Thomas Healy, Hyliion @ World Economic Forum || "Harnessing Technology for Tomorrow’s Energy"

Good afternoon. I'm Arunabha Ghosh. I'm the CEO of the Council on Energy, Environment and Water, which is India's leading climate and energy institution and among the 20 best climate think tanks in the world. Very delighted to be here to welcome you along with our partners for this session, Agility, which is a global leader in supply chain services, infrastructure, and innovation with 65,000 employees across six continents.

The session we have today is called "Harnessing Technology for Tomorrow's Energy." But if we unpack that session title, we actually ask ourselves two types of questions: one, with the technologies we have today for the energy transition, how could we deploy faster? What are the governance mechanisms needed to make sure that we're doing it responsibly and equitably? And the second question that arises is with the technologies that are emerging on the horizon, are we designing them to be sustainable and to drive the sustainability agenda as well?

Now think about it this way: my country, I come from India. India, over the last about 14, 15 years that we've been around as an institution, has very quickly become the world's fourth largest renewable energy market by deployed capacity—over 150,000 megawatts of renewables, over 200,000 megawatts of non-fossil capacity. We have to go to about 500,000 megawatts of non-fossil capacity by the end of the decade. But one of the things we did was thought about: Where was the land? Where are the population concentrations? Where might there be water constraints? Where might there be social conflicts against big infrastructure, and does that slow down or does that help us deploy renewables more responsibly? And we find that India alone has a potential, despite these constraints, of over 7,000 gigawatts of renewables.

It's this kind of hyperlocal assessment that is needed across the Global South, which will drive at least 90% of global energy demand over the next couple of decades. But then we also think about the technologies that are emerging on the horizon. Of course, walk around Davos and you'll hear about AI in pretty much every conversation. You hear about the demand for energy coming from AI, but you also hear about the potential leveraging AI and other emerging technologies to optimize our energy use better. Which way will we end up is determined in part how we design the technology and in part how we experience and live it and how we design the governance of these technologies.

It's to a debate on these issues that we have a stellar panel with us. We have Joanna Messing, who is the CEO of the Growald Climate Fund. We have next to her Thomas Healy, who is founder and chief executive of Hyliion, which is also a company in which Agility has invested. After Thomas, we have László Varró, who's the vice president for strategy insights and scenarios at Shell International. And finally, we have J. Carl Ganter, the co-founder and director of Circle of Blue. We don't have only technologists on the stage here. We've got journalists, we've got analysts, we've got technologists, innovators, investors. We want to have this kind of a multidisciplinary conversation around the issue of harnessing technology in a responsible way to drive the energy transition forward.

So let me start with you, Joanna. You have often talked about a double transition that is needed in energy: on one hand, driving renewables growth; on the other hand, pulling back from fossil energy. We know, at least on paper, renewables plus storage is outcompeting fossil infrastructure in many geographies. And yet, the pace of deployment of clean energy infrastructure is not as fast as we need. What's holding it back? How do we get more investment to flow into geographies where the sun shines the most?

Absolutely. Thank you, and thanks for this session today. I think it's really what you said: it's all about speed and scale. This transition is happening, but will it happen fast enough to meet climate and development goals is the main question.

So there are several things unfortunately holding back this transition at speed and scale. Policies that are incentivizing and enabling transition—permitting in particular is something that can take quite a long time and can be the difference between profit and not profitability. Profitability is another thing: just because the economics work and the costs work does not necessarily mean that an individual project or company is going to be profitable, particularly when you layer on emerging economy risk, some of which is real, some of which is perceived, which makes the cost of capital for many of these projects prohibitively high, again, for speed and scale.

And grid expansion and modernization is another area where I think there's actually quite a lot of space for technology to make sure that it is catching up and up to speed with the deployment of renewables that's required. And then going back to policy, PPA and PPA reform is another major thing holding things back.

However, even with all of these barriers, we see so much opportunity, particularly around financial innovation: thinking about ways to reduce the risk of individual transactions through securitization or through bundling, through aggregation of deals, through the creation of new types of funds that are specifically organized to resource these, through the support of domestic financial institutions that are not carrying the currency risk or the emerging market risk. We also see a lot of opportunity to update the subsidies and policy reform to enable this transition.

And finally, I would say there's a real opportunity around updating the modeling that we're using to assess the financial viability of these projects. A lot of times, the models are backwards-looking, which look at the last 10, 20 years of performance rather than the next 10 to 20 years when we layer in—we're going to be talking obviously about water risk and some of these other risks which will increase costs, climate impacts—and then also not always updated around what are the actual costs and financial potential of solar. So I think there's a real opportunity to update the ways that we're assessing these deals.

I think you've highlighted quite a few of the sort of the nuts and bolts problems that you face when you're trying to actually get projects on the ground. And this difference between what is economic viability and corporate profitability is something that's really interesting, and we should perhaps unpack that a little bit more.

Thomas, let me turn to you. Again, just on that point about corporate profitability, can you tell us what are the technologies that Hyliion has been developing and commercializing for sustainable, low-carbon solutions? I mean, I think you began with the transport sector, but you've expanded well beyond. So what I'm interested in unpacking is: What's the new driver here? Is it the many end users of clean electricity, which you see as an opportunity? Or is there an opportunity to use energy as the sort of spearhead to trigger other types of innovations in transportation systems, in industrial systems, and so forth?

Both you and Joanna have, either explicitly or implicitly, referred to regulation. Just hold that thought; I want to come back to both of you, because is regulation then holding back this kind of more distributed, democratized energy system that we know technology can provide and yet often doesn't get deployed?

But I'll come back to you, László. Turning to the international, to what we've been looking across the globe: you at Shell have been looking at various different scenarios which understand the impact of AI and digitalization on the energy system, on its evolution, and maybe the other way around as well—how not just the demand for energy, but how AI can then help optimize for the way energy is used. Tell us what you have identified, what are you understanding as clear trends, but also what are some of the unknowns that you would still like to unpack?

So in the Shell scenario team, we follow the guidance of the great British economist Keynes, that "when the facts change, I change my mind." We don't operate according to the fixed schedule, but we very much felt that the world has changed with the emergence of AI technology. And so we sat down to analyze and model, and we are going to publish it soon, of how AI is reshaping the world economy and the energy system. And our main conclusion is that it does lead to a significantly higher electricity consumption, but the big story is not the electricity consumption of the data centers; the big story is how a new general-purpose technology is enhancing productivity growth.

Because this excitement with AI, it presupposes that this time for real—that this is not like the blockchain hype five years ago or the metaverse hype five years ago—this is a genuinely new general-purpose technology with comparable impacts to electricity 100 years ago.

Now, different societies have different capabilities to take advantage of the new technologies. We actually identified India as one of the countries which is well-positioned to enhance its development from digital technologies. And 200 million more Indian families in the middle class—that's where the big demand number comes from. So basically, the primary impact on the energy system is the way that it leads to increased prosperity.

Now, the electricity consumption of data centers is a big story because the electricity system in North America and Europe had 15 years of stagnation. Before that, electricity was growing dynamically in Europe and the United States as well. In fact, if you look at the demand projections from 10, 15 years ago, US electricity demand is still surprisingly low rather than surprisingly high.

And if I take—there are different projections on the AI-related electricity demand—with a bit of a summary, the current opinion converges towards: by 2030, a whole Germany; and by 2050, another United States emerging. Now, that is meaningful, but that's well within the capability of the industry to supply if the industry plays its cards right.

Now, where things can go wrong, of course, multiple places. One thing that we announce is that whether there is actually a societal acceptance of the new technology, and that differs from society to society, and that will be one of the big distinguishing factors. And also how society will want to use that. Because let's not forget that both for digital technology and even for solar—I remember from my childhood growing up in Communist Eastern Europe, the Soyuz-Apollo spacecraft, it was the Soviet aircraft which had solar panels, right? One of the early scientific breakthroughs in solar power were done by the Soviet space program, but they did not use this technology to enhance prosperity. They used this technology only for military purposes.

And given that there is a frighteningly frequent discourse of a Cold War 2.0, there is also a possibility that the new digital technology is primarily channeled towards defense-related applications rather than towards growth and prosperity and clean energy. So that's also something to watch.

Well, I mean, all technology can be put to good or ugly purposes, and that's where I guess governance and human agency comes in. But if I may just push you provocatively: you said AI has the potential, because it's a general-purpose technology, therefore it has the potential to significantly enhance productivity, and yet the energy demand surge is a real one. Would you say, you know, when we say one ChatGPT search consumes 10x more electricity than a simple Google search, is that something we should be worried about at all, or, as you say, the market is ready enough to deliver?

So our analysis suggests that there are two factors in play. One is the energy consumption increase triggered by the new technologies, and the second is how digital technology can accelerate innovation and how it can lead to more efficient energy system operation.

And unfortunately, it usually takes a decade or one and a half decades for new energy technologies to reach a critical mass even after the technology maturation happened. Don't forget that we board solar as the sudden and surprising game changer. Actually, Albert Einstein got his Nobel Prize for the photoelectric effect 100 years ago. The first functioning solar panels were designed in the Bell Labs in 1954, so solar had several decades before it broke through as a revolutionary technology.

So our analysis suggests that up until around 2040, the energy demand impact is stronger. So compared to the pre-AI scenarios, we do see a higher fossil fuel demand, primarily higher natural gas demand, because it is natural gas which plays a unique role in power generation. Natural gas demand grows more rapidly until 2040, but then it declines more rapidly, because after one and a half decades, the innovation and the new technologies reach a critical mass, and 2040 onward, it leads to a more rapid and more radical energy transition.

So in a way, we are seeing two things then, László. One is that, as Joanna said, at least on paper, clean energy makes sense. And yet, what you're indicating is there is also a maturation period, not just clean energy, but all related technologies, which will result in a tipping point in the deployment and adoption.

Now Carl, if I could turn to you: we're talking about energy, but other resources also matter. Land matters. You've been tracking the water cycle and the linkages between water and food and energy for a long time. Are some of those other constraints going to delay the tipping point, the rapid adoption of the technologies that László is referring to?

So I think we're living—how many people feel like we're living in a profound moment? Where I'm a storyteller, right? So to get on the ground, figure out who's doing what, where—this is a drama. And when we look at these issues, and we're looking at energy, we're looking at AI, we are facing, as you said, we have to play the cards right. This is an inflection point. Every choice we make, every search we make here today and in the future is actually defining our future.

Every search we make consumes water, right? Because how do we generate the energy first off to power the data centers? Well, today, it's usually hydro, it's coal still, it's maybe natural gas. A lot of those power plants are cooled by water, right? So how do we actually generate the power? What's the other elephant in the room that has become very visible recently is: How do you cool those data centers? So each search, when we say has a water footprint, that's water that's consumed. Most of the data centers use evaporative cooling. So when you're searching using ChatGPT, you're pulling water sometimes out of the ground, out of the river, and pushing it up into the sky. You're consuming that water.

So as this story unfolds, we have, again, we have a good and bad story to tell. How do we define that story? Now, you mentioned the nexus. That's a word I hope you take out of here. If there's one word: the nexus of water, food, and energy in the changing climate. This is a system, and we have a choice of making this a positive system feedback loop or a negative one. How do we use our water? How do we apply our water budget to grow our food, to transport it, and then also how we're using our water to run our data?

Now, the other side is we have the opportunity using AI to optimize the systems. We have an opportunity of optimizing our water systems. Mexico City, which could face a Day Zero moment when the taps either run dry or you might not have water on Saturday, Sunday, Monday, and Tuesday—might have maybe one day a week—Mexico City loses 40% of its water through leaky pipes. Okay, so that's a huge infrastructure problem. How do we fix that? Well, we can use AI to listen better. Microsoft believes that they can solve—they say 20% of those leaks using AI to pinpoint where the leaks are and use the water more efficiently.

One more last little piece here: so anybody here from California? So of course we see the water impact there. But just looking back at some statistics, California uses about 20% of its electricity to move and treat water. So when we're thinking about electricity generation, we have to think about these kind of feedback loops. And so the positive feedback loop: when we save energy, we save water; we also have more water for agriculture. So we can listen better, we can course-correct faster. But really, this is the era of systems change, and it's our choice to—is it a positive system change or a negative? It's the feedback loop.

This is exactly the segue I needed for Joanna. Because Carl is referring to not system change in the singular, but systems change in the plural. There's a systems change needed in energy, but also how we look at water, how we look at land.

So Joanna, again, you talked about subsidies, for instance—reforming subsidies. You talked about permitting, you talked about regulation. And we could say we have those problems just within energy, and then we have those problems just within the food system, and then just within the water system. How do we break the logjam across these silos before we even start thinking about the positive ways in which we can rebuild the system?

I mean, that's the question of the day, and it's obviously different in every geography where each geography, each country has its own regulatory and permitting process. But I think that the thing to consider is indeed when you look at where are coal plants built right now, a significant number are built in water-stressed areas and are using this. So the transition in those places to renewables creates a real win-win in terms of ability to have water.

Then you think about siting. Many of the huge conflicts related to renewable energy deployment have to do with land siting. Again, we're looking at positive benefits: if you look at already degraded land or already used land from coal or other extractive industries, you can think about a transition which is much more viable than simply looking at one individual renewable energy plant or transition of one fossil fuel asset. So I do think if we look at this in the round, we're able to find more win-win situations.

And I do also think that we are facing a time when we're going to be making choices between water and land for energy, for food, for people, for biodiversity. These are difficult choices that we're making. But if we look at this carefully in the round, we can choose—particularly there's a lot of low-hanging fruit or iconic zones or green zones where we can start. It may not be where we go all the way, but we can start.

And India has many of these, and your work has pointed these out. And I highly recommend that everybody read the articles and research papers that CEEW has put out and that they're about to put out about this intersection of how you can actually—when you look at it, the work is so exciting because it takes it from this high-level, "Yes, we can do it," to an actual, "In this geography, here is where we should be retiring these assets, building these assets, deploying water here and there." And then we can figure out the regulatory and policy asks that also relates then to the financing. Because once those costs are integrated, it actually makes the renewables more attractive, it makes fossil fuels less attractive, and also it gives investors the certainty that they need to be able to make those kinds of longer-term investments. So there really is a positive scenario if we can come together on this.

Again, Joanna, you referred to this, but Thomas, in what you were describing earlier, your technologies, there's one theme that's emerging, which is the potential to have more bespoke technological solutions that are deployed rather than a one-size-fits-all story, right? And you've taken an old technology and redeployed it in a way that allows you to look at manufacturing a different way—the use of micro power plants, etc.

So let's come back to this issue of, say, the rising demand for AI, right, or other digital technologies that requires on-site clean power solutions, whether it's for data centers, anything else. How can your technologies and your innovations help to ensure that the energy system within those closed spaces are more optimally designed and optimally used? And how could that then help to link up with other systems? Say, you could be running data centers more efficiently, but also then does AI help you to manage the water system and the food system in a neighboring geography? Are these ways in which you're thinking about the potential use cases for your technology?

But László, you need to come in here because, well, you said earlier that technology—there's a choice that's before us in how we deploy the technology, right? And in a way, Thomas is saying there's also a choice of how we train the technology, right? And when you bring the two together, how do we leapfrog to that more sustainable systems approach?

So a couple of things. One is that in the past 50 years, the energy efficiency of computing—so the amount of electricity used for a mathematical problem—improved by a factor of 10 to the power of 13. So that's 1 and 13 zeros times improvement in the energy efficiency of computing. And almost certainly, there will be very, very strong efforts to continue this efficiency. Now, what will be the speed of the efficiency improvement of computing, that remains to be seen.

Without this astonishing—I mean, today, if you have an iPhone charger, your iPhone charger has a little chip in it which regulates the power flow through the inverter. Now, that chip in your iPhone charger has more computing power than the onboard computer of the Apollo spacecraft. Okay, so this was the technological improvement in digital technology as it continues. I think that the tech industry has a lot of responsibility, but my impression is that they are aware of this responsibility, and they are working very hard to improve the efficiency there.

Another thing about—you mentioned this bespoke engine. We actually think that some of the most important game changers that are unfolding are actually modular technologies, right? Solar is a modular technology, and the manufacturing scale-up can be incredibly rapid. Pre-COVID, China manufactured around 100 gigawatts of solar panels per year. The latest month—the last month—was 65 gigawatts in one month. So that's like 700 gigawatts, and that was like a five times, five years increase. The same for batteries.

Some of the most interesting innov—like nuclear, some of the most promising nuclear innovation is taking processes where the underlying nuclear physics has been well-understood from Cold War weapons research, but trying to apply modern digital technology and imagine that in a modular fashion. And Shell's own R&D team is working on a modular carbon capture and storage facility which will be mounted in containers, because current approaches are this heavy, bespoke chemical engineering. So these modular technologies can be ramped up.

Now, one big thing to watch is that so far, the overwhelming majority of this ramp-up of modular technologies took place in China. Now, it is possible to rely on large-scale imports of Chinese equipment, but there might be political question marks. It is possible to replicate those manufacturing facilities: you can have India, you can have Mexico, you can have other countries coming in. Unavoidably, that takes time.

And currently, there is a gold rush for AI. No tech company is prepared to wait, and that will pull in significant quantities of natural gas, absolutely. And that will change the trajectory in the one- to two-decade time horizon.

Carl, you earlier said this is a drama, but you didn't actually expand on what's the drama. But I'm certainly seeing a drama here, because on one hand, what you and Joanna are referring to, the here and now of getting the energy governance right, but also the other natural ecosystems, right? And what Thomas and László are also saying, that there are other technological disruptions that are also underway that's not going to slow down because we've not fixed these other legacy problems. So what's the governance fix for this?

Ah, so you teed me up perfectly on that, too. So I'm going to say one is: How do we define technology? I'm going to propose we need a—does operating system count as technology, right? So I'm going to propose that we truly need to create some feedback loops that provide the course correction. Because as Joanna said, one of the biggest impediments here is policy, regulation, incentives. So how do we fix policy? Because it's out of sync with the times, right?

How do we see ourselves? How do we set the policy up for the future? But then right now, we're not internalizing, we're not changing fast enough. We have policies that are entirely out of sync with the financial needs that we need to address our big infrastructure issues. But then we also have the ability to listen to ourselves better and course-correct. So think of technology as a radar system, not as a one-way street, but as a ping, as a call and response, so we can monitor our situational awareness.

But then we also have to use the technology to listen better to ourselves and our leaders and tell our story better so they see themselves in the picture. Then we need to use technology to convene and co-convene with humility so that, again, we can look to the future and that we can actually course-correct in real time. Because this isn't a story about today or next year; it's a story about bridging between the two. And right now, they tend to be siloed and they tend to be responsive—we're dealing in the crisis moment—and they tend to be forward-looking, looking in the future.

Before we turn to the audience, I wonder if instead of training technology, we can train governance. You know, unlike the Boston Dynamics dog that was untrained on AI and you spent decades learning and relearning and unlearning, can we do the same for governance? Can we do siting and inclusion and social stability, etc., with this new operating system?

There are actually AI tools which read the documents of public hearings that help the permitting bureaucracy, right? Yes, exactly, better way of public consultation.