@@segasys1339peaker plants can run for days on a small footprint with a natgas pipe. An equivalent solar/Wind peaker plant needs a lot more square footage for both solar/wind generation and the battery array. So now factor in the cost of land and tax rate. And we are only talking about peaker plants. Now let's talk about baseload on the grid. Most grids are 50% base load (meaning electrical usage that is always present 24/7). The amount of square kilometers needed to match say the nuclear fleet in Ontario would be ridiculous. We already have wind along almost the entire Great lakes coastline in Southern Ontario and wind only delivers 1.2% power needed in Ontario. Meanwhile, three locations (Darlington, Pickering, Bruce) deliver 65% to 70% of what we need (and nuclear power is baseload power). The problem with solar and wind is POWER DENSITY. Modern society needs POWER DENSE sources. If you introduced smaller power dense, weather dependent energy sources, you have to seriously OVERBUILD both the generation and storage in area to compensate for the lack of density and on-demand dispatch capacity of nuclear. Germany has tried it. They overbuilt Solar and Wind at 200% of PEAK use nameplate capacity at a cost of 2 Trillion Euro, only to get 49% actual and they are still burning gas and coal, and they have one of the highest electricity rates in Europe. Solar and Wind are not the cheapest when you add in the cost of all the necessary mitigation because of the nature of solar and wind. Solar and Wind DO NOT SCALE WELL. Fine for a small home, not good for a large grid.
And this is why this episode is appropriately called energy crisis. Germany has one because of the way they built their grid. When the commentator says "deindustrialization" he means large industry leaving Germany because of electrical costs AND the cost of LNG as A PHYSICAL MATERIAL OF PRODUCTION, because of the way the German government pushed renewables and natgas hard but banned nuclear. BASF employs somewhere on the scale of 200,000 people. They are actually considering moving out of Germany to the US to continue their chemical plant operations. The amount of oil and gas we use just to make fertilizers, plastics, lubricants, manmade fibres means oil and gas are going nowhere. We can't be dumb about this and ignorant of the scale and numbers involved. If you are a cold climate, heavy industry country, you need nuclear with some nat gas for baseload and peak for generation, and still need oil and gas as materials to make stuff. There is no renewables utopia. We need to find sweet spots if an energy mix and more of those sweet spots will reveal themselves once tech is mature, is cheap, and can scale.
Great listen! Fascinating to hear someone that's focusing on the 'other side' of the doom & gloom needle that so many others have us threading first. Said others are probably correct, but it gets stale after a while. Something bringing forth projections today of what could lay on the other side of said threading is a breath of fresh air in an increasingly stagnant infosphere of talking heads.
Who is this guy ? Why have I never heard of him ? No crying over spilt milk - we only move forward. Thank you, and look forward to future inputs from this huge brain and heart.
I have absolutely no concept of how many square feet are contained in an average skyscraper so an analogy like this just further obfuscates the actual size of these data centers for me.
Working in the Datacenter industry, I can tell you the only limit to growth is the amount of power utilities are willing to provide and they have been very stingy lately.
There is an issue with materials to build the centre as well. I worked on the construction side of Data centres and the amount of material for the build was/is massive from concrete to metal especially for cooling systems and electrics. There will be issues there too.
@Ln-cq8zu As far as I know, there are no supply constraints on concrete, steel, and whatnot. Nor for copper wire, cooling, and electrical equipment. The only exception was during COVID, where certain electrical equipment had lead times as high as 18 months. The contraints now are all on electrical power capacity on the local utilities.
He says there is no energy transition, this is incorrect. Collectively we have decided to take two steps back before we can continue forward. So-called "renewables' are the two steps back and we appear to need a few more before the real transition can begin. Cheap low-pressure/high-temperature nuclear will replace fossil fuels but only when the cost of fossil fuels forces it to happen.
I really enjoyed this episode! It was interesting learning about the future of computer energy consumption, zeta bites and the AI impact on medicine. I am a pathologist. It will be amazing when an AI companion is at hand while I am reading digitally scanned microscope slides for a tissue diagnosis in the future.
Mark P Mills is an excellent guest... regarding aviation 10 - 15 years ago there were very few Chinese holiday travellers now there are 120 million per year So computing network lanes and bytes are increasing 10-fold that rate. There are no solutions only trade-offs.
"We are going to be forced into a more reality-based power posture" - the smartest thing I have heard said about energy since the green revolution started. I like efficiency, and I like wind, solar, and storage when they are able to achieve a favorable lifetime, total cost of ownership / value of power produced. Many green power producers will never create more energy than the value of the energy and labor required to build and maintain their installation. We collectively really need to figure out how to make cost-effective green energy and storage so we can save our fossil fuels for future generations. It will be painful if we burn all of our oil with an ROI above 3/1 in a 200-year period and then struggle to fuel transportation and agriculture for the next 10,000 years. Think ahead beyond your death to the future of our children and grandchildren. This petroleum constrained future is not that far away.
What in the heck are we dealing with if we have a world with a robust nuclear presence for grids and big data while commensurately we have a dearth of fossil fuels required for global transportation and manufacturing and farming? A series of re-localized economies that rely heavily on rail and semi-local farming that still have access to the internet?
The question is not so much how to power something Application Centers or the coming based AI computing but how efficiently to use the energy. In this coming electrical power bonanza, as long as the heat output of the data center can be piped (via water) an stored for use, that will increase the usefulness of the computing. Regrettably though, this residual heat ( 99.9% of all energy going to the data centers ), is no higher then 40-45C (for best computing conditions ), and it can be used only for direct winter heating. If only there could be a "cold steam low-temp" turbine (there are some encouraging independent research in that direction) that can take that low temp fluid and extract energy from it ( with some efficiency ), and concentrated it again, in form of electricity, the energy efficiency of the data centers can be vastly improved, from thermodynamic point of view, as they will not only be consuming but also "producing" energy.
The U.K’s last coal fired power station closes in September in the insane dash to “net zero” . That is 2000 megawatts taken out of the National Grid. This is at a time when demand for energy is on a huge increase, where is all our electricity going to come from? Norway? France? A few windmills and even fewer solar panels?
Solar panels wiith average yearly capacity factor in the single digits and off shore wind turbines with much higher capacity factor, yes, but the size of the statue of liberty with all that concrete, Steel, aluminum, plastics, etc., not to mention the high voltage transmission lines to tie them into the grid.
Any "hype" finishes in one way or another with a crisis. But the crisis is not the death of the hype. It's a necessary phase for moving post the hype. AI will put pressure on energy production and push GAFAM to increase production of their own renewable energy. Nuclear will remain "anecdotic" with a share of around 10-15% of the energy production mix.
It is quite the narrative change to go from green energy / net zero to adding 25% plus fossil fueled energy protection to the grid just for AI. Is the juice really worth the squeeze with AI to use 25% more electricity than our current total production? Kinda reminds me of "the block chain." Is it really worth the resource use we have invested. I say mostly not.
Video doesn't need low latency if it is for consumption only, like Netflix movies or youtube or iCloud. Low latency is crucial only for interactive use cases. On-line gaming and mostly stock trading, which fuels low latency connections. Starlink, commucating through vacuum beats optical cabels on long distances due to a faster speed of light in a vacuum.
ai is a huge problems, and so are phones. but also medicine, we must not get tech into medicine. health is done preventively, not after the fact you lived an healthy lifestyle. when you do something wrong it is deadly, same for a system. so let the bad things kill and learn from it. fear of death is our biggest problem, it is irrasjonal and therefore killing us.
It's very ironic to hear someone dogmatically insisting on the innovative supremacy of capitalism vs industrial policy (right around the 1 hour mark) on a podcast that mostly focuses on nuclear energy (and has recently done a series on the Manhattan project no less!)
Good point, no reason to assume that democracy and freedom will win. Totalitarianism is default historical government for a reason, can make amazing things happen in crisis. Founding fathers knew this which is why they made executive war powers so strong
Same problem about crypto the more crypto there is the more energy consumption so until we figure out a new cheap energy source I think it's only going to be a novelty for the elite and their aspirants in my not so humble opinion.
Interesting podcast. There are things I agree with and things worth challenging. He is correct that latency is critical, but latency in streamed video is not the big challenge, even on real time zoom calls. I disagree about his comments on autonomous cars, the time critical compute will be in the car, not on the edge. The control system needs to have few parts that can fail. The cars will of course be supported by cloud and edge compute. He should be less focused on politics and just make the science based comments.
"No system of energy can deliver sum useful energy in excess of the total energy put into constructing it. This universal truth applies to all systems. Energy, like time, flows from past to future" (2017).
So what's the solution, 15 minute city for you, no car, oh by the way you just maxed out your quota of energy credits, looks like your shitting in the woods
Id be interested in Mark's opinion on the "data" about how much the computing /app / data / AI { & i realise those are all different in various senses ...but bear with me here} economy spectrum is proportionally "real" & doesn't degenerate into industrial fraud {bots talking to bots to increase the number!}. Is the human appetite for information really unsatisfiable? Or is the arms race of AI dominance tag teaming with the bull$hit 2008 financialization of illusions again. A judgement call / market correction might be needed here, or better still proper democratic oversight, independent from current vested interests. I hope that this new Think Tank is not a neoliberal circle jerk & will genuinely tackle collective interest of humanity without assuming markets fix all. I hope im wrong but this sofar {very unprofessional commenting before finishing vid or going away & reading more of his work}, this sounds like techno hopium without addressing systemic structural issues. Big fan of Nuclear, as part of a diverse approach to energy, necessity is the mother of political change, so theres a scary meta crisis mega needle to thread in next 5-20 years. Green Realsim is coming weather tree hugging basket weavers or gordon gekkos like it or not
First, I strongly disagree that we are experiencing deindustrialization in Germany due to the renewable energy sector. This is a myth often spread by right-wing groups, especially the horrible AfD, which bases much of its ideology on climate change denial. Problems arise from the fact that the energy grid is not being built fast enough to transport electricity from the north, where most wind energy is generated, to the south, where many industrial activities take place. However, this is not due to technical issues but due to political turbulence and the reluctance of southern Germany to build the necessary infrastructure. There is a lot of misinformation spread, such as the claim that there is not enough wind in the south, millions of birds dying from wind turbines, infrasound nonsense, and so on. There are about half a dozen peer-reviewed major studies that show how to build a functioning renewable energy network. Of course, some parts only work if renewable energy capacity is doubled or tripled. When it comes to AI, I am quite sure that the big problems do not arise from electrical grids, which are already a challenge on their own. The big problems become visible only when we talk about the fairy tale that AI will create efficiency gains so huge that we will see significant benefits for our environment. At first glance, you might think so. But what are efficiency gains? You find them everywhere, for example in warfare, oil companies achieving incredible improvements in finding oil, and much other nasty stuff we don’t talk about much. The elephant in the room, totally overlooked by AI enthusiasts, is the rebound effect or even worse, the backfire effects of efficiency gains. Consider this: since 1990, we have improved global technical efficiency by about 36%. This means for every calorie of fossil fuel, we get 36% more usable end energy. This is the technical definition of efficiency gains. Meanwhile, our energy usage has risen by about 60%. So where does the claim come from that efficiency will save us? It is as much a childish wish as the hope that God will save us. Do we feel we are getting it wrong? Maybe. The wish for a father figure to regulate it for us might play a role. At least, as long as we see AI in a narrow context, not setting the boundaries large enough to see the consequences of the first, second, or third order, it looks great. But the reality is that efficiency alone can’t save anything at all. We just have a wrong concept of how this works if we believe AI, owned by corporates and used for narrow goals, will help us.
You are right about efficency and energy being a precursor to even more consumption, but you are wrong about deindustrialization of Germany not resulting from.higher energy prices. I dont care which party or person says it, they are right. And to label some one a "climate desire" when you point out an engineering / economic fact is a big cop out strawman fallacy. Come on, 30 cents a KW- hr in Germany versus 10 or 12 cents in the Arab Gulf States , Asia, or American SE is a pretty big comparative economic disadvantage, you have to admit.
@@kenvrinten3450 it's late over here in germany. I have to answer another day. Sure you are right at first glance but the real question is, why! At the moment energy price is expensive.. And the short answer would be, the energiewende is only half finished. Untill it is we have to use a double structure of renewable and Fossile. When finished renewals will be much cheaper but now. And the second answer is. Fossil energy is dying in the next two or so decades. But to explain the details.. Can't be done here on UA-cam in two sentences. When oil goes down.. Everything else will be influenced. And the only thing holding back peak oil is fracking at the moment global. And this is peaking in 5 to 10 years.. So.. Not choosing renewable energy is the most desastrous expensive decision ever
Sadly our dumb government have agreed to build several of these Energy Greed Data Centres in Auckland While our South Island hydro lakes are at critical crisis low water levels 🧐☹️
101215. Nanoscale uniformly aligned parallel diode arrays may be a practical way of absorhing ambient heat which is uniform on a mesoscale and non uniform on a nanoscale. Plausibly, this heat absorbtion would yield, by conversion without gain or loss, an equivalent amounts of electrical energy. Johnson Nyquest thermal electrical noise power is difficult to deny. AND Diode voltage / current characterstics can be displayed on an XY tracing oscìlloscope. Creditable replicatable proof of concept protypes beyond improper influence are needed. Aloha
Yes we need nuclear power again, maybe more than ever as us old folks die from heat. The only thing is it needs to be small modular molten salt thorium reactors, which can reduce the cost of nuclear to make desalination of seawater irrigation cheap enough to use.
Where you get all these people smart people Peter zahan's right? And so is this Mills guy? I hope AI can come up because we don't have enough smart people man this has never been done before infrastructure and we got stupid politicians
America is not a democracy, it is political oligarchy serving corporatocracy. The paradigm is changing during the capital capture of the nation state, to include an entirely new mode of production through techno-feudalism (Yanis Varoufakis). Meanwhile, this guest is so embedded in managing the neoliberal establishment, and probably invested in Two-Party political tribalism, that presumably none of the analysis escapes market and imperialist friendly framing. The following presents alternative concerns about the AI/energy relationship. AI is being developed by the establishment as integral to national security - the National Security Commission on Artificial Intelligence (NSCAI), which program is intended to keep pace with the Chinese social credit system. AI is also being developed as integral to American surveillance capitalism/techno-feudalism. AI will additionally be deployed as the technology asserted to finally be capable of stabilizing the capitalist business cycle, which achievement however, will require a Centralized Bank Digital Currency (CBDC) with integrated Digital ID compelling social credit system compliance. AI will also be deployed to affect an individual-based carbon tax (Palantir Technologies), where each individual’s carbon footprint will be taxed under the pretense of controlling climate change, which program will also require the CBDC/Digital ID including surveillance capitalism/techno-feudalism compelling social credit system compliance. Switching over to this new AI paradigm living hell will require a shock to the system, which could be anything, but probably an egregious “business cycle” (Second Great Depression), world war, false flag, or even the systemic energy deficiency problem incurred by the authoritarian establishment doubling-down on AI. Of course, the rationing of energy consumption will also be deemed necessary, and integrated with individual-based carbon tax compelling social credit system compliance.
I have a solution. I once heard a story about some super metal made from silver that would allow electricity to circle the globe on a single wire. Meaning a power company on one side of the country could easily send power to the other side without a need for breaks in the line. Then I never heard anouther story about it again. Ask AI how we could scale materials like this. Even making the chips out of this super metal. For anyone who understands how chips work and how electricity flows, this super metal basically removes 80% of the resistance on the metal so atoms flowed faster and cooler.
Within the first three sentences you came out the mouth with nonsense comparing the political position of America and Russia. If you're this blind right out the box, then i don't think there is much point in listening
As ai becomes more efficient, the power supply problem will disappear. This is already happening with improved hardware and software, especially with quantum computers, once they become mainstream.
Terrific to listen to a scholar who knows what he's talking about rather than uninformed politician or activist.
The renewables crowd on Reddit felt a shiver down their back when he said solar/wind+ batteries is more expensive.
"NO THAT IS WRONNNGG!11!!1 LOOK AT THIS LAZARD REPORT! NUCLEAR TOO EXPENSIVE AND DANGEROUS!!!! AUSTRAILIAAAAAAA!"
Source? Most renewable bros are quite convinced batteries will become cheap enough to replace nat gas as peaker plants. I’m not sure they are wrong.
Your backyard tree is probably cheaper than coal 5,000 feet underground. But scale matters and renewables dominate the scale.
@@segasys1339peaker plants can run for days on a small footprint with a natgas pipe. An equivalent solar/Wind peaker plant needs a lot more square footage for both solar/wind generation and the battery array. So now factor in the cost of land and tax rate. And we are only talking about peaker plants. Now let's talk about baseload on the grid. Most grids are 50% base load (meaning electrical usage that is always present 24/7). The amount of square kilometers needed to match say the nuclear fleet in Ontario would be ridiculous. We already have wind along almost the entire Great lakes coastline in Southern Ontario and wind only delivers 1.2% power needed in Ontario. Meanwhile, three locations (Darlington, Pickering, Bruce) deliver 65% to 70% of what we need (and nuclear power is baseload power). The problem with solar and wind is POWER DENSITY. Modern society needs POWER DENSE sources. If you introduced smaller power dense, weather dependent energy sources, you have to seriously OVERBUILD both the generation and storage in area to compensate for the lack of density and on-demand dispatch capacity of nuclear. Germany has tried it. They overbuilt Solar and Wind at 200% of PEAK use nameplate capacity at a cost of 2 Trillion Euro, only to get 49% actual and they are still burning gas and coal, and they have one of the highest electricity rates in Europe. Solar and Wind are not the cheapest when you add in the cost of all the necessary mitigation because of the nature of solar and wind. Solar and Wind DO NOT SCALE WELL. Fine for a small home, not good for a large grid.
And this is why this episode is appropriately called energy crisis. Germany has one because of the way they built their grid. When the commentator says "deindustrialization" he means large industry leaving Germany because of electrical costs AND the cost of LNG as A PHYSICAL MATERIAL OF PRODUCTION, because of the way the German government pushed renewables and natgas hard but banned nuclear. BASF employs somewhere on the scale of 200,000 people. They are actually considering moving out of Germany to the US to continue their chemical plant operations. The amount of oil and gas we use just to make fertilizers, plastics, lubricants, manmade fibres means oil and gas are going nowhere. We can't be dumb about this and ignorant of the scale and numbers involved. If you are a cold climate, heavy industry country, you need nuclear with some nat gas for baseload and peak for generation, and still need oil and gas as materials to make stuff. There is no renewables utopia. We need to find sweet spots if an energy mix and more of those sweet spots will reveal themselves once tech is mature, is cheap, and can scale.
It seems to me there are two parties, original poor. If the rich needs the power they will get it. The poor will live off the crumbs.
Powered by solar and wind w top down control mechanisms that use straight power to digitally direct the populus..
Great listen! Fascinating to hear someone that's focusing on the 'other side' of the doom & gloom needle that so many others have us threading first.
Said others are probably correct, but it gets stale after a while. Something bringing forth projections today of what could lay on the other side of said threading is a breath of fresh air in an increasingly stagnant infosphere of talking heads.
Great podcast. Wonderful topic. Can't wait to hear Mark on batteries
Who is this guy ? Why have I never heard of him ? No crying over spilt milk - we only move forward. Thank you, and look forward to future inputs from this huge brain and heart.
I have absolutely no concept of how many square feet are contained in an average skyscraper so an analogy like this just further obfuscates the actual size of these data centers for me.
My brain is 10 times bigger after listening to Mark which means my brain is now 10 percent as big as Marks
What happens when the AI bubble bursts because there's no way for anyone to make any money with it and it works poorly? That energy won't be needed.
Fingers crossed!
Interesting comment 🤷♀️
Running the internet at the moment is the USA's greatest usage of electricity
jarvens paradox .. look it up
@@ronwalker4998 Jevon's paradox. I know what it is and how to spell it. It is not relevant to my comment.
Working in the Datacenter industry, I can tell you the only limit to growth is the amount of power utilities are willing to provide and they have been very stingy lately.
There is an issue with materials to build the centre as well.
I worked on the construction side of Data centres and the amount of material for the build was/is massive from concrete to metal especially for cooling systems and electrics.
There will be issues there too.
@Ln-cq8zu As far as I know, there are no supply constraints on concrete, steel, and whatnot. Nor for copper wire, cooling, and electrical equipment. The only exception was during COVID, where certain electrical equipment had lead times as high as 18 months. The contraints now are all on electrical power capacity on the local utilities.
Looking forward to the battery conversation to follow. thank you both.
Priority given to residences? In what country?
In the state of New South Wales in Australia the government forces an aluminium smelter to turn off when the load is too high.
I thought he said "will be." It does seem like an oxymoron.
He says there is no energy transition, this is incorrect. Collectively we have decided to take two steps back before we can continue forward. So-called "renewables' are the two steps back and we appear to need a few more before the real transition can begin.
Cheap low-pressure/high-temperature nuclear will replace fossil fuels but only when the cost of fossil fuels forces it to happen.
CHEAP? low pressure hi temp nuclear?? Closer than fusion BUT VERY DIFFICULT.
@@briancam_2000 It has nothing to do with fusion, 100% fission. The reactor that TerraPower is building in Wyoming is an example.
Brain washed heathen
Keep your spent fuel cool for 200 generations.
Did you want Mad Max?
And what happens when uranium runs out?
@@Zanderzan1983 the sun will burn out before we run out of uranium and thorium.
I really enjoyed this episode! It was interesting learning about the future of computer energy consumption, zeta bites and the AI impact on medicine. I am a pathologist. It will be amazing when an AI companion is at hand while I am reading digitally scanned microscope slides for a tissue diagnosis in the future.
Good talk. Thanks!
Great guest!
What a knowledgeable guest.. brilliant stuff
Mark P Mills is an excellent guest... regarding aviation 10 - 15 years ago there were very few Chinese holiday travellers now there are 120 million per year So computing network lanes and bytes are increasing 10-fold that rate. There are no solutions only trade-offs.
"We are going to be forced into a more reality-based power posture" - the smartest thing I have heard said about energy since the green revolution started.
I like efficiency, and I like wind, solar, and storage when they are able to achieve a favorable lifetime, total cost of ownership / value of power produced. Many green power producers will never create more energy than the value of the energy and labor required to build and maintain their installation.
We collectively really need to figure out how to make cost-effective green energy and storage so we can save our fossil fuels for future generations. It will be painful if we burn all of our oil with an ROI above 3/1 in a 200-year period and then struggle to fuel transportation and agriculture for the next 10,000 years. Think ahead beyond your death to the future of our children and grandchildren. This petroleum constrained future is not that far away.
Is it realistic to think two solutions to energy needs are possible? IDK.
Mark is the man!
In Belgium, residents pay full price of 32 euro cent per kWh, while industry pays 9.
What in the heck are we dealing with if we have a world with a robust nuclear presence for grids and big data while commensurately we have a dearth of fossil fuels required for global transportation and manufacturing and farming? A series of re-localized economies that rely heavily on rail and semi-local farming that still have access to the internet?
Billions of blue blistering barnacles! I love that captain Haddock in the background!
The question is not so much how to power something Application Centers or the coming based AI computing but how efficiently to use the energy. In this coming electrical power bonanza, as long as the heat output of the data center can be piped (via water) an stored for use, that will increase the usefulness of the computing. Regrettably though, this residual heat ( 99.9% of all energy going to the data centers ), is no higher then 40-45C (for best computing conditions ), and it can be used only for direct winter heating. If only there could be a "cold steam low-temp" turbine (there are some encouraging independent research in that direction) that can take that low temp fluid and extract energy from it ( with some efficiency ), and concentrated it again, in form of electricity, the energy efficiency of the data centers can be vastly improved, from thermodynamic point of view, as they will not only be consuming but also "producing" energy.
Is this Doombergs dad?
Nice catch. Good ear. That guy IS doomburg
That's definitely doomburg without voice modifier
@@Briand-ei1gs the only way to be sure is to have both on at the same time.
@@pin65371 damn your on fire
To be fair Mark very critical of Putin….Doombetg does his sit on the fence stuff
What did I get from this? People don't plan very well and there will be a construction boom in power plant construction.
The U.K’s last coal fired power station closes in September in the insane dash to “net zero” . That is 2000 megawatts taken out of the National Grid. This is at a time when demand for energy is on a huge increase, where is all our electricity going to come from? Norway? France? A few windmills and even fewer solar panels?
Folks are mostly incapable of thinking in reality.
Solar panels wiith average yearly capacity factor in the single digits and off shore wind turbines with much higher capacity factor, yes, but the size of the statue of liberty with all that concrete, Steel, aluminum, plastics, etc., not to mention the high voltage transmission lines to tie them into the grid.
Any "hype" finishes in one way or another with a crisis. But the crisis is not the death of the hype. It's a necessary phase for moving post the hype. AI will put pressure on energy production and push GAFAM to increase production of their own renewable energy. Nuclear will remain "anecdotic" with a share of around 10-15% of the energy production mix.
Wisdom
@@bayleybomber 1:17:08
1:17:08 😊😊
Only in the EU and possibly in the ÚS, which electricity projection and production on state level, without any national grid.
Plus the problem of seasonality and transmission (blocked by the Greens).
It is quite the narrative change to go from green energy / net zero to adding 25% plus fossil fueled energy protection to the grid just for AI. Is the juice really worth the squeeze with AI to use 25% more electricity than our current total production? Kinda reminds me of "the block chain." Is it really worth the resource use we have invested. I say mostly not.
Video doesn't need low latency if it is for consumption only, like Netflix movies or youtube or iCloud. Low latency is crucial only for interactive use cases. On-line gaming and mostly stock trading, which fuels low latency connections. Starlink, commucating through vacuum beats optical cabels on long distances due to a faster speed of light in a vacuum.
ai is a huge problems, and so are phones.
but also medicine, we must not get tech into medicine.
health is done preventively, not after the fact you lived an healthy lifestyle.
when you do something wrong it is deadly, same for a system.
so let the bad things kill and learn from it. fear of death is our biggest problem, it is irrasjonal and therefore killing us.
If ever there was a plea for physicists and engineers, this is it. Not so many of them in politics! But first we need to use our heads!
An euphemistic but not entirely inaccurate term for an apartment building would be "Vertical Village."
It's very ironic to hear someone dogmatically insisting on the innovative supremacy of capitalism vs industrial policy (right around the 1 hour mark) on a podcast that mostly focuses on nuclear energy (and has recently done a series on the Manhattan project no less!)
Good point, no reason to assume that democracy and freedom will win. Totalitarianism is default historical government for a reason, can make amazing things happen in crisis. Founding fathers knew this which is why they made executive war powers so strong
Same problem about crypto the more crypto there is the more energy consumption so until we figure out a new cheap energy source I think it's only going to be a novelty for the elite and their aspirants in my not so humble opinion.
Turns out crypto was extremely useful for government transactions
@@bayleybomber you mean for the _not insider trading_ ?
Interesting podcast. There are things I agree with and things worth challenging.
He is correct that latency is critical, but latency in streamed video is not the big challenge, even on real time zoom calls.
I disagree about his comments on autonomous cars, the time critical compute will be in the car, not on the edge. The control system needs to have few parts that can fail. The cars will of course be supported by cloud and edge compute.
He should be less focused on politics and just make the science based comments.
"No system of energy can deliver sum useful energy in excess of the total energy put into constructing it.
This universal truth applies to all systems.
Energy, like time, flows from past to future" (2017).
This doesn't seem universally applicable.
Thankfully, we didn't have to make the energy systems we just tap into them. That's way easier.
I do not believe so.
I know if humans were serious about conservation of energy we would use a 18 volt domestic grid
So what's the solution, 15 minute city for you, no car, oh by the way you just maxed out your quota of energy credits, looks like your shitting in the woods
Really love to see Mark join the Trump team
Measured and mesmerising data as always
Wearhouse scale computing = wearhouse hardware = were-ware
Need to put a small tax on data processing. Problem solved.
Not sure ai will be necesssary if most humans are battling flood, fire, drought, conflict, migration etc.
Id be interested in Mark's opinion on the "data" about how much the computing /app / data / AI { & i realise those are all different in various senses ...but bear with me here} economy spectrum is proportionally "real" & doesn't degenerate into industrial fraud {bots talking to bots to increase the number!}. Is the human appetite for information really unsatisfiable? Or is the arms race of AI dominance tag teaming with the bull$hit 2008 financialization of illusions again.
A judgement call / market correction might be needed here, or better still proper democratic oversight, independent from current vested interests. I hope that this new Think Tank is not a neoliberal circle jerk & will genuinely tackle collective interest of humanity without assuming markets fix all.
I hope im wrong but this sofar {very unprofessional commenting before finishing vid or going away & reading more of his work}, this sounds like techno hopium without addressing systemic structural issues.
Big fan of Nuclear, as part of a diverse approach to energy, necessity is the mother of political change, so theres a scary meta crisis mega needle to thread in next 5-20 years.
Green Realsim is coming weather tree hugging basket weavers or gordon gekkos like it or not
First, I strongly disagree that we are experiencing deindustrialization in Germany due to the renewable energy sector. This is a myth often spread by right-wing groups, especially the horrible AfD, which bases much of its ideology on climate change denial.
Problems arise from the fact that the energy grid is not being built fast enough to transport electricity from the north, where most wind energy is generated, to the south, where many industrial activities take place. However, this is not due to technical issues but due to political turbulence and the reluctance of southern Germany to build the necessary infrastructure. There is a lot of misinformation spread, such as the claim that there is not enough wind in the south, millions of birds dying from wind turbines, infrasound nonsense, and so on.
There are about half a dozen peer-reviewed major studies that show how to build a functioning renewable energy network. Of course, some parts only work if renewable energy capacity is doubled or tripled.
When it comes to AI, I am quite sure that the big problems do not arise from electrical grids, which are already a challenge on their own.
The big problems become visible only when we talk about the fairy tale that AI will create efficiency gains so huge that we will see significant benefits for our environment. At first glance, you might think so.
But what are efficiency gains? You find them everywhere, for example in warfare, oil companies achieving incredible improvements in finding oil, and much other nasty stuff we don’t talk about much.
The elephant in the room, totally overlooked by AI enthusiasts, is the rebound effect or even worse, the backfire effects of efficiency gains.
Consider this: since 1990, we have improved global technical efficiency by about 36%. This means for every calorie of fossil fuel, we get 36% more usable end energy. This is the technical definition of efficiency gains.
Meanwhile, our energy usage has risen by about 60%.
So where does the claim come from that efficiency will save us? It is as much a childish wish as the hope that God will save us. Do we feel we are getting it wrong? Maybe. The wish for a father figure to regulate it for us might play a role.
At least, as long as we see AI in a narrow context, not setting the boundaries large enough to see the consequences of the first, second, or third order, it looks great. But the reality is that efficiency alone can’t save anything at all.
We just have a wrong concept of how this works if we believe AI, owned by corporates and used for narrow goals, will help us.
You are right about efficency and energy being a precursor to even more consumption, but you are wrong about deindustrialization of Germany not resulting from.higher energy prices. I dont care which party or person says it, they are right. And to label some one a "climate desire" when you point out an engineering / economic fact is a big cop out strawman fallacy.
Come on, 30 cents a KW- hr in Germany versus 10 or 12 cents in the Arab Gulf States , Asia, or American SE is a pretty big comparative economic disadvantage, you have to admit.
@@kenvrinten3450 it's late over here in germany. I have to answer another day. Sure you are right at first glance but the real question is, why! At the moment energy price is expensive..
And the short answer would be, the energiewende is only half finished. Untill it is we have to use a double structure of renewable and Fossile. When finished renewals will be much cheaper but now.
And the second answer is. Fossil energy is dying in the next two or so decades. But to explain the details.. Can't be done here on UA-cam in two sentences.
When oil goes down.. Everything else will be influenced. And the only thing holding back peak oil is fracking at the moment global. And this is peaking in 5 to 10 years.. So..
Not choosing renewable energy is the most desastrous expensive decision ever
Requiring milliscond responses from a stock market is simply gambling for wall street.
This does not benefit main street or industry.
Sadly our dumb government have agreed to build several of these Energy Greed Data Centres in Auckland While our South Island hydro lakes are at critical crisis low water levels 🧐☹️
101215. Nanoscale uniformly aligned parallel diode arrays may be a practical way of absorhing ambient heat which is uniform on a mesoscale and non uniform on a nanoscale. Plausibly, this heat absorbtion would yield, by conversion without gain or loss, an equivalent amounts of electrical energy.
Johnson Nyquest thermal electrical noise power is difficult to deny.
AND
Diode voltage / current characterstics can be displayed on an XY tracing oscìlloscope.
Creditable replicatable proof of concept protypes beyond improper influence are needed.
Aloha
Yes we need nuclear power again, maybe more than ever as us old folks die from heat. The only thing is it needs to be small modular molten salt thorium reactors, which can reduce the cost of nuclear to make desalination of seawater irrigation cheap enough to use.
Sounds like a plan. When will it begin in earnest.
Burn natural gas AI would never use fissionable material.
Thumbs down for promoting nuclear energy
why? I must have missed the truth.
Is that you, Doomberg?😊
UK need facility to enrich nuclear fuel.
Tunnel to Northern Ireland .
And develop of hydrogen engine
Are these folks using the term cloud to refer to the internet ?
. .yes I think they are
Not everything on the Internet is in a cloud
Only in their minds.
Where you get all these people smart people Peter zahan's right? And so is this Mills guy? I hope AI can come up because we don't have enough smart people man this has never been done before infrastructure and we got stupid politicians
Data centers is not a bad name; I would name them data mills. I think it is appropriately industrial :)
America is not a democracy, it is political oligarchy serving corporatocracy. The paradigm is changing during the capital capture of the nation state, to include an entirely new mode of production through techno-feudalism (Yanis Varoufakis). Meanwhile, this guest is so embedded in managing the neoliberal establishment, and probably invested in Two-Party political tribalism, that presumably none of the analysis escapes market and imperialist friendly framing.
The following presents alternative concerns about the AI/energy relationship. AI is being developed by the establishment as integral to national security - the National Security Commission on Artificial Intelligence (NSCAI), which program is intended to keep pace with the Chinese social credit system. AI is also being developed as integral to American surveillance capitalism/techno-feudalism. AI will additionally be deployed as the technology asserted to finally be capable of stabilizing the capitalist business cycle, which achievement however, will require a Centralized Bank Digital Currency (CBDC) with integrated Digital ID compelling social credit system compliance. AI will also be deployed to affect an individual-based carbon tax (Palantir Technologies), where each individual’s carbon footprint will be taxed under the pretense of controlling climate change, which program will also require the CBDC/Digital ID including surveillance capitalism/techno-feudalism compelling social credit system compliance. Switching over to this new AI paradigm living hell will require a shock to the system, which could be anything, but probably an egregious “business cycle” (Second Great Depression), world war, false flag, or even the systemic energy deficiency problem incurred by the authoritarian establishment doubling-down on AI. Of course, the rationing of energy consumption will also be deemed necessary, and integrated with individual-based carbon tax compelling social credit system compliance.
I have a solution. I once heard a story about some super metal made from silver that would allow electricity to circle the globe on a single wire. Meaning a power company on one side of the country could easily send power to the other side without a need for breaks in the line. Then I never heard anouther story about it again. Ask AI how we could scale materials like this. Even making the chips out of this super metal. For anyone who understands how chips work and how electricity flows, this super metal basically removes 80% of the resistance on the metal so atoms flowed faster and cooler.
The rich and the poor
Within the first three sentences you came out the mouth with nonsense comparing the political position of America and Russia. If you're this blind right out the box, then i don't think there is much point in listening
As ai becomes more efficient, the power supply problem will disappear. This is already happening with improved hardware and software, especially with quantum computers, once they become mainstream.
Everything needs to be saved
AI will increase CO2 load in our atmosphere .. efficiency only increases demand and more electricity use .. jarvens paradox look it up