From Coal to Compute: China’s Grid Meets the AI Boom with David Fishman

David Fishman is a Principal at The Lantau Group who advises on energy development, infrastructure, and electricity markets across East Asia, with a focus on China. His expertise spans power-sector policy and economics, grid development, project bankability, and transaction support, backed by regulatory and economic intelligence across China’s solar, wind, coal, nuclear, hydro, transmission, and power markets. He has led work on policy forecasting and tariffs, renewable-asset due diligence, China business matchmaking, and green-power procurement for multinationals.

In our conversation, David unpacks how China’s decades-long planning underpins its energy transition and how renewables, storage, and grid build-out are looking to be able to meet AI-era compute demand. We also touch on China’s East Data West Compute and how it leveraged strong geographical planning, as well as discuss the cultural and commercial reasons behind the global retail adoption of solar energy.

For me, the most interesting point he brought up is that electricity used to be bound to scarce resources, but as the saying goes, the sun shines, wind blows, and water flows everywhere. Access to reliable power will become more evenly distributed, which can raise living standards in places left out of prior industrial revolutions - and Chinese technology is driving that change.

In today’s world, there’s no shortage of information. Knowledge is abundant, perspectives are everywhere. But true insight doesn’t come from access alone—it comes from differentiated understanding. It’s the ability to piece together scattered signals, cut through the noise and clutter, and form a clear, original perspective on a situation, a trend, a business, or a person. That’s what makes understanding powerful.

Every episode, I bring in a guest with a unique point of view on a critical matter, phenomenon, or business trend—someone who can help us see things differently.

For more information on the podcast series, see here.

Chapters

Introduction to China’s Energy Landscape

The Evolution of China’s Energy Demand

Nuclear Energy: Pros and Cons

Data Centers and Electricity Consumption

Main Drivers of China’s Growing Electricity Demand

Challenges of Renewable Energy for Data Centers

Geographical Dynamics of Energy Supply in China

Infrastructure Challenges in Southeast Asia

Commercial Reasons for Renewable Energy Adoption

China’s 2030 Renewable Energy Goals and Beyond

The Transition to an Electricity Civilization

Transcript generated by AI

Grace Shao (00:00)

David, welcome to Differentiated Understanding. Thank you so much for joining us today. I have been following a lot of your work on X and LinkedIn and you’re such a prolific writer yourself. And thank you so much for dissecting the industry, but really also breaking down the jargon on energy-related industry policies. So today I think we’re going to cover quite a broad range of topics, but really starting off from the high-level China energy planning, how it came about, and why their leaders are right now.

How that plays into right now, obviously, the energy competition within the AI boom, and then the companies that are backing these developments and growth, whether it’s like PV or solar. But yes.

Thank you so much. So to start off with, why don’t we start with your work and what you do at Lantau Group?

David Fishman (00:44)

Yeah, so I’m, I’m a principal, I’m a principal consultant at the Lantau group. We’re an energy economics consultancy. We’re focused on the commercial and economic aspects of the business of electricity or energy broadly around the region. And in China, I’m focused entirely on the business of electricity. That could mean either working with generators, right, producers of electricity, or those who invest in generation projects.

It could mean working with the markets, the grid, either the physical infrastructure of the grid or the commercial or virtual infrastructure of power markets that help connect power generators to power buyers. And then the ultimate user of the electricity, the end user, which could be a large producer of physical goods or IT infrastructure like data centers. Anyone who has a lot of exposure to electricity as a buyer or seller would be somebody that we work with in our business.

Grace Shao (01:40)

Perfect, so you are the perfect person to go to help us understand China’s energy buildup then. We always hear about China being the biggest emitter, But then at the same time, they’re now the leaders in renewable energy generation. How did that really come about? What’s the kind of dynamic there now?

David Fishman (01:56)

it’s all driven by the need for electricity consumption demand, which has been rising incredibly rapidly at the same pace as the Chinese economy, right, electricity or energy consumption really closely correlated with GDP growth. So as long as your, your economy is doing more things, it starts needing more energy as well. And we went from a period where, you know, the the pace of the growth was even outpacing the ability of energy sector players to meet.

that need. They were not able to build enough electricity infrastructure. They were not able to find enough primary energy sources to meet the growing demand. We’re talking about the 90s, the 2000s. And then in the last 20 years or so, we swung around more towards being in a position of relative abundance on the energy side, where it’s possible to have more energy than the economy is currently calling for. And that’s where we got to where we are now. We built up a huge energy

electricity generation base that was primarily powered by what at the time was the best option, which was coal. Coal-fired power became the backbone of the entire Chinese electricity grid. And because China is huge, whatever is the leading share of something in China becomes just massive in the world. It becomes massive overall. The renewables didn’t really come on the scene until about 2011, 2012 is when the first really strong installation capacity subsidy programs were put into place to really encourage generators to build wind and solar farms and started ramping up the scale of the industry overall. But at that point you already had tons and tons of coal fired generation that was, you know, the backbone of the entire fleet. And that wasn’t going to go away so quickly. So over the last 10 years, we can add tons and tons of wind and solar, but it doesn’t, you know, it only stems the growth of coal. It hasn’t really even started taking away from the total generation of coal fired power. So that’s how you end up having, you know, the largest coal generation sector in the world, the most emissions in the world, and the largest renewable sector in the world all at the same time.

Grace Shao (04:01)

But how did China become so dominant in solar and all the renewable manufacturing? was like scale, was it cheaper capital, cheaper labor, policy support? How do we understand this? Because right now we’re seeing that the US is talking about transitioning to a more, you know, green energy, but like economy.

David Fishman (04:18)

Yeah, well, mean, every to have an industry that is somewhat speculative, relies on new emergent or unproven technology. And in 2011, China didn’t necessarily set out to say, we’re going to go all in on wind and solar, and we’re going to count on this becoming world beating in the next 10 to 15 years. At the time, it was just, you know, we’ve built up a lot of production capacity or a decent amount of production capacity for for solar panels and wind turbines, and we’ve been exporting them.

And now we would like to start installing some of them domestically. But it’s not going to be so competitive or so profitable if we do it right now. So let’s make sure there are very comfortable, good incentives in place so that anybody who builds a wind or a solar farm will be guaranteed a good rate of return on their expenditures. So you start out by saying state support. We need to offer state support to incentivize certain types of things to happen that the market wouldn’t want to build on its own.

⁓ And then we need to be able to, you know, apply pressure throughout the value chain, wherever there would be somebody who’s unhappy about not getting an acceptable rate of return on their activities, generating electricity or producing solar panels or lending money to people who want to do this. Everybody needs to be incentivized to participate in this game. We’re trying to create electricity from wind and solar and it’s not very economically competitive right now. So how do we help them, right? That’s where you get Yes, your subsidies, your state support, your low interest loans, your affordable land access, things like that, all of those things, those help thing scale up, right? Those give you scale. And once you start getting that scale, you start to enjoy the economies of scale, you start to enjoy the effects of competitors on the production side entering into price wars to try to maintain market share. You get these jumps forward in innovation. I’m going to squeeze an extra 1 % out of my solar panels. going to I’m to beat the other guys, right? The scale turns into a bit of a snowball where all these other effective, enjoyable benefits, your economies of scale, you’re increasingly lower costs, you’re increasingly more attractive technology are all piling up. And all along the way, you’ve still got the state presidents hanging out in the background saying, will help, will make sure that your rate of return is acceptable. If things look like they’re getting out of hand, we’ll tweak things so that you still make enough money to keep yourself solvent. And then finally, one really important factor in all of this is the state owned enterprises themselves. The state owned enterprises are mostly involved in the capital intensive section of the industry. So that means building and operating the solar farms and the wind farms. That’s the really risky part of the entire value chain.

And so they’re willing to take on a lot more risk than a private developer might be able to. They’re willing to accept lower profit margins than private capital might be willing to. And that acts as a great big lubricant for the whole system so that you can continue keeping capacity numbers high even if rates of return sometimes are a little questionable.

Grace Shao (07:20)

So, how should we understand this, actually? Most of the energy players, whether throughout the supply chain, are SOEs or do we have major private players as well?

David Fishman (07:30)

Well, so the equipment makers are almost all private companies, right? Your solar panel manufacturers, your wind turbines, your batteries, those are all private companies. They get support from the state. They have subsidies, they have land grants and things like that, but they’re all private companies. And then on the generation investor project development side, you have mostly SOEs. Private companies are certainly happy to participate when they find good opportunities with acceptable rates of return. But for everyone else where maybe the rate of return is only a 6 % return on capital, right? That’s not attractive to private equity. That’s not attractive to a fund usually, but state-owned enterprises are happy to do that kind of work because that’s their mandate. So it really is a partnership between the private sector doing certain things and the state sector doing other things.

Grace Shao (08:21)

That’s quite commonly actually seen in China, Like across a lot of industries. How do we understand the grid mix right now in China?

David Fishman (08:28)

Yeah, so right for for 2024, which is the last year we have full full data sets for I think was something like 58 % coal fired power and that’s been declining as a percentage even as it increases in total volume, right? Because other sources have been growing faster than coal has been growing but 58 % coal and then we had wind water solar

Wind, solar together came out to, I think it was about 32%, something like that, with wind and solar being in the 18 % range and hydro in the 14 % range. Hydro has a lot of annual variability, depending on whether it’s been a good year for rainfall or not. And then the balance, that last 10 % is made up of nuclear, which about five, five, six percent. You’ve got gas to power, which may be two or three percent, and then other, includes things like biomass, waste to power, experimental technologies like that, but still very fossil fuels dominant.

Grace Shao (09:28)

How does that compare to other major economies? We might look at Germany or even the US.

David Fishman (09:33)

Yeah, it varies a lot depending on their natural resource endowment. So the United States has been shutting down a lot of its coal capacity and instead leaning heavily into gas to power. And gas to power is now one of I think it should be the largest generation source of the United States. There’s a lot of gas in the United States. Then you take a large country like ⁓ France, right? France has nuclear, built nuclear decades ago. Nuclear is still the largest contributor to France’s electricity mix. You switch over to somewhere like Brazil. Brazil’s got lots and lots of hydropower and hydropower is the main driver of Brazil’s economy, of their electricity economy. you know, natural resource endowment really matters. Major countries that are still using coal a lot, like China, you’ve got places like China, India, ⁓ Indonesia, those are the ones that come to mind, especially in East Asia.

And then throughout Europe, course, Poland is well known as being a huge, huge coal consumer. But if you’ve got access to gas, you use gas to power instead. South Korea uses a lot of gas. Currently, Japan uses a lot of gas. Gas is very common throughout Southeast Asia, usually imported gas or LNG. So it’ll be quite, quite expensive if you’re using that for electricity, which can sometimes contribute to high power costs, lot of gas in Europe as well.

Grace Shao (10:51)

That was really good context. I actually want to double click on the nuclear topic. It seems like there’s quite a bit of controversy with a lot of the SMRs being like the small ⁓ nuclear plants that are being reactivated right now. How do we understand, I guess, the pros and cons of nuclear? Is it really still quite dangerous for local communities or is it like potentially a big possible solution for us as we’re seeing electricity shortage?

David Fishman (11:16)

That’s like three different, very different questions. So I’ll try to answer them in order there. So SMRs are next generation kind of experimental ideas for nuclear that they could be maybe more cost effective or more flexible in a smaller format, something like 100 or 200 megawatts instead of a thousand megawatts. Right now there are only a couple of SMRs operating around the world for commercial civil power use.

⁓ China’s got one, for example. and mostly in the United States, the conversation has revolved around restarting some previously retired or mothballed but not shut down or decommissioned, nuclear power plants. That now that there’s this energy crunch, an electricity crunch that some of that retired generation that maybe wasn’t competitive in the market landscape of whatever year it was, decommissioned in or retired in, mothballed in,

Currently the market climate has changed that there’s such demand for electricity now in such a scarcity that the buyer of the electricity is willing to finance the refurbishment of the plant and willing to finance the long-term operations of the plant by becoming a buyer of electricity. So when you look at your major tech companies that have just signed an agreement to restart a nuclear power plant, that’s because the operator, the owner of the nuclear power plant has secured a very lucrative contract to sell electricity from that power plant to that data center for a long period of time. On the safety aspect, yeah, look, we’ve had a couple of notable kind of headline grabbing, world attention grabbing accidents over several decades of the nuclear industry’s operations. I always consider it to be kind of like one of those airplanes versus cars thing, right? A lot of people are afraid of flying, although it’s incredibly safe relative to driving on a statistical basis, right? But you know, an airplane crash grabs the headlines in a way that a car crash never will. And that’s, you know, a similar situation with the safety of different energy types, right? A nuclear power incident once every several decades grabs headlines, but the long term damage to human health and livelihood caused by combusting fossil fuels has been immensely larger, incredibly, incredibly larger. So everything has its own kind of trade-offs and how you evaluate it is up to you, but I do invite everyone to think of it as an airplane crash versus a car accident risk scenario.

Grace Shao (13:40)

That’s a very interesting way of framing it. I think like you said, a lot of times the headlines really focus on what’s big or what’s more exotic right? And that’s kind of in the case that nuclear is just not as commonly heard about, therefore everyone actually pays attention to it more. I really want to double click on what you just talked about on the big headlines of big tech buying up, or reviving plants to power data centers. Because when we were talking before this interview, you joked you said,

Hey, look, I’ve been an energy guy forever. No one really wanted to talk to me that much before. For now, all of a sudden, everyone wants to talk about energy, right? We are in the midst of an AI boom. And the bottleneck, especially for the US, or the choke point right now, for a lot of these big tech companies, is securing enough energy to power data centers that are required to power their training or influence whatnot, right? So help me understand this. The AI boom is capital intensive.

So is energy. What, how do we understand the relationship actually between energy and data centers and AI right now? Why don’t we start with that big question?

David Fishman (14:42)

Yeah, well, mean, there are a few other productive activities in the world that rely so much. On electricity as a primary input as the operation of chips in a data center. I mean, maybe the only thing that’s similar is Bitcoin, right? Mining Bitcoin because the operations you’re doing are just constant calculations that require consumption of electricity to be performed. data center operations are similar. In the physical world, the only thing I think comparable is something like ⁓ non-ferrous metals smelting, like aluminum or copper smelting is also just, we call it solidified electricity.

That’s what aluminum is. So data centers are constantly performing actions and tasks that use electricity. It’s a direct relationship to do what it needs to do to be productively useful at all. needs to consume electricity. so from an economics perspective, you just say this is a demand driver of electricity, almost a direct demand driver of electricity. It’s not an aluminum smelter, it’s a data center, but we need more electricity to serve its needs so that it may serve its function, its customers, which are asking for computing power. And so from the electricity sector perspective, I just see that as a number that used to be 1 % year on year demand growth, and now it becomes 3 % year on year demand growth, or 5 % year on year demand growth.

And that’s something where, if we build one or two generation assets, we will meet the anticipated demand for the future. Now it’s, need to build eight generation assets or 10 generation assets, or I can’t allow that large electricity user to actually connect to the grid and start demanding power. Cause I don’t have enough. I don’t have enough electricity to serve them as a grid operator. Maybe that’s my perspective, right? You’re not allowed to connect to my grid, you need too much electricity. And when you’re in that circumstance as a data center operator, you’re saying, well, I got to bring my own power, right? Bring my own electricity. And so that’s kind of the way they’re thinking about it now. If I want to set up my compute in an area that, you know, for whatever reason, it’s beneficial for me to be sited here, but this local grid doesn’t have enough electricity for me, I got to come up with my own solution. I could build my own assets. I could have a captive power plant, perhaps I could support ⁓ local generators to build something and sign a long-term contract with them so that they’ll build the asset, I'll build all the power. Or maybe I’m looking at something like, you know, something more creative where I say, Hey, there’s already an asset. It’s in the grid system. It’s nearby. It’s this nuclear power plant that’s been mothballed for 15 years. Let’s get that going again. I’ll buy all the electricity. Whatever it is, it’s it’s, you know, because of this very strong direct relationship.

A data center operator is a wonderful customer for an owner or operator of a power plant, right? I’m in the business of making electricity, you’re in the business of using electricity, like, surely we can work something out. So that’s where you how you end up with this kind of like, we call it a PPA, a power purchase agreement. That’s the direct relationship that’s established between the generator and the consumer.

Grace Shao (17:48)

Who are you seeing as biggest consumers of power right now amid this AI boom that you’re seeing in China specifically.

David Fishman (17:54)

Well, yeah, it’s going to be your big Chinese tech companies, right? Your Tencent, your Alibaba, and ByteDance, companies like that that just have massive need for tech, you know, and then your emerging AI companies. And of course, the big tech companies have their own AI plays always. And then they’re going to be independent or third party AI companies that are looking to train or whatever it is. And they’ll

They’ll need a lot of that too. So in the IT space, that’s who you’d expect. The big tech companies and your frontier model creators.

Grace Shao (18:27)

And what do you think, if you had to put a number on it, how big is this AI driven load as a percentage in terms of the incremental electricity demand we’ve seen, I guess, in the last three, four years and comparing it to next three to five years of production?

David Fishman (18:40)

So it’s interesting, it grabs all the headlines, but remember, Chinese power consumption is already growing at a stunning rate for every other reason already. So this is just one additional thing. So in the last five years, think it’s been ⁓ a modest amount of the growth can be attributed to ⁓ data center needs.

⁓ Looking at something less than 15 percent I’d say maybe 10 10 15 percent of the growth can be directly attributed to data center needs And then over the next five years I’ve seen forecasts from like state-grade energy research Institute where they’re saying you know of all the different sources that are driving power consumption growth Maybe maybe 20 % will be attributed to to AI so or to data center so not not like nothing, but also not as much as you guess or expect, certainly not compared to other countries where it really is maybe 50 % of the load growth or more can be attributed in that way.

Grace Shao (19:36)

I see. What are actually the top drivers for China’s demand, growing electricity demand, other factors?

David Fishman (19:42)

Yeah, so among those, the same report that I saw that said, you 20 % can be AI. In that same report, it was another 10 to 15 % or so. it was 10 % is charging for EVs as a driver of growth. And then another 10 % was electrolysis of hydrogen. So you can produce hydrogen by running a current through water, essentially electrolysis of hydrogen, which replaces our other ways of producing.

Grace Shao (19:57)

Okay.

David Fishman (20:09)

hydrogen, which are usually quite dirty and fossil fuels intensive. And hydrogen is very useful as a as a input for for production of ammonia or methanol, many other very useful chemicals. So that’s about 40 % of the growth was attributed to those three things, these three emergent sectors. And so the other 60 % of the growth will be traditional sectors. So that’s your heavy industry, and uses tons of electricity all throughout it. That’s your services or your tertiary industry. So you’re looking at growth of AC use in shopping malls and hotels and things like that. And then the smallest portion of that is residential power use. All three of those traditional sectors are all rising in China as well. And in addition to these new three sectors that are emerging as part of the clean tech revolution.

Grace Shao (20:58)

That’s really interesting. I wish we hadn’t time to go into the EV talk today, but see if we do later, but let’s focus on AI first. I actually want to understand why can’t we go full renewable with these data centers right now? What’s the issue with, you know, using renewable solutions instead of traditional solutions? Cause there’s been a huge debate around that, right?

David Fishman (21:16)

Yeah, well, mean, and so when you have a power load, doesn’t matter what it is. If you have a power load that draws on power pretty continuously, the easiest way to meet that kind of load will be something that generates power pretty continuously. It’s not impossible to meet it otherwise. It’s just the easiest, the most straightforward way to meet that load. If it has certain types of flexibility in the way it operates, then you can also start to address it with more flexible generation sources. But overall, a ⁓ data center load is a more stable large load. And large stable loads are really good matches for our conventional generation sources. Our hydropower short, but you know, coal, gas, nuclear, great matches for the way a data center needs electricity. Now we can meet those needs with a combination of solar and wind and storage and flexible generation sources like that. But operating those in that way generally incurs a larger cost, either a direct cost in terms of operating cost of the assets or a systemic cost.

And in order to have all those variable assets in there, we also needed to keep some gas generators on standby. We needed to pay some capacity payments to a battery station, something like that. Those are systemic costs that also usually end up being assessed to someone, ideally the one who caused the need. But sometimes it’s assessed to an end user who didn’t cause the need for that variable generation, but they still paid for it. So when we look at can we meet the ⁓data center needs with renewable, certainly we can. It’s just more complex and it takes more planning and maybe it incurs more system costs and sometimes the system costs aren’t really well tracked. So China does intend to start meeting a lot of its data center growth with renewable sources. We talk about where the new data centers are going in, the parts of the country that they’re going in and there’s this program called the National Hub Nodes.

Which is part of the East Demand West Compute program. And so under this scenario, any new data centers added into those national hub zones need to be consuming at least, I think it’s 80 % of their electricity needs to be renewable. It’s a pretty high percentage. So if it weren’t possible, there wouldn’t be such a requirement, but it is trickier, it’s more complex. And so that’s how China is proposing to drive most of its data center build out now is with these renewable energy sources. It’s a way to avoid this growth of a new load sector resulting in increased fossil fuels consumption.

Grace Shao (23:57)

I see. I’ve heard about like, you know, the challenges with battery solutions right now for the intermittent kind of nature of renewables. Do we have actually strong enough battery solutions right now to solve that issue?

David Fishman (24:10)

I mean, batteries, the question is, can you you scale it cost effectively? Right? Can you can you get your battery storage to be able to cover what two hours, four hours, eight hours? How long of a backup solution do you need there? How what type of gaps are you expecting to need to meet in the context of all the other power that you’ve contracted for or that you have available to you? If if you’re in a region where there’s tons and tons of solar power, it’s sunny all the time. The sun is up most days and it’s producing well most days. Okay, well, you’ve got a solution for part of the day. And then it’s like windy in the evening. So you’ve got a solution for the evening. But how do we cover that evening peak period when it’s not particularly sunny or windy? Okay, batteries, sure. How long are the batteries good for? Can we be sure that the batteries will be charged and ready with no loss of load incident every day? Or at least at a very, very slow failure rate, something like that. When you start doing some type of probabilistic assessment, how many batteries are enough so that we can ensure a failure happens once every 10,000 operating years or something like that, whatever your threshold is, you start to realize, woo, it’s a lot. I need a lot of backup. I need a lot of storage. And so maybe that’s where you run into some of the problems where the project economics might not work. They might not make sense anymore. Ideally, I’m describing an extreme scenario, but ideally you can find a solution that involves not using massive, massive amounts of batteries that you barely use most of the time.

Grace Shao (25:38)

I see. It just sounds like there’s a lot more operational risk, right? I want to understand East data, West compute. From your perspective, like how should we understand that, I guess, from the energy framework? And can you help us understand the geographical importance of having the energy suppliers on Western regions of China?

David Fishman (25:57)

Yeah, it’s absolutely driven by geography. maybe you’ve seen the famous line before, you can draw a line through the center of China and everything on the west side of the line is 7 % of the population and everything on the east side of the line is 93 % of the population. It’s roughly half and half of the country. That’s just where all the Chinese people, all the load, all the industry is, is along the coasts in the east and the south of the country.

And so your need for electricity, your need for, your demand for any energy use of any kind is mostly going to be out there. But the good energy resources in China are in the West. They’re in the North, the Northeast, and the Northwest, and the West. So you’ve got to find a way to get them to each other. You want, you know, the wind blows all day, the sun shines all day, and the mountains are just heaped with coal in Western China.

So we got to connect them to Eastern China. In the case of data centers, you’ve got two options, right? Either we’re going to bring power lines from the West to the East, or we’re going to bring fiber optics from the East to the West, right? So either I bring the electricity to where the load is, or I send the compute to where the electricity is. And they are doing a mixture of both. So this is where you end up with a bifurcation of the types of computing needs that you have.

The short term, very rapid response compute needs to stay in the east, close to where the demand is. And so for those, we bring the electricity to the compute. And then for like a longer lead time, you know, maybe we’re training a model or something like this, we can send it out to the west, send it closer to where the cheap electricity is. So this is a consideration that makes sense for China, because China is huge and it’s got very different ⁓ dispersion of its resources. A smaller country could do it on a smaller scale, of course. I know in the UK they joke about London data computed in Scotland or something like that because they’ve got the offshore wind up there in Scotland. So it’s a similar concept, just in China’s case it has to happen on a continental scale.

Grace Shao (27:56)

I’m also interested in actually how the relationship between the provinces are. How do they work together? Do they have different mandates for each of the provinces or is it just a very top-down kind of mandate from the federal level?

David Fishman (28:10)

Yeah, so in China, is a tiered, a hierarchical relationship between kind of national government, provincial governments, and municipal governments. Generally, you’d expect planning at the more macro level to come out of Beijing for the whole country, and execution is going to be left to provinces and municipalities. So when they say you need to enable cross-regional power transmission, okay.

Like that’s that’s something that comes out of Beijing, but now it’s up for the individual provinces to work out how to do that. Eventually, they’ll execute their trades through the state grid power exchange, which is up in Beijing. But all the negotiations and all the interactions have to happen at the provincial level or even lower at municipal levels. That’s that’s broadly true for for almost all Chinese policy, not just the power sector and not just data centers, but in general, broad strokes laid out at the top and then executional happening at provincial or low.

Grace Shao (29:05)

That’s interesting to hear. I kind of want to shift gear and kind of double click on the private sector right now. You you just talked about along the supply chain, lot of the coin makers are actually private companies. We’ve got the PV makers, the battery makers. think, you know, people in the West will probably know of CATL, Longi, Trina. There were some also even international ambitions of these companies kind of set up manufacturing hubs in the US or even across Southeast Asia. Can you kind of help us understand their global ambitions and where they’re at now.

David Fishman (29:34)

Yeah, mean, so the initially the Chinese market was very initially, they were built in China to be sold internationally. And then the Chinese market got so much larger than everywhere else that many of these equipment makers had plenty to do just selling to the Chinese market. But because of the very, very stiff domestic competition, there’s always that interest of like, well, could we could we diversify a little bit out of this incredibly competitive environment by selling?

broad as well. And so initially they could sell abroad from China again, as some of those trade barriers started going up that really incentivized the expansion into Southeast Asia, into Vietnam, for example, that you could have Trina solar panels produced in Vietnam and exported to markets that were maybe closed, or tariffed for for Chinese exports. And then we saw, I mean, with the United States and its recent tariff policies, that tariffs were placed on the Southeast Asian countries as well. And then specifically if there was evidence of trans shipment of Chinese panels heading to Vietnam and then going to the United States that they would be taxed or tariffed even more. So a lot of that has, you know, eventually ended up coming down to just one place, right? If you’re a Chinese solar producer, and you want to sell to the United States, you’d better be ready to produce in the United States. That seems like the way I’ve talked to a couple of them, and that’s the way they’re thinking about it. If they had production already, they say, okay, we’re gonna keep producing for the American market from our production facilities there. If they’re thinking about starting a new, opening a new factory, well, it’s been a scary environment for that kind of thinking recently, right? Of course, there was the recent⁓ raid on a South Korean under construction battery facility in the United States, right? There’s a lot of concern about, you know, that kind of capex heavy investment for a market that just maybe isn’t geopolitically that friendly, and it would come back to bite you in the end. There was another case, I think, where a Chinese solar panel producer built a facility in the United States, and then it wasn’t going to be eligible for the IRA tax credits and they were forced to sell it to an American company, which is now producing solar panels from that facility. So if you’re a Chinese solar panel manufacturer or battery manufacturer and you’re looking at this kind of geopolitical climate, certainly you’re willing to consider building an overseas factory to get around tariffs, whether it’s in Southeast Asia or even in your target market. But some markets seem like they’re a little bit more geopolitically favorable than others right now.

Grace Shao (32:08)

It’s obviously quite challenging for a lot of these companies trying to sell to the US right now. Just like you said, given the kind of backdrop. Okay, I want to ask about Southeast Asia. We know that a lot of these companies are building out, you know, manufacturing hubs, even data centers in Southeast Asia. A lot of the big tech companies, Chinese big tech companies are now said to be the biggest hyperscalers across Southeast Asia as well. Like they’re buying up all the data centers.

But there are some concerns around infrastructure challenges in Southeast Asia, right? So there have been complaints about, you know, the utility prices rising. There’s issues in Johor, Malaysia, where frankly, local infrastructure such as water, land, roads, it’s not really there yet. How can you, how do you help us make sense of all of this? Like, who are the buyers? Who are the investors? Who are the users? Is this really helpful right now? Are they actually helping the local economy? And are they actually, are these companies getting what they need in those regions?

David Fishman (33:00)

Yeah, so anywhere in the world, including China or even very, very transparent power markets ⁓ like parts of the US or in Europe, it has proven to be remarkably difficult.

to attribute changes in the price of electricity to any one thing. Data centers have become a ready and available target. And I think it’s surely true in some cases that it is the thing that’s driving higher power prices for that region or for that node. But in other cases, I think it would be more difficult to make that assessment. And it’s something that is speculated in the media and then kind of becomes true by default without really having a good investigation into it. I haven’t looked at Malaysia or Vietnam enough to comment whether I could attribute the data center, no matter who owns it, the data center demand to rising power costs. It does make sense usually in an economic sense, right? Your demand load has increased, your supply hasn’t changed, and so we end up at a higher point on this cost curve for more periods of, you know, throughout the day.

So that’s, mean, it’s something that is theoretically a problem. If it gets to the point where the data center load starts to really affect the competitiveness of electricity for other parts of the economy, either the industrial, other industrial segments are saying our energy inputs are too expensive now, or you’ve got residential power costs skyrocketing, you’ve got a developing economy, and you’ve got, you know, residential people can’t necessarily afford the costs of higher energy costs, then you’d expect the national policymakers, regulators to step in and say, if you’re going to be allowed to operate in this way, we need you to bring your own electricity, for example, ⁓ or you need to be willing to cover the increased costs of other sectors of the economy through some type of ⁓ payback scheme or kickback scheme, right? That kind of thing is a reasonable thing to ask of a data center operator when they’re coming in and creating those types of problems for a generation overall. But I mean, remember, they are also creating their own economic benefit. They’re not just leeches on the system, which I think many would argue something like cryptocurrency mining is that you’re just sucking up tons and tons of electricity, and you’re only creating value for a smaller group of people who believe in the value of the cryptocurrency. Data centers, on the other hand, are performing tasks that are considered to be productive and valuable tasks for their clients, for their customers.

Maybe you could argue you think that using AI video generation is not a productive and useful task, but hey, like somebody thinks it’s valuable and useful. Enough people think it’s valuable and useful. So that’s what you weigh the two against each other, right? You’ve created a burden on the grid system, a burden that is having knock-on effects on other parts of the economy. You need to account for yourself. You need to take responsibility for the burden that you’ve created on the system.

But also, you you’re not a leech. You’re not a parasite on the system. You are in fact doing a useful thing and it’s the responsibility of energy planners and economic planners to try to make energy available to you. So it’s a delicate balance. And it’s, I wouldn’t, it’s not fair, I think, to attribute blame to anyone unless they’re intentionally trying to get around policy or screw over other people.

Grace Shao (36:28)

Yeah, I have a question that’s a bit more anecdotal, but when I drive through Europe or China, right, you see like solar panels on private citizens houses across the countries, right? Like, you know, in Spain, Italy and Germany, you really see across China, like Hebei in Songsu, you really see people embracing it as a society. Can I do you think there’s a cultural difference in terms of kind of embracing renewable in the US versus China or Europe or what kind of incentivizes them.

David Fishman (37:02)

There’s a cultural difference, but in the case you cited there, there’s a very strong commercial difference as well. In Europe, residential solar is primarily motivated by the decision making of the people that own the house, that own the rooftop where the residential solar goes in. In China, rooftop solar is driven by developers who don’t see the they see the rooftops as real estate.

So developers go in, they knock on doors and say, I see you have some nice unoccupied rooftop space up there. I will pay you to lease your rooftop. I would like to put solar panels on it and sell electricity back to the grid. So the Chinese business model is very unique in that case. I would say the United States and Europe are more similar in the ways that the mechanism for motivating, motivating residential installations. And then China is off doing its own thing because it has this incredibly effective, but unique scenario where developers are the ones promoting solar panel development to residential users who weren’t very aware of, you know, solar energy at all. And then also, they don’t even have much electricity consumption, frankly. So they’re not super motivated to go install solar panels. ⁓ residential electricity costs in China are very cheap. So it’s not like they’re trying to offset a high power bill.

The whole motivation for the residential sector in China is ⁓ collecting a rent on their rooftop space so that some developer can generate electricity and then sell it to the grid.

Grace Shao (38:31)

I see. I totally thought it had something to do with just kind of this overall societal embrace of new technology and this 2030 renewable energy goal. Can you actually, I guess the last question I have for you, can you help us understand what is this grand 2030 goal of China to really convert, you know, transition majority of the energy consumption to renewable? How do the EV sector, the battery sector, and various sectors related to energy really play a role in this?

David Fishman (38:58)

So the 2030 goal is the peaking goal, the emissions goal. So they want to peak carbon emissions by 2030. And then they say carbon neutrality neutralize the emissions by 2060. So we got to reach a peak, then we got to draw down and neutralize. So by 2030, I mean, maybe we’ll be down to a 50 to 55 % share of coal in the power sector. I don’t know if it will slip below 50%. By that point, it’ll be getting close to it.

And then you’ve got all your other non power sources of emissions, right? Transportation, building, heating, and industry. That’s the major one, industrial inputs. So by 2030, the goal was to just peak the emissions across all emissions generating sectors. Now it looks cautiously like we might have reached an early peak this year. I say cautiously and I hedge my phrasing here because, well, there’s lots of different segments that create emissions.

Power is maybe gonna stay flat, maybe gonna be, you know, peaked by 2025, 2026, but we’ll see what happens to new capacity for wind and solar next year. If wind and solar and hydro and nuclear can’t keep up with the pace of consumption growth, then the only way to meet the additional consumption growth is by using more coal, right? So that’s the power sector. Fingers crossed that they managed to figure it out. Transportation fuels maybe is going to peak in this year or next year. We’re talking about the EVs coming in such massive numbers that they’re replacing petroleum ICE vehicles.

Grace Shao (40:28)

What’s the percentage actually? What’s the percentage of EV cars on the roads right now, do think?

David Fishman (40:33)

It’s you know, it varies widely by province in Shanghai. It’s over 50 % now for sure. Major cities is 50%. But you know, I was recently in, like Shanxi, a northern, you know, Chinese province, and it was probably less than 10%. ⁓ So it’s I don’t know what the number is nationally. It’s it varies so incredibly by city, I think it’s or by province, I think it’s best to look at that on a provincial level. But the idea being that new

Grace Shao (40:38)

Wow.

David Fishman (40:57)

New EVs should have already started outpacing the sales of ICs and that should continue so that we would be able to, know, more vehicles are going on the road but the EVs are being added more rapidly and that petroleum consumption for passenger transportation should be peaking, you know, within the next two years or so. And you’ve still got aviation and maritime fuels and long-distance trucking and all the other things. Excuse me.

And then finally, you’ve got the industrial sector and there’s different aspects. Some are growing, some are rising, know, steel is maybe flat, cement is dropping for the next 15 or 20 years, but hey, petrochemicals and coal chemicals are rising for the next 15 or 20 years. So you’ve got all these different, you know, sources of emissions to consider. That’s what’s all going into that 2030 goal. So if they manage to peak by 2030, like wonderful, lovely, enough of those sources have peaked and started drawing down that we can make up for the ones that have not peaked and that are not drawing down. And then we’ve got another 25, 30 years beyond that to figure out how to get to neutrality after peaking in 2030. If they happen to incidentally peak this year or next year, I don’t think they’re gonna claim it. I don’t think they’re gonna crow about it. I think gonna stick to the 2030 emissions peak, give a little bit of time just in case a little bit of buffer.

Grace Shao (42:08)

Really?

David Fishman (42:13)

But it seems very likely that we’re flirting with the peak a little bit early versus their own targets.

Grace Shao (42:20)

What’s the practical impact on society and the economy then?

David Fishman (42:24)

Well, the ideal goal is to have as little impact as possible, right? That business as usual growth as usual continues, economic abundance of energy is compatible with energy transition. That should be the goal that the only outcome in fact is that the business of the energy transition is good business, that it contributes in fact to GDP.

⁓ installing solar farms and building wind turbines and upgrading industrial equipment is all like good economically productive activity. And so that should be the only impact. That it’s a good impact and that otherwise everybody just goes on driving cars, they just happen to be EVs now. They go on using electricity, it just happens to be increasingly clean electricity. They go on using steel, it just happens to be green steel. That’s the best case scenario.

Grace Shao (43:16)

That’s a good world to be looking forward to. Okay, David, thank you so much for your time today. What is one differentiative you you may have? It could be about your industry or it could be about something else broader in life.

David Fishman (43:28)

One differentiating view that I have.

Yeah, well, I mean, I’ll talk about my other great passion, which is kind of economic development in general. I got into electricity because I cared about development. And I stay in electricity because I continue to see it as important for development. And I am pretty agnostic when it comes to the different isms of the world, the different political science factions that say it’s this ism or that ism.

jWhat works is the best ism, whatever it is, I don’t care what label it has. And my observation right now about where we are as a civilization is that we are right on the precipice of moving from the fossil fuels phase of humanity to the electricity phase of humanity that electricity will be an interim period, because eventually we’ll probably figure out fusion power and harnessing plasma, and that will take us to the stars. But in the interim phase, we’ll be in the electricity age. We went from biomass to mechanical energy to fossil fuels to electricity generated from the natural elements. And that this electricity phase should be the phase where energy stops being additive.

And starts being substitutive instead. All the previous energy phases that we went to, we had to use more energy to upgrade. But to upgrade to electricity should allow us to start using less primary energy. That using wind and solar and hydro and nuclear are less intensive for primary fuels, which means we’ve freed up space for developing nations in the world to start to enjoy a fraction of the energy abundance, the civilizational abundance that so much of the rest of the world is already enjoying. They’re already there, right? As long as we stayed locked in the fossil fuels era, they’re not allowed to enjoy that abundance. They’re locked out of it because we have no more carbon budget to spend. We need to work on this, you know, civilizational existential challenge of transitioning our energy. And for the moment, there’s no more carbon budget for the developing countries of the world.

Getting into the electricity era, the electricity civilization era, frees up that carbon budget, enables them to start pursuing the same type of abundance that everyone else is already enjoying in developed nations.

That’s the great thing that motivates me throughout all of this, that the electricity civilization era is one where regardless of what ISM introduced it or ushered it into the world, is one that is going to be incredibly beneficial ⁓ for humanity. And at this moment, I think I’m working in the country that is most likely to usher in that era. And here’s the controversial portion, right? I think China is the one to bring the world right now into the electricity civilization era. And that will be such an incredible boon for humanity in general.

Grace Shao (46:27)

That’s a really, really interesting way of framing it. I really appreciate that. Thank you so much for your thoughtful answers and just your time and your insights today. I really appreciate your time and just your generosity in sharing with us.

David Fishman (46:38)

Thank you for having me. It was a great pleasure.

AI Proem is a reader-supported publication. To receive new posts and support my work, consider becoming a free or paid subscriber.

Get full access to AI Proem at aiproem.substack.com/subscribe