We haven't heard much climate BS from AOC in a while, have we? And she was one of the loudest climate alarmists. In fact, I think we only have 6 years left on her 12-year timeline so I'm not sure they'll have enough time for AI to need any of those new energy sources anyway.

DallasAg 94 said:

---

Translation: "The amount of money we can make from AI greatly exceeds our fantasies from Climate Change. So we're moving to our next big grift."

Divest from real estate and alternative fuels. AI is a huge money making boondoggle.

---

I timed divesting my green energy investment well and made a bundle.

The lack of demand to curtail energy expenditure to advance AI is one of many "tells" that the whole climate crisis is just a power/money grab.

The knots the alarmists will have to tie themselves in is going to be interesting to watch. In this case it's unfortunate in my Luddite perspective because I don't personally see exponential growth in AI (faster than we can effectively manage) as being a net good for mankind or the planet we steward

javajaws said:

---

Lets just hook up all the unemployed people sitting at home that collect unemployment and play video games to the Matrix and turn them into batteries.


Seriously though - I'm not sure I would put much faith in these estimates just yet. All technologies mature, and as such they generally get more energy efficient per measure of value they produce. Energy consumption will obviously go up over the next 5-10 years, but I think eventually it will come back down as well. The current approach to "AI" (which isn't AGI) is very much a brute force approach. So long as this is the "state of the art" consumption will increase as usage/demand increases. But eventually we will develop better technology that consumes less energy.

---

Great post. One of the key questions is how many winners their will be in the coming shake out?

I would bet on google and microsoft. With better odds, I would take a flyer on X w/Grok.
Obviously, if there are fewer entrants the "powers that be" will not have to be worried as much about smaller, more nimble competitors.

Fewer competitors, less racing to be first, means AI's power use will be smaller. Still will be a lot, though,

Over_ed said:

---

javajaws said:

---

Lets just hook up all the unemployed people sitting at home that collect unemployment and play video games to the Matrix and turn them into batteries.


Seriously though - I'm not sure I would put much faith in these estimates just yet. All technologies mature, and as such they generally get more energy efficient per measure of value they produce. Energy consumption will obviously go up over the next 5-10 years, but I think eventually it will come back down as well. The current approach to "AI" (which isn't AGI) is very much a brute force approach. So long as this is the "state of the art" consumption will increase as usage/demand increases. But eventually we will develop better technology that consumes less energy.

---

Great post. One of the key questions is how many winners their will be in the coming shake out?

I would bet on google and microsoft. With better odds, I would take a flyer on X w/Grok.
Obviously, if there are fewer entrants the "powers that be" will not have to be worried as much about smaller, more nimble competitors.

Fewer competitors, less racing to be first, means AI's power use will be smaller. Still will be a lot, though,

---

The AI players are in a scaling race to gobble up market share and outlast their competitors. They aren't thinking one bit amount a bubble or slowing down spending or building of data centers. They're all going as fast as they can, throwing every dollar they can at it, and they won't let lack of energy get in the way. They'll go for whatever gets them power the fastest, which is why there is 4 year backlog on gas turbine orders.

Now Microsoft, Google, and Meta also want to be net-zero carbon emissions and are also exploring zero emission power and CO2 removal strategies at the same time. But speed comes first so the green energy and CO2 removal investments will have to wait while the AI data center race is taking place.

The climate hoaxers will just shift their energy back to nuclear, so we still won't have any relief from the hysteria. It won't matter that places like France have for decades safely used nuclear to supply a good percentage of their grid needs - it will still be two or three times more expensive here than it should be simply because of all the weak-minded legislators and regulators listening to the new grift of the month.

Maroon Dawn said:

---

Rex Racer said:

---

Yep. I knew as soon as AI took off that nuclear energy was about to make a huge comeback, too. It's the only way. There's too much money to be made.

---

This. I'm betting those changes they made after 3 mile island get revisited so it's no longer impossibly expensive to build nuclear

---

They're building a mini nuke at RELLIS. This the way.

Texarkana said:

---

Exactly right - the Tech companies have waved their magic wand and nuclear power plants and gas power plants are no longer bad for the environment.

---

Time to load up on Exxon-Mobile and Energy Transfer.

javajaws said:

---

Lets just hook up all the unemployed people sitting at home that collect unemployment and play video games to the Matrix and turn them into batteries.


Seriously though - I'm not sure I would put much faith in these estimates just yet. All technologies mature, and as such they generally get more energy efficient per measure of value they produce. Energy consumption will obviously go up over the next 5-10 years, but I think eventually it will come back down as well. The current approach to "AI" (which isn't AGI) is very much a brute force approach. So long as this is the "state of the art" consumption will increase as usage/demand increases. But eventually we will develop better technology that consumes less energy.

---

Great post and I think you are right. I asked AI to list companies that are in the (energy efficiency) space for potential investment. Here is what it gave me:

Groq: Develops specialized AI inference chips that prioritize speed and energy efficiency, aiming to provide cost-effective alternatives to traditional GPUs for running AI models with lower power draw.

research.aimultiple.com +1

Tenstorrent: Focuses on scalable AI hardware with a emphasis on power-efficient architectures, including chips designed for high-performance computing that consume less energy than conventional solutions.

yolegroup.com

Celestial AI: Innovates in optical interconnects and photonic computing for AI, which can drastically cut energy use in data transfer within AI systems compared to electronic methods.

yolegroup.com

SambaNova Systems: Builds full-stack AI platforms with custom chips optimized for efficiency, enabling faster training and inference while reducing overall power requirements.

yolegroup.com

Hailo: Specializes in edge AI processors that deliver high performance with ultra-low power consumption, ideal for devices where energy efficiency is critical.

yolegroup.com

Ambiq: Leads in ultra-low-power semiconductors for edge AI, enabling battery-powered devices to run sophisticated AI tasks with minimal energy.

ambiq.com

SiMa.ai: Designs chips and software for low-power AI at the edge, combining hardware and algorithms to minimize energy use in real-time applications like vision processing.

datacenterdynamics.com

Untether AI: Creates energy-efficient AI accelerators that decouple compute from memory, reducing power consumption for inference workloads.

research.aimultiple.com

BitEnergy AI: Develops algorithms like L-Mul that replace power-hungry operations in AI models, claiming up to 95% reductions in energy for certain computations.

reddit.com

Emerald AI: Builds software to dynamically manage and optimize energy use in AI data centers, helping adjust consumption to avoid grid strain.

virginiabusiness.com

DeepMind (Alphabet/Google): Advances AI efficiency through techniques like data center cooling optimizations and model compression, achieving significant energy savings in large-scale deployments.

aimagazine.com

Intel: Invests in neuromorphic computing (e.g., Loihi chips) that mimic brain-like efficiency, alongside general low-power AI hardware initiatives.

quora.com +1

Starcloud: Explores space-based data centers to leverage natural cooling and solar power, potentially slashing energy costs for AI computations by a factor of 10.

Any suggestions, texags?

Great thought. I am not in this space enough to guess winners and losers. The value of these energy-efficient AI companies, in part, depends on what happens in the "primary" shakeout, which will start next year.

Right now the prevailing thought is AI companies don't care how much power we have to make/buy. "We need to win."

But the next cycle will focus on performance per watt.

I've been saying all along that the current AI tech isn't scalable and power consumption is the reason why. To think that there is any real money to be made is just foolish. The power requirements are too great, but let them pour their money into power generation anyway. We will all benefit from this eventually.

YouBet said:

---

DallasAg 94 said:

---

Translation: "The amount of money we can make from AI greatly exceeds our fantasies from Climate Change. So we're moving to our next big grift."

Divest from real estate and alternative fuels. AI is a huge money making boondoggle.

---

I timed divesting my green energy investment well and made a bundle.

---

WWR?

Ironically, a hard pivot to nuclear energy will do more for the environment than solar or wind ever did.

Bird Poo said:

---

quora.com +1

Starcloud: Explores space-based data centers to leverage natural cooling and solar power, potentially slashing energy costs for AI computations by a factor of 10.

---

This right here. When Starship becomes fully operational in a couple years it's going to be way cheaper to put up orbital data centers than it's is to build and maintain terrestrial ones.

Will land prices for prime data center tracts be lower than the price of a launch?

Are solar panels less expensive than ongoing power bills? Did you have to improve/extend the transmission lines?

No property taxes in orbit.

Water/energy bills for evaporative cooling? One time cost in orbit.

The up/down link will be the only argument against orbit data centers/AI.

deddog said:

---

YouBet said:

---

DallasAg 94 said:

---

Translation: "The amount of money we can make from AI greatly exceeds our fantasies from Climate Change. So we're moving to our next big grift."

Divest from real estate and alternative fuels. AI is a huge money making boondoggle.

---

I timed divesting my green energy investment well and made a bundle.

---

WWR?

---

Ha.

PLUG.

The green energy grift will continue until it is no longer possible for people to make money off of all the rubes who believe rainbows and unicorn farts can power the country.

Rex Racer said:

---

Yep. I knew as soon as AI took off that nuclear energy was about to make a huge comeback, too. It's the only way. There's too much money to be made.

---

Small nuke plants at the datacenter. This is the way!

Rex Racer said:

---

Yep. I knew as soon as AI took off that nuclear energy was about to make a huge comeback, too. It's the only way. There's too much money to be made.

---

Hope your right but the coal and O&G lobby will be fighting nuke expansion

The current server farms used to train AI systems and run inference are very inefficient.

Do you need powerplants rivalling state power plants to power and train a human brain over several decades? No. Not anywhere close. There needs to be R&D into making this way more efficient.

TexAgs91 said:

---

The current server farms used to train AI systems and run inference are very inefficient.

Do you need powerplants rivalling state power plants to power and train a human brain over several decades? No. Not anywhere close. There needs to be R&D into making this way more efficient.

---

Actually you do, or you will very soon. The compute power needed to train the latest AI models is growing faster than Moore's law can provide the efficency gains needed to keep power requirements steady.

Sq 17 said:

---

Rex Racer said:

---

Yep. I knew as soon as AI took off that nuclear energy was about to make a huge comeback, too. It's the only way. There's too much money to be made.

---

Hope your right but the coal and O&G lobby will be fighting nuke expansion

---

I don't think they are going to win this one. Ultimately, conservatives want nuke power because it's the most efficient and cleanest and the left will want nuke because it's the lesser of evils in their mind from an environmental impact standpoint.

Bird Poo said:

---

javajaws said:

---

Lets just hook up all the unemployed people sitting at home that collect unemployment and play video games to the Matrix and turn them into batteries.


Seriously though - I'm not sure I would put much faith in these estimates just yet. All technologies mature, and as such they generally get more energy efficient per measure of value they produce. Energy consumption will obviously go up over the next 5-10 years, but I think eventually it will come back down as well. The current approach to "AI" (which isn't AGI) is very much a brute force approach. So long as this is the "state of the art" consumption will increase as usage/demand increases. But eventually we will develop better technology that consumes less energy.

---

Great post and I think you are right. I asked AI to list companies that are in the (energy efficiency) space for potential investment. Here is what it gave me:

Groq: Develops specialized AI inference chips that prioritize speed and energy efficiency, aiming to provide cost-effective alternatives to traditional GPUs for running AI models with lower power draw.

research.aimultiple.com +1

Tenstorrent: Focuses on scalable AI hardware with a emphasis on power-efficient architectures, including chips designed for high-performance computing that consume less energy than conventional solutions.

yolegroup.com

Celestial AI: Innovates in optical interconnects and photonic computing for AI, which can drastically cut energy use in data transfer within AI systems compared to electronic methods.

yolegroup.com

SambaNova Systems: Builds full-stack AI platforms with custom chips optimized for efficiency, enabling faster training and inference while reducing overall power requirements.

yolegroup.com

Hailo: Specializes in edge AI processors that deliver high performance with ultra-low power consumption, ideal for devices where energy efficiency is critical.

yolegroup.com

Ambiq: Leads in ultra-low-power semiconductors for edge AI, enabling battery-powered devices to run sophisticated AI tasks with minimal energy.

ambiq.com

SiMa.ai: Designs chips and software for low-power AI at the edge, combining hardware and algorithms to minimize energy use in real-time applications like vision processing.

datacenterdynamics.com

Untether AI: Creates energy-efficient AI accelerators that decouple compute from memory, reducing power consumption for inference workloads.

research.aimultiple.com

BitEnergy AI: Develops algorithms like L-Mul that replace power-hungry operations in AI models, claiming up to 95% reductions in energy for certain computations.

reddit.com

Emerald AI: Builds software to dynamically manage and optimize energy use in AI data centers, helping adjust consumption to avoid grid strain.

virginiabusiness.com

DeepMind (Alphabet/Google): Advances AI efficiency through techniques like data center cooling optimizations and model compression, achieving significant energy savings in large-scale deployments.

aimagazine.com

Intel: Invests in neuromorphic computing (e.g., Loihi chips) that mimic brain-like efficiency, alongside general low-power AI hardware initiatives.

quora.com +1

Starcloud: Explores space-based data centers to leverage natural cooling and solar power, potentially slashing energy costs for AI computations by a factor of 10.

Any suggestions, texags?

---

MSAI in Houston/Beaumont.

amercer said:

---

The US is on pace to add record amounts of green energy to the grid this year and over the next few years.

Data centers don't think that some elections are woke and others are patriotic.

---

Those projects have been papered up and in the queue for years. Lots of long lead times on these projects.

Meanwhile..

Global Warming and the "impossible" Radiation Imbalance

Quote:

---

This analysis demonstrates that a global warming scenario driven solely by greenhouse gases (GHGs) is inconsistent with more than 20 years of observations from space and of Ocean Heat Content. The standard anthropogenic global warming (AGW) hypothesis, which attributes all observed warming to rising GHG concentrations, particularly CO2, cannot explain the observed trends. Instead, natural factors, especially long-term increase in incoming solar radiation, appear to play a significant and likely dominant role in global warming since the mid-1970s.

The observed increase in incoming solar radiation cannot be accounted for by the possible anthropogenic side effects of Albedo- and Cloud-feedback. All evidence points to the conclusion that this "natural" forcing with a trend of about 0.035 W/m2 /year is equal to, or even exceeds the greenhouse gas related forcing of about 0.019 W/m2 /year. Based on these values, only 1/3rd of the observed temperature trend can be of anthropogenic origin. The remaining 2/3rd must stem from natural changes in our climate system, or more broadly, in our entire Earth' thermal system.

Moreover, the observed increase in Earth's radiation imbalance appears to be largely unrelated to GHGs. Instead, it correlates strongly with natural processes driving increased incoming solar radiation. Claims of "acceleration" in the radiation imbalance due to GHG emissions are not supported by the trend in accurately measured GHG concentrations. If any acceleration in global warming is occurring, it is almost certainly driven by the increasing flux of solar energyan inherently natural phenomenon not induced by greenhouse gases.

In summary, this analysis challenges the notion that GHGs are the primary drivers of recent climate change. It underscores the importance of accounting for natural variability, especially in solar input, when interpreting warming trends and evaluating climate models.

---

And....

Antarctic Amundsen-Scott Station Sees Coldest October in 44 Years…Mainstream Media Silent!

Quote:

---

On October 15, the Amundsen-Scott South Pole Station registered an astonishing temperature of minus 61.3 degrees Celsius and it isn't even winter there. It's springtime and temperatures should be on the rise.

Coldest October since 1981

According to Report 24, the numbers are clear: It was the coldest October measured at the station since 1981.

This extreme cold is not an isolated event. As the article points out, even CNN reported in 2021 that the continent had experienced its coldest winter since records began.

The data from stations like Amundsen-Scott, Vostok, and Dome C show that instead of a linear, CO-driven heating trend, the South Pole is dominated by naturally occurring, extreme temperature fluctuations, including pronounced cold snaps.

---

This cold Antarctica (is that redundant?) story is an example why they went away from global warming to climate change.

Always right, plus more things to publish and get governmental grants.

Shiner, the story before is confirmation of what many of us on F16 believed for a very long time. The sun, who would have guessed?


Great posts, btw.

Over_ed said:

---

This cold Antarctica (is that redundant?) story is an example why they went away from global warming to climate change.

Always right, plus more things to publish and get governmental grants.

Shiner, the story before is confirmation of what many of us on F16 believed for a very long time. The sun, who would have guessed?


Great posts, btw.

---

An unfalsifiable hypothesis is an unscientific one.

For those that think AI demand for electricity will end renewable power, think again. It will drive up electricity demand and prices making them more economical and potentially without any subsidies. AI makes it economical for adding essentially all power sources especially nuclear. Hopefully SMR nuclear power plants are about to be deployed in large numbers.

PS. This isn't me saying I support the climate change agenda. Just the capitalist agenda of making money.

But if that's true, why are these tech companies pursuing nuclear instead of solar and wind?

Kansas Kid said:

---

For those that think AI demand for electricity will end renewable power, think again. It will drive up electricity demand and prices making them more economical and potentially without any subsidies. AI makes it economical for adding essentially all power sources especially nuclear. Hopefully SMR nuclear power plants are about to be deployed in large numbers.

PS. This isn't me saying I support the climate change agenda. Just the capitalist agenda of making money.

---

I do think there are use cases for renewable energy but you have to match the geography and climate to those solutions and not shoe horn green energy into a spot where it's not compatible. It can certainly be a supplemental energy source that helps overall production, and getting the left to acknowledge that practical compromise would help this country so much.

ShinerAggie said:

---

But if that's true, why are these tech companies pursuing nuclear instead of solar and wind?

---

Because if you are going to try to power a massive data center with one single source, nuclear is your only real option especially if you still believe in climate change like a lot of tech companies and their customers still believe since nuclear is the ultimate 0 CO2 emission source. .

My point is the widely held view being supported by high prices in most of the country that electricity prices are going to be higher due to increasing electricity demand. Higher prices help incentivize all energy sources including solar and wind.

Many posters are saying this will end renewables which doesn't make economic sense and is the opposite of the original article.

Kansas Kid said:

---

ShinerAggie said:

---

But if that's true, why are these tech companies pursuing nuclear instead of solar and wind?

---

Because if you are going to try to power a massive data center with one single source, nuclear is your only real option especially if you still believe in climate change like a lot of tech companies and their customers still believe since nuclear is the ultimate 0 CO2 emission source. .

My point is the widely held view being supported by high prices in most of the country that electricity prices are going to be higher due to increasing electricity demand. Higher prices help incentivize all energy sources including solar and wind.

Many posters are saying this will end renewables which doesn't make economic sense and is the opposite of the original article.

---

The data center operators do want to minimize theirCO2 emission. They are primarily exploring looking at nuclear, geothermal, biomass, and natural gas with carbon capture. They're chasing all of them because one source just isn't enough power to meet their needs. Solar and wind are not seriously considered for data centers because of the required land area and variability.

More climate change policy failure on the horizon in the UK.

https://www.wsj.com/opinion/keir-starmer-net-zero-climate-britain-labour-party-04c069aa?st=UxGMNA&reflink=article_copyURL_share

Quote:

---

Prime Minister Keir Starmer is preparing to ditch his promise to shift 95% of Britain's electricity generation to renewables by 2030, the Guardian newspaper reported this week. Mr. Starmer's office denied it, but every other word and deed from his government suggests the report is correct.

The political stakes are rising. Not long ago climate was a bipartisan preoccupation for Mr. Starmer's Labour Party and the Conservatives; a Tory government introduced a legally binding net-zero target. No longer, as Tory leader Kemi Badenoch now promises to scrap the whole idea. Nigel Farage's insurgent Reform UK party is winning over Labour voters by heaping scorn on net zero.

---

Costs are rising because they are having to subsidize the green energy myth.

Quote:

---

The cap for household energy prices increased 2% this autumn, even though the regulator noted that wholesale energy costs have fallen recently. One line item that keeps growing is subsidies for power generation, which have exploded in absolute terms and as a share of a household's total electricity billto about 20% today from 8.5% in 2015.

Meanwhile, the government's main vehicle for subsidizing renewables is a "contract for difference" program that guarantees generators a minimum return. No one can say how expensive those subsidies will be since the subsidy increases as energy prices fall.

---

And because they have to offset when mythical green energy doesn't work because it's not 24/7 like traditional power sources.

Quote:

---

Consumers also are paying more owing to the growing cost of building new power lines to connect wind and solar, and escalating "balancing costs" to supply electricity when intermittent renewable sources aren't generating.

---

YouBet said:

---

More climate change policy failure on the horizon in the UK.

https://www.wsj.com/opinion/keir-starmer-net-zero-climate-britain-labour-party-04c069aa?st=UxGMNA&reflink=article_copyURL_share

Quote:

---

Prime Minister Keir Starmer is preparing to ditch his promise to shift 95% of Britain's electricity generation to renewables by 2030, the Guardian newspaper reported this week. Mr. Starmer's office denied it, but every other word and deed from his government suggests the report is correct.

The political stakes are rising. Not long ago climate was a bipartisan preoccupation for Mr. Starmer's Labour Party and the Conservatives; a Tory government introduced a legally binding net-zero target. No longer, as Tory leader Kemi Badenoch now promises to scrap the whole idea. Nigel Farage's insurgent Reform UK party is winning over Labour voters by heaping scorn on net zero.

---

Costs are rising because they are having to subsidize the green energy myth.

Quote:

---

The cap for household energy prices increased 2% this autumn, even though the regulator noted that wholesale energy costs have fallen recently. One line item that keeps growing is subsidies for power generation, which have exploded in absolute terms and as a share of a household's total electricity billto about 20% today from 8.5% in 2015.

Meanwhile, the government's main vehicle for subsidizing renewables is a "contract for difference" program that guarantees generators a minimum return. No one can say how expensive those subsidies will be since the subsidy increases as energy prices fall.

---

And because they have to offset when mythical green energy doesn't work because it's not 24/7 like traditional power sources.

Quote:

---

Consumers also are paying more owing to the growing cost of building new power lines to connect wind and solar, and escalating "balancing costs" to supply electricity when intermittent renewable sources aren't generating.

---

---

Saw this as well. Governments realizing this forces economic hardships that their citizens can't afford to accept. And from an energy perspective the UK is much better off than Germany. France's saving grace are their nuke generators.