How energy will shape the future of AI (and vice versa)
from Everything Energy
31m 31s
Artificial intelligence is transforming industries and economies, but its growth comes with significant energy demands. This report reveals that data centers—where AI models are trained and run—consume about 1.5% of global electricity today, rising to nearly 950 terawatt-hours by 2030, equivalent to Japan’s annual usage. While this increase poses challenges to energy systems, especially in regions like the U.S. and Europe where data centers cluster in key urban areas, the impact remains within a small fraction of global emissions. AI also presents major opportunities for energy efficiency: it improves solar forecasting accuracy by over 90%, helps detect methane leaks in oil and gas, and can reduce energy demand in industry by an amount equivalent to Mexico’s annual consumption. However, risks include cyber threats, where AI is used to mimic voices and launch sophisticated attacks on energy infrastructure, and supply chain vulnerabilities, such as China’s dominant role in rare earths like gallium. The report underscores that while AI’s energy footprint is substantial, its potential to drive innovation and efficiency in clean energy technologies is significant. Yet, without coordinated policy, transparency, and investment in grid resilience, the full benefits of AI in energy may not be realized, and its integration into energy systems remains a complex, evolving challenge.
Welcome back to the IEA's podcast Everything Energy. I'm Dan Hewitt. This week we're
looking at energy and artificial intelligence. In just a few years, artificial intelligence
has transformed from an academic idea into a trillion dollar industry. Today it's on
our phones, laptops, and in the tools we use to create images, film clips, even podcasts.
AI has the potential to reshape economies and the future of work. But without electricity,
there's no AI. Generating a single image with generative AI consumes roughly as much energy
as charging a smartphone. But policy makers and the markets have lacked the tools to fully
understand the links between AI, energy, and what they really mean. The IEA's major new
report on AI has tried to fill that gap with the most comprehensive analysis on this topic
to date. On the one hand, it showed the scale of the challenge. It found by 2030 that electricity
demand from data centers is projected to more than double to about 945 TWh. That's about
the same as the total electricity consumption of Japan today. But it also found that AI
offers real opportunities for the energy sector too. Here to discuss both the opportunities
and the challenges are the reports to lead authors, i.e. analysts, Thomas Spencer, and Sid Arthasing.
Alright, Sid Artham, Thomas. Thank you for joining me. So let's start with why this report
on the links between AI and energy is needed now. And there are a few big moments that got
us here. The first is to do with ChatGPT. The second is about a Nobel Peace Prize for
Biomedicine, so you could you explain a bit more about how they're linked?
Yes, so these are two breakthrough moments where artificial intelligence really rose up
the agenda in terms of the public awareness of the topic. So ChatGPT I think was the first
commercial application of artificial intelligence that people in their day-to-day life started
using. And with that came increasing concern about the growing energy footprint of artificial
intelligence because all of a sudden people are using an application, ChatGPT, which is
quite energy intensive, you know, querying ChatGPT is much more energy intensive than putting
in a Google search, for example. So with the rise of ChatGPT as a commercial application
for AI, people started growing concerned about the energy demand implications. On the other
hand, with the Nobel Prize last year, which was awarded to three AI scientists for breakthroughs
that they made in modeling the 3D structure of proteins, it sounds complicated, but really
this is critical to drug discovery because the 3D structure of proteins determines how they
interact with molecules in the body and also with drugs that we introduce into the body.
So with this Nobel Prize, again in the public awareness, it really became clear that AI could
solve some big problems. And so people started to ask also in the energy sector what problems could
AI solve. And so with these two moments, you had the realization that AI might have significant
implications for energy demand, but on the other hand could also solve a whole range of problems
in the energy sector. And so these two breakthrough moments for AI also had a lot of implications
for how we think about the use of AI in the energy sector. And there is a third issue which is
to do with uncertainty or a lack of understanding about what AI is and where we are with it. And in
many ways though, we actually did deep seek was a really useful example of that lack of understanding.
So when deep seek R1, which is the reasoning model by this group based in China, was launched
in January 2025, it took a few days for the markets to understand what had really happened.
What this model basically did was that it was able to synthesize vast amounts of data in innovative
ways that took far less electricity, far less energy to actually do those calculations.
The markets responded by basically selling stocks of chip manufacturers, of energy companies.
Basically the tools that would help us understand what this means for the energy sector
didn't really exist. So this really is where the AI comes in. This is where we are able to
bring together different sets of stakeholders, different sets of data to really understand
the implications of disruptive changes like this. So firstly, we found that in experimental
conditions, the large language model, the reasoning models consumed 10 to 20 times more electricity
than simple queries using small language models. So already, even though deep seek R1
may have had innovative approaches, it's still because of the nature of the model itself,
consumes far more electricity than then small models were. And the second is that many of these
innovations were in the training of the model itself. But we find that the training of all large
language models over the past several years have consumed as much as about two terawatt hours of
electricity. Compared to 400 electricity that's consumed every single year, at least that was
the number for 2024, by all data centers for all types of calculations, not just for artificial
intelligence. Now we've mentioned data centers a lot here. So perhaps this is a good moment just to
step back and remind ourselves what data centers are. Because for example, they're not just used
for AI. So maybe Thomas, you could tell us what's a good way of understanding what we're talking
about. I think Dan, in simple terms, you can think of a data center as the backbone of the internet,
right? So data centers are big buildings where we have lots of computers, essentially, unlike the
computers that we have in our homes or offices. They don't have screens because we don't need to
interact with them directly, but they do similar things to the computers in our homes and offices.
They store and process data. So for example, when you go online and watch a video from Netflix,
for example, that video is obviously not stored directly on your TV or on your laptop.
It's stored in a data center, probably one that's quite close to where you are located. So here in
Paris, it might be just outside Paris. And essentially your TV or computer is sending a signal to that
data center saying, I want to watch this particular video, this particular TV show, send me the data
so that I can stream it on my TV or laptop. And they're critical for artificial intelligence,
essentially because artificial intelligence models, like chat GPT, are too big to run on your
laptop. They need to be run on specialized hardware, which is located in data centers. And that's
why looking at the electricity demand from data centers is critical when we think about the
footprint of artificial intelligence because basically all of the calculations related to
artificial intelligence take place in data centers and therefore all of the electricity consumption
that is linked to artificial intelligence takes place in data centers as well.
How do you go about working out how much energy data centers use? Because that's what this
will report was about. And that's what people haven't known much about before. So how do you do it?
Yeah, it's a great question. It's almost like piecing together different pieces of information.
And trying to build up a picture of the of the whole. The reason it's like that is that as of
today, we don't have any comprehensive global statistics on the electricity consumption of data centers
in the same way that we have comprehensive statistics, for example, on the energy consumption
of the iron and steel sector or the transportation sector. Governments don't collect comprehensive
statistics yet on the electricity consumption of data centers. And while most big tech companies do
report on their total electricity consumption, they don't necessarily break out the share of that
that comes from data centers versus their warehouses in the case of Amazon or their office buildings.
What we did is compare a whole range of sources. First of all, we gather data on all of the existing
data centers around the world. There's about 11,000 of them as of today. Secondly, we gather data
and analysis and we talked a lot to the tech industry about how these data centers are run.
And so once you have an understanding of the number and how they are run, you can make an estimate
of their electricity consumption. Then we compare it with what national statistics do exist.
For example, in Ireland, they collect very good statistics on data centers and their electricity
consumption in Ireland is a very big market actually for data centers in Europe. So we can compare
our estimates with the national statistics that do exist. And finally, another source for us was
to compare with company reports. So we gathered the electricity consumption as reported by
the major tech companies globally. So your Amazon's, your Google's, your Microsoft's,
but also your bidus, your Tencent's, your Alibaba's in China and a whole range of companies.
And so really in a kind of detective approach, piecing together all of these different sources,
we came up with our really first of a kind comprehensive estimates at the global level but also
for major countries and regions of the electricity consumption of this really important sector.
And the thing is about how much energy data centers use our fairly extraordinary. The big data
centers currently in operation use as much energy as 100,000 households. But the report says
that the largest data center announced could use as much energy as 5 million households,
which is the same as a big city popping up every 18 months. So just tell me a bit more about how
governments can prepare for adding that amount of electricity demand.
So the analysis that we conducted also included an element of looking at where data centers
exist through satellite imagery to understand what the relationship may be to energy systems
[BLANK_AUDIO]
and our urban life.
So we found a few interesting facets and characteristics
of these data centers, right?
Firstly, they tend to be very spatially concentrated.
So unlike other forms of energy-consuming infrastructure,
say for example, iron and steel, mills,
or aluminum smelters, or factories, data centers
tend to accumulate around each other.
The second thing is that these data centers
tend to be closer to urban areas compared to these other forms
of infrastructure.
So they tend to come up where there's
already a big center of energy demand.
And thirdly, these data centers are growing larger
and larger in size.
So as you mentioned, an accelerated data center which
caters to AI demand today could be typically
consuming as much electricity as 100,000 households.
The largest in construction is as much as 3 million households.
And the largest that's being planned is 5 million households.
And these come up very, very quickly.
So compared to energy infrastructure,
it could take as much as five to 10 years,
depending on the types of infrastructure
that needs to be built.
The data center itself could come up
as quickly as 18 months, as you rightly mentioned.
The governments and regulators have
a very important role to play.
In firstly, understanding the demand
that may come out from these data centers.
So ensuring there's greater transparency,
we're able to have this dialogue between the technology
industry, the governments, and so on.
So this is also a place where the IA has a role to play
as a convening body we are able to bring together
these different stakeholders.
There were better sense of where this demand will come from,
where it will be geographically situated.
And therefore, how can the energy industry plan
for something like that?
Regulators also ensure that better decisions
are made as far as the geographical sighting
of these data centers are concerned.
So they really come up in places where the energy systems
are able to handle the increased load.
What concerns are there about overbuilding?
Because if it takes 18 months to build
one of these enormous data centers,
what concerns are that the grid won't be able to keep up?
So you may have far more data centers,
but you can't actually supply the energy and eat.
Already that there have been interesting cases
where the distribution companies are unable to meet the demand
coming out of these data centers.
So many of the new connection requests have either been paused
or they're in a very long queue.
So it really puts the build out of the AI infrastructure
in some form of a risky situation.
And there, of course, billions of dollars
that are being lined up to ensure
that this infrastructure is ready in time
for when people need it.
Now, there are some very big figures mentioned in this report.
So let's just remind ourselves what share
of global electricity demand comes from data centers today.
Today, data centers consume about 1.5% of global electricity
consumption.
In absolute terms, that's about 415
terawatt hours of electricity consumption.
You can roughly compare that with the electricity consumption
of a country like France today.
Going forward, our projection is that by 2030,
the electricity consumption of data centers
will more than double, so reaching about 945
terawatt hours of annual electricity consumption.
To put that in perspective, that's
roughly equivalent to the annual electricity consumption
of Japan.
So that doubling is a big challenge for the energy system
to meet as said, rightly highlighted.
But again, we need to put it in perspective.
Even with that doubling, our projection
is that data centers will still account for around 3%
of global electricity consumption.
And of course, we need to also remember
that there are other drivers of electricity consumption
growth around the world.
So particularly in lower income countries,
it's economic growth and industrialization,
but also growing demand for cooling because of climate change,
but also because of rising incomes.
And then one of our objectives of decarbonization
is to electrify the energy system to reduce emissions.
And so growing demand for electric vehicles,
for example, or the electrification of heating
is also driving up electricity demand.
So at the global level, we actually
find that data centers account for slightly less than 10%
of the electricity demand growth.
Again, that's not the absolute value.
That's the growth between now and 2030.
So yes, doubling is a big challenge for the energy system.
Yes, it's happening very quickly.
But again, we need to put it in perspective
of this huge size of the global energy system.
Let's talk about where we're going to find these data
centers, because places like Japan, America, Malaysia,
they're predicted to have a lot of them.
And almost half of electricity demand growth in the US
by 2030 is predicted to come from data centers.
But there's also a huge amount of regional variation
expected.
So we find that data centers tend to be
clustered in a few key parts of the world.
So of course, in the United States,
it's Northern Virginia and a few other states, including Texas.
In Europe, it is the cities of Frankfurt,
Tyres, Amsterdam, and of course, London.
And you just mentioned that example in Northern Virginia.
Can you tell us about how much those data centers
there are going to be using?
It's 25% in the case of Northern Virginia.
It's highest 20% in the case of Ireland.
So really, in different parts of the world,
data centers account for a larger and disproportionate
share of the total electricity demanded in those regions.
We find that the growth that is emerging now
from the United States, from other parts of the world,
really highlight the importance of integrating
data centers into the larger energy system.
So for example, in the United States,
in 2030 already, the amount of electricity
consumed to process data will be more through data centers
than the total electricity demanded
from the iron and steel sector, from chemicals,
from cement, from aluminum, and all of that combined.
But it really comes down to where these different economies
are in their respective energy trajectory.
So for example, take China.
China is the world's second largest market of data centers.
And yet, it plays a much smaller role in the growth in China,
because China is also seeing very high levels of electrification
with the demand coming out of rising electric vehicles.
So already, for example, last year, nearly half of the vehicles
that were sold were either electric or hybrid electric.
So that really highlights how the other parts of the system,
including air conditioning, including heat pumps,
including electricity demand from industries,
that are also driving consumption dead.
You just mentioned China.
And I think that's a really good way
for us to move on to talk about supply chains.
People possibly think about supply chains with batteries,
but not necessarily for AI.
And AI needs all kinds of rare earths and minerals.
And a huge percentage of them are refined in China.
So can you give us a bit more context
about the complexity of AI supply chains?
Yeah, I think, you know, when people ask a question
to chat GPT, they typically don't have in mind the fact
that behind this apparently disembodied technology
is a huge supply chain that cuts across many, many sectors
and is very international.
So for example, to connect a data center to an electricity grid,
you need transformers.
Transformers today are a large global market.
Many countries import most of their transformer demand.
We have seen that waiting times for transformers
have increased recently because of the growing demand
for investment in our electricity grids and so on.
So transformers are a critical part of the value chain
for data centers and therefore for artificial intelligence.
If we look at the components that go into data centers,
they need a whole range of critical minerals and metals.
So for example, the wiring in a data center
to transfer port electricity all around the different servers
in the data center, that's copper.
Copper demand is growing very strongly
as we invest more in electric vehicles
as we build out our electricity grids, for example.
And by the way, because data centers require
a very, very high level of redundancy,
so they can't ever go offline because of an accident,
they duplicate all of the network copper network
in data centers.
So they're quite copper-intensive as well.
If we take rare earth metals like, for example, gallium,
gallium is increasingly being used in the next generation
of computing chips because it helps them
be much more power efficient than the current generation.
And 99% of refined gallium supply is in China today.
We estimated that the growth in gallium demand
from data centers between now and 2030 could account
for 10% of the total gallium production today annually.
So that's a very significant increase in this metal.
So one of the critical messages from our report
was that we need to take a careful look at what you can call
in simple turns, the artificial intelligence supply chain
and how the growing links between data centers
and the energy system may raise different and new kinds
of concerns around energy security,
because ultimately the security of our transformer supply
chain, the security of our gallium supply chain,
may impact the energy sector, but also the artificial.
intelligent sector. Now we know of course that data centers we're going to need more energy
to meet demand. So let's talk a bit about the energy mix because is that increase in demand
going to be met from renewables? Maybe Sid you could just tell us a bit about how that demand
is being met today but what we're projecting in 2030-2035 for sure. So I think firstly let's
try to understand the nature of the electricity demand that comes out of data centers in the first
place right. So take air conditioners. Now air conditioners or air conditioning demand tends
to peak it of course the the warmest parts of the day. So for example in places like India that
could be 2 or 3 pm electric vehicles people tend to charge them overnight at homes or sometimes they
do fast charging at different times of the day. Unlike these different types of infrastructure
data centers tend to have very consistent demand through the course of the day. So the demand profile
is quite flat for which we need to ensure that therefore the electricity that is supplied to it
is able to meet that really flat nature of demand coming out of data centers. Typically firm loads
of power. So this is electricity that comes out from sources that we can control such as natural
gas or nuclear energy tend to be preferred as the suppliers for data centers. But of course as the
system becomes more complex as we have more variable renewable energy. So this is of course electricity
from solar from wind and so on in a well integrated system where they're complemented by more firm
loads of power where they also have storage on the side that can ensure that even for very flat
demand profiles such as coming out of data centers you can have increasing shares of renewables
of solar of wind and so on to also play a part of the demand you know for the data centers.
So we find that the supply really comes down to where or which country you're looking at. So
all these various sources have a very important role to play but we find that in the United States
for example natural gas is increasingly playing a larger role in ensuring that electricity demand
is met in China it is coal that is a significant part of the growth but really renewables are
increasing their share in growth in fact we find that over the course of the next decade
renewables as well as nuclear will have a very strong role to play and therefore it may also help
decarbonize the demand that that comes out of data centers especially post 2030.
Now a lot of people will want to know the relationship between AI and carbon emissions because
on the one hand there are optimists that think that AI can solve the problems of climate change
and on the other hand there are concerns about the added burden to the energy grid and that could
potentially derail climate goals. So what is the data tell us because perhaps the truth is
somewhere in the middle of those two ideas yeah I like the way you put it because it's a real debate
between the optimists and the pessimists or you could say the do-mongerers and the people
who are really bullish on the potential of AI to solve climate change and essentially our report
found that neither of these very opposing positions are justified. So first looking at the emissions
increase how does AI affect carbon emissions well obviously the electricity that is consumed in
data centers may have been produced from fossil fuel powered sources and therefore would have
emissions. We find that the electricity consumption of data centers today accounts for about
180 million tons of CO2 emissions every year. It sounds like a lot but again we need to put this
in perspective of the total energy system. Today global energy sector emissions are about 37,500
million tons per year. So data centers account for less than 1% of the global total today.
Even in our worst case scenario we project that emissions from the electricity consumption in
data centers could rise from 180 million tons today to about 500 million tons by 2035. That would
still equate to less than 1.5% of total global emissions in 2035. So it's not good news for climate
but it's not going to change the game in terms of making our climate goals out of reach.
So that's on the pessimist side of the equation. On the optimist side of the equation
essentially what the optimists are saying is that we can use AI to solve climate change
in very simple terms. So this could be for example using AI to integrate more renewables into our
power systems or for example using AI to innovate new low emissions technologies like batteries
or new nuclear reactor designs and so on. And what we find in our report is that there is a lot of
potential for AI to reduce emissions and accelerate innovation. No data about it. That potential
probably exceeds the increase in emissions that you could have from data centers and AI at
least in the near term. So to 2035 but it's not a given that we will tap this potential.
So for example we would need policy and regulation to promote the uptake of AI related solutions.
And secondly even if we tapped all of this potential it would still only account for a small
share of the emissions reductions that we need to address climate change. So it doesn't really
change the game for climate mitigation either. We still need to be doing what we have needed to do
for a long time which is deploy low emissions technologies and energy efficiency as fast as we
can across the entire energy system. AI can help with part of that but it's not a silver bullet.
And you mentioned efficiency so perhaps we could just dive into that a little bit more and
said could you give us some real world examples about how AI has made our energy system more efficient
with grids with something else that we can point out and say okay this is what's happening right
now. Really AI is already being used for a wide variety of efficiencies for a wide variety of
optimizations of the energy system in the first day. So what we did for the first time ever we put
together some of the best case studies to understand the implications that may have on the energy
system at a broader level. So we find for example let's take the oil and gas sector.
We start with the oil and gas sector because this sector really has been one of the early adopters
of AI of supercomputers. In fact some of the world's fastest supercomputers today are operated by
the oil and gas sector just because the sheer vast amounts of data that's generated and that needs
to be processed in order to extract oil and gas. We find that for example these types of
applications of AI can help spot methane leaks to help reduce unwanted methane emissions which
which is quite a potent greenhouse gas. AI can help ensure that the oil and gas sector works
in a much more sustainable manner. Similarly in the case of the electricity sector again AI is
increasingly being used for vast and very different types of optimizations. In grids for example
we find that in certain parts of the energy system the grids are congested because they congested
because the capacity that the long distance transmission lines have are not able to meet the rising
amounts of electricity that's being generated in those regions and the rising amounts of demand
that's arising from those regions. So we find that AI led interventions can increase the capacity of
these long distance transmission lines by slightly just a little bit but that ensures the overall
balancing of the system happens in a much more efficient manner. AI is already being used also to
ensure that weather forecasting is improved in terms of its accuracy. I was just in India
very recently and we were speaking to the Solar Energy Association of the country and they mentioned
how in one of the states one of the key solar energy producing states there was a 20% unexpected
decline in solar energy generation and that led to massive issues in terms of balanced the grid suddenly
because that was completely unexpected. AI led forecasting has helped improve the accuracy
of forecasting by over 90% and that has helped avoid those kind of issues since that incident occurred
a few years ago. Let me just maybe leave you with a couple of data points. So in the industry sector
AI led optimization and automation and improving the fuel mix and so on has the potential to reduce
the energy demand the equivalent of what the entire country of Mexico consumes today over the
course of the next decade. Okay well that's that's the positive picture but there is another side
that as well because people are concerned about for example the increased possibility of cyber
attacks on grids on energy infrastructures. So Thomas what are the things that people are
concerned about? Yeah so I think the link between AI and energy security it can be described
as a double edged sword so if we take for example cyber attacks there's no doubt that
cyber attacks on energy infrastructure and energy companies are increasing. In our data,
it has roughly tripled over the last four years, and this is cyber attacks specifically targeting
energy companies or critical energy infrastructure. It's also very clear that malicious
actors are using AI to make cyber attacks more sophisticated. For example, in our everyday lives,
we might one day receive a call from someone that sounds a bit like our mom asking for
a transfer of money because they've been locked out of their bank account, for example.
That could be an AI bot that is imitating a person. This is just an example from our everyday lives,
but there are similar applications of AI for cyber attacks that target corporate actors
or critical infrastructure. So that's the sort of negative side of the of the coin. On the positive
side, it's very clear that to stay ahead in this arms race, it is an arms race. Companies
are using AI increasingly as part of their defense protocols. So energy companies are using AI
to monitor their infrastructure, to monitor their digital infrastructure for malicious attacks as well.
[BLANK_AUDIO]
Podcast Summary
Key Points:
Artificial intelligence is rapidly growing and now consumes significant electricity, with data centers projected to use nearly 950 TWh by 2030—equivalent to Japan’s annual electricity demand.
The energy footprint of AI is substantial, with current data centers using about 1.5% of global electricity and 180 million tons of CO₂ annually, though this remains less than 1% of global emissions.
AI offers transformative opportunities for the energy sector, such as improving renewable integration, detecting methane leaks, and optimizing grid forecasting, with real-world examples showing up to 90% accuracy gains.
Data centers are spatially concentrated near urban areas and are growing rapidly in size, with some new facilities consuming energy equivalent to 5 million households—raising urgent concerns about grid capacity and overbuilding.
The AI supply chain involves critical minerals like gallium and copper, with 99% of refined gallium currently sourced from China, posing risks to energy and AI security.
While AI demand may increase electricity consumption, its potential to drive energy efficiency and innovation could offset emissions, though this depends on policy, investment, and implementation.
There is growing concern about cyberattacks on energy infrastructure, with AI being used by malicious actors to mimic voices and launch sophisticated attacks, though energy companies are also using AI for threat detection.
Governments and regulators must improve transparency, plan for regional energy demand, and ensure that AI development is aligned with energy system resilience and decarbonization goals.
Summary:
Artificial intelligence is transforming industries and economies, but its growth comes with significant energy demands. 5% of global electricity today, rising to nearly 950 terawatt-hours by 2030, equivalent to Japan’s annual usage. S.
and Europe where data centers cluster in key urban areas, the impact remains within a small fraction of global emissions. AI also presents major opportunities for energy efficiency: it improves solar forecasting accuracy by over 90%, helps detect methane leaks in oil and gas, and can reduce energy demand in industry by an amount equivalent to Mexico’s annual consumption. However, risks include cyber threats, where AI is used to mimic voices and launch sophisticated attacks on energy infrastructure, and supply chain vulnerabilities, such as China’s dominant role in rare earths like gallium.
The report underscores that while AI’s energy footprint is substantial, its potential to drive innovation and efficiency in clean energy technologies is significant. Yet, without coordinated policy, transparency, and investment in grid resilience, the full benefits of AI in energy may not be realized, and its integration into energy systems remains a complex, evolving challenge.
FAQs
Data centers currently consume about 1.5% of global electricity, equivalent to 415 terawatt hours annually.
By 2030, data center electricity demand is projected to more than double to about 945 terawatt hours, equivalent to Japan's annual consumption.
Data centers have a consistent, flat demand profile throughout the day, requiring reliable and stable power supply, unlike peak-demand sources like air conditioning.
There are growing concerns about cyber attacks on energy infrastructure, with malicious actors using AI to create more sophisticated and deceptive attacks.
Yes, AI can optimize renewable integration, improve weather forecasting, detect methane leaks, and reduce energy demand—potentially saving energy equivalent to Mexico’s consumption.
AI supply chains rely heavily on copper for wiring and rare earth metals like gallium, with 99% of refined gallium currently supplied by China.
Chat with AI
Loading...
Pro features
Go deeper with this episode
Unlock creator-grade tools that turn any transcript into show notes and subtitle files.