The AI-Energy Nexus

[MUSIC] There is a lot of concern out there about, you know, what is the impact that the demand of data centers? And what the impact of those data centers are on the electricity grid? Is that going to drive out electricity prices? Great reliability problems. And the problems that our team is trying to solve are exactly the ones that try to prevent those things from happening and make sure that we actually have a positive impact on the grid. We get a chance to use our role as outsized buyers of clean electricity to advance solutions in that space. [MUSIC] This is where the internet lives. A show about the unseen world of data centers and the incredible advances they make possible. I'm Stephanie Wong and I'm your guide to the people and places that make up the internet. This season, we're exploring how AI is fueling the next industrial revolution, redefining everything from farming and healthcare to design and manufacturing. And we're asking how data centers are adapting to power at all. In this episode, pushing the frontiers of energy. [MUSIC] Once upon a time, wind, solar, and batteries were considered to be alternative energy. They were experimental and very niche. Wind has great potential as an energy source in America. A potential that is in the process of being realized. In the 1980s and 90s, wind and solar started gaining attention, but utilities still view them as expensive science experiments with small commercial uses. The Department of Energy is working with NASA to develop large-scale wind machines. Four large experimental turbines are generating power for electric utility companies. It took a combination of private dollars and government investment to de-risk these technologies and drive down costs. And today, they're not alternative anymore. Wind, the solar, and batteries now make up 85% of all new power capacity added to the grid. And they're often the cheapest resources on the grid. As manufacturing costs decrease with mass production. As new technology and stronger and lighter materials are developed, wind energy will find an even more important place under the sun. Remember, at one point, wind and solar were also new technologies coming down the cost curve. But now they're pretty much scale and mature. And just last year, our broader energy team did eight gigawatts of renewables. That's Lucia Tian. She leads advanced energy technologies at Google. To put that eight gigawatt number in perspective, Lucia is saying that Google deployed three times more wind and solar capacity last year than the entire market installed back in 2005. It's a staggering amount, especially for a single company. And it's still just a start. There are technologies that we know we're going to need in three, five, ten years that aren't quite ready today. Wind and solar are doing the heavy lifting today, but Lucia is looking for the next set of breakthroughs. She's targeting a suite of technologies that push the limits of physics and chemistry, harnessing the power of the atom, drilling for the heat inside the Earth's crust, and finding new ways to trap carbon emissions. These innovations are just reaching the commercial stage. And Lucia is focused on getting them to scale. Our team's role is to help Google advance and scale up the energy technologies that we can't yet buy at scale on the market today, but that we know we're going to need to meet both our electricity needs in the future and our clean energy goals. [MUSIC] Lucia is an engineer with expertise in economics and data science. She loves figuring out complex systems. She's the kind of person who always picks the windows, and not to look at the clouds, but to look down, taking pictures of big infrastructure projects. Well, I can send you the pictures. I've always been really fascinated by big machines and systems. I love planes and trains and ships and the electricity grid is probably one of the largest and most complex engineering systems out there. Some kids like to take toys apart to see how they work. Lucia was the kind of kid who wanted to understand the design theory behind the toy. I was maybe more of a theoretical kid. I love to think about complex math problems and solve little quantitative puzzles with my dad. One of the proudest moments of my life was the first time I solved a math puzzle faster than him. And I think that naturally translated into an interest in science and engineering as I got older. She took that system thinking to the highest levels of government and industry. Over her career, Lucia has modeled supply chains for fighter jets and analyzed massive data sets to understand voting behavior. She uses data to make sense of a chaotic world. And now she's applying that logic to the energy transition. The amount of data out there and almost every realm these days is absolutely overwhelming. And the amount of input we're faced with on a day-to-day basis is overwhelming. For me, it's all about how to synthesize and abstract in a way that leads to good decision-making across realms. And that's intelligence. Also, what artificial intelligence is about, right? And when did you first get interested in energy specifically? That's a good question. A North Star, for me, is to always run towards what I saw as the most important and complex problem of the day. And that's certainly the inflection point we're in at the energy system today. And so I think for the last, let's say, five, ten years, I've been very interested in topics related to energy systems and climate change and have somehow managed to work my way into this industry. In 2022, Lucia brought that system thinking to the Department of Energy. Her mission to figure out how to take advanced energy technologies out of the lab and scale them up for the real world. Yeah, it was a perfect crash course in the full range of advanced energy technologies. You know, their role in the energy system and what each needs to overcome to achieve commercial scale. Wind, solar, and lithium ion batteries are mainstream technologies today. But the grid still needs lots of complementary firm resources, things like advanced nuclear, hydrogen, and geothermal. Her team at the Department of Energy created a series of instruction manuals, exploring how to build commercial pathways for each of these technologies. And through that process, we got to talk to and learn from everyone that was in the ecosystem in each of those verticals from. The tech OEMs, to the developers, to the financeeers, to utilities, and think about how to scale up those technologies from a public investment perspective. Now I get to solve that same problem, but from the Google side, from the point of view of a large corporate buyer of electricity. Back in 2020, Google set a goal to run on 24/7 carbon-free electricity by 2030. That means serving offices and data centers with clean electrons every single hour of every single day, which is much, much harder than just buying offsets. We see a Spencer days trying to tackle that ambitious goal. What's exciting about that goal is that it requires us to not hit the easy button, right? It requires us to build out a portfolio of technologies. And that includes things like nuclear, geothermal, and power generation with CCS or carbon capture and sequestration. It also includes technologies that help improve the grid itself, advanced transmission, technologies, and grid enhancing technologies. In the last five years, the scale of that mission has grown. With the AI boom fueling a new wave of construction, data centers are consuming more energy than before. For Lucia, this isn't just a procurement challenge. It's a design challenge. When we're bringing large data center loads to the grid, we have to think about how we can be good grid citizens and make sure that that large load doesn't impact the rest of the ratepayers or consumers on that grid. So we always want to make sure that we are actually having a positive impact on the grid from a reliability, from a affordability, and from a clean perspective. Lucia sees a silver lining in this energy demand growth. She views it as a lever. By acting as a partner and buyer for these new technologies, Google can inject the capital and confidence needed to jumpstart them. When we scale a technology like a geothermal or a nuclear, we're doing that so that Google has that option in the future to be able to procure energy from those sources at less expensive rates. but ultimately we're bringing those technologies down a cost curve for everyone. And why do these technologies need a blueprint for scaling? Like why can't you just simply buy electricity from them today? Scaling new technologies is really hard. You have to overcome a bunch of different hurdles, right? Well first you have to manage the technology risk piece of it. Depending on where the technology is in its life cycle, there may be more R&D involved. There may be fundamental science hurdles that you have to overcome. And at least on the buyer side or as a public investor in scaling these technologies, you need to think about, okay, how do you actually cross all those milestones and create the right market structures and incentives to do that? And then you have to actually, you know, even as the technology gets more mature, you have to actually build projects. And these are big energy infrastructure projects, right? So it's actually a pretty challenging transition for some of these early technology companies to then transition to developing real projects because the problems that you have to solve in developing those large projects is very different from developing a technology. You know, things like supply chain and APC and sighting and permitting, that's a different skill sets, a different set of talent you need to develop projects. And all of those challenges can get in the way of, you know, on time, on budget, project delivery that you need to see. For the last 20 years, electricity demand in America was basically a flat line, but that era is officially over. We're now in what experts are calling an electricity super cycle. From the electrification of cars to the return of manufacturing and the growth in AI, everything is plugging in all at once. The grid is waking up and it is hungry for power. I think we are really at a very interesting inflection point where, you know, the demand for AI services is intersecting with a critical moment in the energy sector. And so what's really interesting is that companies like Google and our peers can have an outsized role in the energy sector. I think we can play a catalytic role by making investments and by bringing innovation to the table in catalyzing the development of new energy technologies. So let's look at the toolbox, Lucia is working with. Each technology presents its own unique hurdles. First up, nuclear. The US used to lead the world in this industry, but our capacity slowly atrophied. Because demand was flat, we stopped building new plants. There hasn't been that need, right, to bring large-scale new generation resources online. So, you know, just a few years ago, we were still in an environment where nuclear reactors were being shut off. And that points at a challenge of we've lost a lot of the experience and supply chain and the ability to build these large-scale energy infrastructure projects, especially in nuclear. To help reverse that trend, Google is taking an all-of-the-above approach. That means revitalizing nuclear plants that already exist while simultaneously working with startups that are piloting brand-new reactor designs. In the short run, we need to maximize our existing nuclear fleet, right? And that means restarts like the Duane Arnold nuclear plant restart we just announced. It also includes things like upgrades. That means scaling up the power output of existing plants. Those are capacity expansions that can come more quickly on the grid. When it comes to new nuclear, that's going to take a little bit longer to build out real large new capacity on the grid from nuclear. And in my mind, there's a role for both large reactors and small modular reactors. Then there's geothermal, tapping the hot water or hot rocks beneath the Earth's surface for power generation. This is another area where America dominated back in the 1980s. But when the commercialization of fracking took off, investment dollars were pulled away from geothermal in favor of gas extraction. Later, wind and solar overtook the renewable market, leaving geothermal on the sidelines. The challenge with geothermal has always been two things. One is geography, right? The resource is limited to the places where you can get to pretty high temperatures at relatively shallow depths. And sometimes you drill and you don't get to that heat. And then, you know, financial investors will lose out on that initial investment. That's the second challenge. Resource exploration risk. So that initial investment in resource exploration can be pretty risky. The solution was to stop looking for the perfect geology and start engineering it. Google partnered with Furvo Energy, a company with a drilling technology that can create geothermal reservoirs almost anywhere. They're now scaling up to a commercial-sized project that will provide 115 megawatts to serve a Google data center in Nevada. So these are fracking techniques. And those techniques, called enhanced geothermal, allow us to reduce the resource exploration risk and also expand the geography in which we can access geothermal power. So it's a really exciting moment in the geothermal industry, right? And I think geothermal is ready to be scaled right now. Next up, batteries. These are critical for a wide variety of applications, including in cars, homes, and paired with renewable power plants. Costs have come way down and lithium-ion batteries can now scale to provide 8 to 10 hours of storage. We're even seeing batteries integrated directly into data centers to provide more flexibility. But the sun doesn't always shine and the wind doesn't always blow. And those periods of lower renewable generation are often longer than the 8 to 10 hours. Right? And so we need for grid reliability reasons for renewable shifting reasons, battery storage solutions that last 8 to 24 hours. That's what we call kind of medium-duration storage. And for the really longer periods up to 100 hours. This area is brimming with innovation. But that actually creates a new problem. Google has to weigh through a sea of options to figure out which chemistries and technologies are actually promising enough to bet on. There's different battery chemistries, there's thermal solutions, there's mechanical solutions, right? There's just so many different types of storage and hundreds of different companies. I feel like every day there are new energy storage startups knocking on our door. And sifting through that and being able to figure out what really is going to scale and what's going to be economic in the future is a real challenge. On the market structure side, it's also a really big challenge because the electricity markets today don't fully value long-duration energy storage for all of the different benefits it brings to the grid. Even with all this new green tech, the grid is still heavily dependent on burning gas. Since we can't delete those plants overnight, Google is looking for ways to clean them up. Which brings us to the last tool in Lucia's current toolkit, capturing carbon right at the power plant. Power generation with carbon capture and sequestration has to be part of the decarbonization set of solutions. So we've long believed this, but it took us a little while to figure out where and how we can do our first project. CCS is a technology that's been around for decades. But it hasn't gotten the mainstream traction many expected. And that's because it's incredibly difficult to execute. Carbon capture and sequestration requires a very complex value chain. You have to get the gas, you have to ensure that it is low carbon intensity gas, you then have the power plant, then you have the capture and then you have to transport and sequester the CO2. So there are many different steps in that whole chain. From an economic perspective, it's also a technology that ultimately is always a cost. So someone has to be willing to pay the premium to capture the CO2 from that gas plant. Because running the gas plant without CO2 is always going to be cheaper. That led to a project at the Broadwing Energy Center in Illinois. It's designed to prove that CCS can work at scale using advanced turbines to scrub 90% of the carbon out of the exhaust. But the goal isn't just to build one nice plant. Google is working with its partners to turn this specific design into a repeatable blueprint that they can deploy over and over again. This is a really exciting first project. It sits over the first first operational, the only operational class six wells in the country for permanent CO2 sequestration. Our partner there ADM has been doing CCS on site for over a decade now. So there's lots of operational experience and lots of community buy-in for it. We also love the project because it starts to set what we hope can be a combination we can repeat elsewhere. We're also going beyond that first site so there's a pipeline of projects to come for which this can become a sort of reference. So when you look at these technologies, are there commonalities in how you're solving the challenge of scaling first-of-a-kind energy projects? There's so many unexpected challenges that we run into on the path from first engaging with a technology provider and ultimately getting to an executed contract with them. We have a team of nerds that love to learn about new technologies and love to dive into the research papers and talk to technical experts and go on site visits. And then we also have a very creative commercial team because in many cases for these first-of-a-kind deployments, capital is a really big challenge. And these are all hard texts, right? They're not software startups where investment sizes are on the order of millions. Some many cases, you know, first-of-a-kind planned could be in the hundreds of millions or billions. And so there's a lot of capital that's needed and we can also bring some other tools to the table. For example, Google can provide some equity, project development capital, tax equity. These are all financial tools we can bring to the table to give the market further confidence and attract or crowd in other sources of financing. Are you optimistic? I'm always optimistic. I really believe in the power of technology and innovation to help us solve some of these challenges and that's kind of exactly what my team works on. You've described Google's energy strategy as energy for AI and AI for energy. What do you mean by that? We are just beginning to see what AI ML tools can do for us. And we need energy to serve the AI load. And therefore, we have an opportunity as a buyer of electricity to impact the electricity grid. But on the flip side, the AI tools themselves can help us improve the operations and resources that we actually connect onto the grid. One of the most exciting uses of AI tools will be to accelerate the way that we perform grid studies and able to connect both loads and generation more quickly. But the example that I'm thinking of is perhaps a little more unusual. Earlier this year, we began partnering with Westinghouse, the company behind the large AP 1000 nuclear reactors. And we're using our cloud AI tools with Westinghouse's incredible wealth of data on nuclear operations and construction. And we're using AI to generate and optimize construction work packages. So when the last large nuclear reactors were built in this country, literally pieces of paper had to change hands every time a cable had to be pulled at the construction site. We can change that and we have to change that. So there are many paths from here. And I think we're trying to set ourselves up to have a bunch of really good options. There is a pervasive feeling right now that America has lost its muscle memory for building big things. We're great at inventing things in the lab, but we struggle to build them in the wild. Experts call it the missing middle. The gap work great innovations fizzle out before they ever reach commercial scale. Lucía's team exists to bridge that gap and approve that we can still build ambitious infrastructure. I think there's incredible innovation happening in the US and many of the advanced energy technologies that we've been talking about have their origins in US national labs and universities and there's just an incredible R&D ecosystem here. What many people have talked about as the missing middle is that we have a hard time commercializing and scaling those technologies right here and we have a hard time building big infrastructure projects. And so I think that's definitely a piece of the challenge we have to solve. We can bring commercial tools and financial tools and policy tools to the table. And here's again where I think AI can have a transformative role to play. I think we've got to use all of the AI digital tools that we can bring to the table to actually build these projects in a more efficient and cost effective way. What makes you feel hopeful right now? I think what gives me the most hope is the fact that we are sitting at an intersection of two industries, right? The tech industry and the energy industry where the future of those two industries have become intertwined. And I think each can have an accelerating and positive impact on the other. Lucia Tian is the head of advanced energy technology at Google. Where the internet lives is produced by latitude studios and collaboration with Google. You can subscribe to the show anywhere you access your podcasts. Please give us a rating if you are enjoying our journey together. If you want to see some of these stories for yourself, we have a series of short documentaries on YouTube. They're linked in the show notes. And if you want to learn more about how Google's data centers are benefiting communities around the world, go to datacenters.google. I'm Stephanie Wong. Thanks for listening.