't need to ask AI the same questions over and over?