When Brian Janous took charge of Microsoft’s clean energy strategy in 2011, the company’s data center demand was modest. He was measuring new demand in the tens of megawatts.
Over the years, that grew to hundreds of megawatts of new demand as hyperscale computing expanded. And then everything changed in the spring of 2023, with the public launch of ChatGPT 3.5, which ran on Microsoft’s data centers.
“That was the moment that I realized we were going to need a bigger boat. This is a massive leap in a period of like six months — and the amount of time that it takes to actually build infrastructure was measured in years,” said Janous.
Projections show data center energy demand could double in the next couple of years, driven by artificial intelligence. Janous saw the hockey stick growth coming, and he realized the disconnect between how fast AI is moving and how the core input to data centers — electricity supply — is struggling to keep pace.
After decades of flat demand, load forecasts are doubling because of data centers, expanding ports, new manufacturing plants supported by the IRA, and electric cars.
Janous recently co-founded a company, Cloverleaf Infrastructure, to help utilities unlock grid capacity with grid-enhancing technologies, batteries, and other flexible resources to meet the onslaught of new demand. He also advises LineVision, a dynamic line rating company that is helping expand transmission capacity.
This week: we talk with Janous about why we don’t need energy miracles — we need to think creatively about planning, and squeezing more out of the system.
This episode is brought to you by The Big Switch. In a new 5-episode season, we’re digging into the ways batteries are made and asking: what gets mined, traded, and consumed on the road to decarbonization? Listen on Apple podcasts, Spotify, or wherever you get your shows.
Transcript
Stephen Lacey: So last week our team was at DISTRIBUTECH, this very large conference on transmission and distribution infrastructure. It’s been around for more than 30 years. It had been a while since I attended, I think eight years, and I was struck by a bunch of things that felt different and some things that haven’t changed.
What hasn’t changed? Well, if you go out on the floor and start talking to companies, you hear the same thing: Selling to utilities is really hard. Getting good data is even harder, and even with many of the big industry proclamations about net-zero targets, most power companies are still very incremental in their adoption of things like virtual power plants as a replacement for gas plants, which we covered on the show a couple of weeks ago.
But a few major things have changed. The conference has evolved from a bunch of equipment companies to a lot more services companies for everything from managed EV charging to advanced distribution management systems. AI is definitely everywhere. Sure, some of it might’ve been an optimistic play, but it’s becoming foundational to the services that many companies are building.
And then there’s this mega-trend that is unfolding really quickly, which we’ve also covered on the show. Suddenly, after decades of flat demand, load forecasts are doubling thanks to data centers, expanding ports, new manufacturing plants supported by the IRA and electric cars. And so I ended up logging more than 18,000 steps a day at the conference talking about some variation of this trend, including a few thousand steps with this guy.
Brian Janous: So I’m Brian Janous. I most recently was the VP of energy at Microsoft. That’s a role that I had for over 12 years when I built the energy team there. And now I’m also founding a company called Cloverleaf Infrastructure.
Stephen Lacey: Brian was responsible for procuring renewables for Microsoft’s operations. He stepped into the role in 2011 when Microsoft’s data center load was pretty modest. This was at the front end of the boom in hyperscale computing.
Brian Janous: We weren’t a huge consumer of energy at the time. I think when I joined Microsoft in 2011, there was somewhere between 100 and 200 hundred megawatts of data center capacity, which is not nothing, but compared to what it is today is pretty minuscule.
Stephen Lacey: So there was extra space on the system for companies building data centers to grow. Brian had to think about serving tens of megawatts of new load with renewables and then later hundreds of megawatts. And although he faced some grid constraints and time pressures, it felt manageable. And then in the fall of 2022, everything changed fast.
You had this front row seat to this booming demand. Microsoft of course partnered with OpenAI and then with the launch of ChatGPT, we see a hundred million people sign onto the platform very quickly. And then suddenly you have billions of visits to ChatGPT and it throws data center demand through the roof. What happened?
Brian Janous: It definitely caught people by surprise, not that there was AI or that we were working with OpenAI, but that adoption, like how quickly that happened. I distinctly remember ChatGPT-3 being released in November of 2022, and that it was just a few months later in April of ’23, that 3.5 came out and it was a huge leap in terms of its capabilities.
And that was the moment that I realized we were going to need a bigger boat. This is a massive leap in what this thing can do in a period of six months, and the amount of time that it takes to actually build infrastructure to support that was measured in years.
Stephen Lacey: Suddenly projections show data center energy demand could double in the next couple of years. Brian saw that hockey stick growth coming and he realized there was a disconnect between how fast AI is moving and how the core input to data centers, electricity supply, was struggling to keep pace.
Brian Janous: So it takes two things to build an AI model. I need a chip, I need to make sure that I have my GPU, and I’ve got to have an electron to power it. And one of those things was going to constrain it.
Stephen Lacey: And it’s starting to look like the power side might be the bigger problem.
Brian Janous: We’ll find a way to make more chips. There’s a lot of capital in that space. I’m more concerned about can the electrons keep up with the growth?
Stephen Lacey: There’s reasonable debate inside the tech industry over the demand picture for data centers. Today, we can build them much more efficiently to reduce power and water use. And while internet use has exploded, data center demand has stayed relatively flat. There are also ways to train models to make them less computationally intensive. But it’s clear that data centers are much more power hungry than they were just a couple of years ago, and our power infrastructure is going to have a hard time satiating it.
Brian Janous: The tranche sizes each year that you’re having to go out and procure are now being measured in the gigawatt scale. So you’re no longer immaterial to the utility’s load forecast. In fact, in many cases you are the utility’s load forecast.
Stephen Lacey: This is The Carbon Copy. I’m Stephen Lacey, this week thinking creatively about how to prepare the grid for an onslaught of new demand. I sat down with Brian Janous at DISTRIBUTECH to talk about why we don’t need energy miracles, we just need to think creatively about planning and squeezing more out of the system.
So there’s no debate that demand is far beyond what anyone foresaw just a couple of years ago, but there’s debate about how you handle that demand and ultimately how big it will be. The International Energy Agency recently came out with a study showing that data center energy use will double in the next two years. Does that feel right to you?
Brian Janous: I think that’s going to be hard to achieve. It’s really about whether we can actually physically interconnect that much capacity. Again, because even if you just take all the big tech companies and ignore the rest of the data centers, that alone is probably at least 15 to 20 gigawatts of capacity. And so talking about doubling that, that’s building a lot of stuff. And so, that’s where I’m skeptical that we’re actually going to achieve that growth rate without some significant movement in terms of how fast we’re investing in grid infrastructure.
Stephen Lacey: Not just energy, but we need clean energy. We need a lot of new renewable energy in other forms of low carbon energy and we’re running in place.
Brian Janous: Right. And of course, it’s not just data centers that are looking for this, right? We’ve had this huge uptick in domestic manufacturing in the United States. We have big focus on hydrogen. All of these things have the same raw material, which is electrons. So they’re all competing and they’re all looking for clean electrons as well.
I saw that Brian Kemp, the governor of Georgia, was at the World Economic Forum last month, and he was talking about Georgia being the EV capital of the world in terms of manufacturing production. But the thing that I thought was so interesting that he said is that he recognized that in order to do that, they needed a lot more clean electrons. He realized that that’s what those customers were going to be looking for, and he realizes that’s the raw material that they have to offer.
And so when you have a Republican governor saying, “Clean energy is the thing, this is what we have to be about,” it’s pretty encouraging that there’s that recognition. But it’s a huge challenge as you know to build out that much infrastructure.
Stephen Lacey: So there are two extreme ends of the spectrum when talking about how to solve this problem. On one end of the spectrum are the techno-optimists who say, “We just need to build a lot of nuclear power, maybe nuclear fusion, and that will solve the problem.”
Then on the other end of the spectrum, you have a lot of data center stalwarts who say, “Look at all the amazing innovations we’ve had in the efficiency of computing and the build out of these data centers. Data center energy demand wasn’t nearly what we thought it would be from the ’90s into the 2000s just because we got better at building data centers.”
Where are you in that spectrum?
Brian Janous: I guess I’m probably a little bit in the middle. I definitely think that we over-index on counting on, quote, “energy miracles,” because again, going back to what we were talking before, it takes a long time to build this stuff. So we are not going to have significant contribution of say, nuclear energy, certainly not fusion, that’s going to meet the challenge that we’re talking about right now, which is we got to build a lot of stuff this year and next year and the year after that.
So we have a lot that has to be done between now and then. I definitely am a huge proponent of that and we should keep investing and focusing on that because we will need that the coming decades, but I think we just have to be realistic about what we can achieve in the near term.
I’m also skeptical of the idea that efficiency is going to save us. In fact, I’ve heard that a ton over the years and my response to that has always been, the only time that efficiency is going to reduce the input needed for any process is when you’re starting to exhaust the demand for that resource. And when you think about something like data, we’re nowhere close to satiating human demand for data, which means that if you make it more efficient, you just accelerate up that S-curve faster and you actually use more of it, not less of it.
And AI is a great example. What happened with AI is suddenly we could do a lot more with data in terms of image generation and text generation, and so we didn’t use less computation, we’re using more energy as a result. We’re doing things that we didn’t think we’d be able to do for 10 years and we’re doing them today. So it just pulls innovation in and increases demand. It doesn’t go the other way. So efficiency, at least in this world right now, means we use more, not less energy.
Stephen Lacey: Then let’s narrow in on the infrastructure challenge. If these data center portfolios are not adding tens of megawatts, but now gigawatts of demand, if manufacturing facilities are doing the same, and we’re seeing roughly the same type of demand in manufacturing, and we have all these EVs hitting the grid, we need to build quickly to service all this new demand. And what’s that infrastructure challenge? How much do we need to expand the grid itself to accommodate integrating all those new resources?
Brian Janous: Well, we are going to have to build a lot more infrastructure. We’re going to have to build a lot more transmission lines, a lot more power plants, but we also have to be realistic about how hard that stuff is.
I’ve just started listening to a fantastic podcast called The Big Dig. A lot of people are familiar with the Boston tunnel project that took way too long and was way over budget. But what I find so fascinating about it is so much of it has been about just the on-the-ground politics and the community engagement and all the stuff that makes building stuff really hard. It’s like, you look at these things and like, “Well, why did this take so long?” It’s like, “Well, there’s a million reasons why it took so long.” And that’s the stuff that I think people vastly underestimate.
And so yes, we need to be building new things, but we also have to be realistic about the fact that we’re going to have to find growth inside the existing system too. We have to find ways to better optimize the system we have and get more out of the system we have rather than just counting on, well, we’re just going to build a bunch more power plants because we will, eventually. But we’ve got a lot we need to do between now and then in terms of meeting load growth, interconnecting customer requests, etc.
Stephen Lacey: And that’s what you have focused on since you left Microsoft. You took on an advisory role at LineVision. Talk about your thinking around grid-enhancing technologies, what LineVision does specifically, and why you started getting interested in grid-enhancing technologies to expand the existing grid capacity that we have now?
Brian Janous: So LineVision is a dynamic line rating company. So this is a sensor based technology that allows us to better understand the actual load flows of power on the system. I mean, think about the same way you use real-time traffic insights as you’re driving around, it’s helpful to know, oh, this highway is congested, I’m going to go this way. It’s that sort of system for the grid so we have a better understanding of what can actually flow over a line at any particular time.
And that allows us to unlock more capacity because historically, utilities have used a static line rating, which is, well, this line is rated for X, but it doesn’t take into account ambient temperatures and other things that are going on in the grid. And the reality is there is more capacity than what that static rating would say. And so dynamic line rating creates value out of the current system.
And why I find this so compelling is, again, in my old role at Microsoft, we were constantly running into these constraints where utilities would say, “Well, no, we don’t have power.” And I’m looking at the system going, “I think you do. I think there is power there. I think there is capacity on the system, but your way of planning the system is quite antiquated.” And if we had better technology and better insight, we could find ways to optimize more in real time what capacity can actually be delivered where.
Stephen Lacey: And if we look at grid-enhancing technologies, like the portfolio of grid-enhancing technologies, what are they generally? What are the other technologies that could help us expand existing capacity?
Brian Janous: There’s a few different things. I mean, dynamic line rating is one. You also have different forms of transmission optimization. Think more like ways, for instance, where you can actually control traffic. One thing is even knowing that there’s congestion in the system, but the other is can we start to redirect power flows over different parts of the system? So there’s different ways that we can use these technologies and use them in a combined way as well because they all sort of do different things. But as you put more intelligence on that system, you can incrementally start to eke out more capacity where otherwise you would think there’s not any.
Stephen Lacey: And do you have any numbers behind what we could potentially unlock?
Brian Janous: Well, yeah. In fact, there was a great RMI study that just came out very recently that looked at the PJM system and determined that you could unlock in excess of six gigawatts of extra capacity. I mean, PJM’s a 90 gigawatt system, so six gigawatts is not nothing. That’s pretty substantial. And this is for an investment of about a hundred million dollars that would return annually in excess of a billion dollars to ratepayers.
So I mean, that’s a huge return on investment without having to build a single new transmission line. And so when you think about the cost of doing that compared to the cost of simply putting better intelligence into the existing system, it’s a massive return on investment.
Stephen Lacey: So since Microsoft, you’ve also been focusing on building your own company, and just yesterday before we recorded, you announced that you have co-founded a company called Cloverleaf Infrastructure. You’re taking a lot of your understanding of site development and procurement from your days at Microsoft and combining this with your understanding of all these different technologies, both on the generation side, on the grid-enhancing technology side, and putting it together to help utilities figure out how to manage this load growth. What are you guys doing?
Brian Janous: Yeah, so really excited about the announcement. This is something that’s been in the works for a little while for us, and it goes back to what you said is, my passion is around how do we grow the grid? How do we meet these customer demands? And I saw firsthand all the challenges of doing that, especially with the burden of having to deliver so quickly.
And being able to step out of that for a little while and start to think about how can we really be more intentional in working with utilities from a customer perspective to say, “Let’s create a better runway for customers that come to you and are going to start asking you for increasingly large interconnects.” Can we actually make that process easier where utilities can start to better understand the investments they can make on their system, how those investments will be received by those future customers, and really just provide them a signal to what’s coming?
Because we’ve seen a bit of a disconnect there too in that there’s not always a great signal from these companies as to what they need. Oftentimes, they don’t even know. Going back to the conversation we were having about ChatGPT, no one was fully prepared for the scale and the speed that that was happening.
And again, if technology is moving that quickly and infrastructure is lagging it by years, we’re going to have a huge deficit of powered land in this country to serve, again, not just data centers, but also all the domestic manufacturing that’s looking for the same thing, and hydrogen, and other things that are also going to be looking for clean energy to power their businesses.
Stephen Lacey: So then how are you actually mapping out projects?
Brian Janous: The opportunity set in front of us is sit down with the utility, understand where are the points on your system where you think there might be opportunities to deliver a significant amount of power capacity. It’s really the inverse of what you do when you’re doing renewable development. You’re looking for what are spots on the grid I can actually connect and inject power in, a significant amount of power? We’re doing the reverse to say, “Okay, what are the points on the grid where you can pull a significant amount of power out?”
And then to dive in further to say, “Okay, well, what are the constraints that we’re seeing?” Like you say, “Okay, we think there’s 300 megawatts here.” Okay, well, what would it take for that to be 500 megawatts or a gigawatt? What sort of investments would you have to undertake thinking about things like grid-enhancing technologies, thinking about how we use storage as a flexible resource?
Because the thing that I think people that work in this industry understand, but most people don’t understand is when a utility says that they don’t have power, they don’t mean that most of the time there’s not power. They just mean there might be a handful of hours a year where there’s not power, so they can’t commit to delivering 24/7, 365 a year. And so the questions then become, okay, how can we take the time to think about what investments would be necessary to start to expand the amount of capacity you could actually offer a customer?
Stephen Lacey: How complicated is it?
Brian Janous: It’s pretty complicated, but it’s not rocket science either. And this is one of the things that when we think about going back to the question about do we need these miracles? It’s like, well, yeah, it’d be great to have fusion, but there’s a lot of really boring and unsexy stuff like transmission censoring and dynamic line rating, those things that people don’t get all that excited about it. But that is the nuts and bolts of what it takes to get stuff done in this industry.
The other piece of course being the regulatory piece of it is, how are the tariffs structured? Are they structured in such a way that incentivize investment the way we need it done? That stuff is kind of boring, but we don’t really get anything done unless you spend time digging through a utility tariff and really understanding how the thing works and how we can apply new technologies and new ways of operating into that system.
Stephen Lacey: As someone who founded a company in this space, I know you don’t consider it boring. And as someone who covers this space, I don’t consider it boring, but how do we make this a bigger piece of the conversation? Because I feel like a lot of this gets missed when we talk about specifically how to serve new data center demand driven by AI. This doesn’t even feel like a legitimate part of the conversation. Maybe it is on the utility level and we’re just not hearing it, but it’s not really a part of the discourse.
Brian Janous: It’s not fully yet, and I think what it’s going to take, going back to the comment I made about Governor Kemp, it’s going to take more states and more communities realizing that the biggest opportunity for economic development, job growth, tax revenue is power. If you have power, you can attract jobs, you can attract investment, you can attract that tax revenue, and clean power. That’s it.
And so, if I’m the governor of a state and my utility is turning that investment away because the utility has not invested, or if it’s PUC is keeping the utility from making the sort of investments necessary ahead of time, then if I’m the governor, I’m picking up the phone and I’m calling the head of the PUC saying, “Why am I losing to my neighboring state, this neighboring community?”
So I think it really comes down to this recognition that the job opportunities that are going to be created over the next decade are massive, and they’re all tied to, is there going to be power available and is it going to be clean?
Stephen Lacey: What are your thoughts on using AI to actually make the grid more efficient? So LineVision is a great example of a company that uses artificial intelligence for dynamic line rating, and there are dozens and dozens and dozens of vendors out there that are using machine learning and other forms of AI to orchestrate virtual power plants, for weather forecasting, et cetera. And so this can be an extremely powerful unlock for clean resources to serve this demand. How do you think about the positive impact of AI to actually solve the power problem?
Brian Janous: Oh, I think it’s huge, and I think the return on investment we get, because I’ve gotten this question a lot of, wow, data centers gobble up so much of the world’s electricity when in fact data centers use 1 to 2% of the world’s electricity. And when you think about what we can do as a result of that and then even add on AI … And crypto notwithstanding. I won’t get into whether that’s a good return on invested electrons.
But when you look at what happens in the cloud and AI, there’s a massive return on invested energy for what we can do in that space. And so I’m very bullish on the role that AI will play in helping us to optimize the grid because again, going back to what we were saying before is there’s so much that can be done in real time and so many flexible resources that exist on the grid that are completely untapped.
But that requires a tremendous amount of intelligence and orchestration that we’ve been talking about for years with smart meters and all these other things, but it’s really difficult to implement in practice. And so we do need better data layers to help understand what are truly the potential unlocks that we have in the existing grid.
Stephen Lacey: This is something that utilities haven’t thought a lot about, or they haven’t fully invested in grid-enhancing technologies and they don’t know fully what’s happening on the grid and it’s constraining renewables. And suddenly, if they’re taking this really seriously and they’re having to think about how to use the grid more efficiently, how to expand it, and then how to serve that new load with additional clean resources, that there is a pretty good opportunity here if we do it right?
Brian Janous: Yeah, I’m actually quite optimistic. And part of the challenge is we are coming out of this period of stagnation, and so we weren’t really forced to have to think about these issues that much over the last decade. We didn’t have a huge problem with interconnecting renewables. We didn’t have a problem getting the power we needed for data centers to grow over the last decade.
But as we’ve come into this decade, all these problems have started to crop up, and we’re dealing with the fact that load is looking like it’s going to continue to accelerate, and you have these utilities restating their IRPs and suddenly realizing, oh wow, we have to double the size of our system in the next 15 years. And so I think there’s an urgency that simply hasn’t been there for the last 15 to 20 years. And so it is pushing utilities to have to think more creatively about how they’re going to invest in their system and how they’re going to meet this customer load.
Stephen Lacey: So we have two limiting factors here for AI, and that is chips and energy. So we’ll ask Sam Altman to carve off maybe a trillion from that $7 trillion he’s looking to raise for chips and put it into grid technologies.
Brian Janous: Yes, we need both. And of course, Sam is working on a fusion company as well, and so he is trying to solve that problem long term. We’d love to see some of that go into grid-enhancing technologies in the short run though.
Stephen Lacey: And then when we talk about the short run, is this a problem that can be solved in a year? What time scales are we talking about?
Brian Janous: Yeah, that’s the beauty of it, it’s very low impact, and you can install these things very quickly. You’re not reconductoring, you’re not running new lines, you’re simply outfitting existing infrastructure with sensors. This can happen really quickly, and so that’s what’s exciting about this technology.
And so that’s where, as I think about this problem, I mean, building that portfolio of a runway of technologies, things we can do today, things we can do 3 to 5 years from now, things that we can do 5 to 10 years from now, we need to be doing all of those things in concert together because they’re all part of … Nothing’s a silver bullet. None of these things are ever going to solve all the problems that we have. You need all of these things to work together if we’re going to grow the system and decarbonize it simultaneously.
Stephen Lacey: When you sit down in the room with utilities or a grid operator, are they just as surprised as you were when you saw the rise of demand from ChatGPT?
Brian Janous: Yes, and I think there’s a group of utilities that really get it. I think there’s another group that’s sort of getting there, and I think there’s a group that still has their head in the sand of not wanting to recognize what’s happening. So I think everyone is coming along at different paces, but I think you’re seeing it month in and month out, more utilities coming out and saying, “Oh, we need to revise our IRP up again.”
So that drumbeat is really getting out there, and I think it’s starting to truly sink in that we are facing a significant challenge in this country on how we’re going to meet these demands. But again, it’s all driven by a massive investment in new technologies and new jobs. So it’s good for everyone, just it’s going to take some time for us to get there in terms of getting the industry all moving in the same direction.
Stephen Lacey: Brian Janous, thank you so much.
Brian Janous: Thank you, Stephen. Appreciate you having me.
Stephen Lacey: That’s it for the show. The Carbon Copy is a production of Latitude Media. It’s produced and written by me. Sean Marquand is our technical director. He also mixes the show and he wrote our theme song.
We have much more podcast goodness for you at Latitude Media. Of course, we’ve got our companion show, Catalyst, with Shayle Kann, deep dive conversations on the future of decarbonization. We also have The Latitude, which is our show featuring Lisa Jenkins, our editor, reading some of our best feature stories. And so, if you want our news on the go, while you’re listening to your pods, check that out. You can get it anywhere you listen to podcasts or over at Latitudemedia.com. And of course, we got show notes and transcripts for all of these shows.
Latitude Media supported by Prelude Ventures. Prelude backs visionaries accelerating climate innovation that will reshape the global economy for the betterment of people and the planet. Learn more about their portfolio at preludeventures.com.
And please spread the word about this show wherever you are active on these issues, LinkedIn, X, Bluesky. Wherever you are talking about these issues, we would love to be a part of that conversation. We will catch you next week. Thanks for being here. I’m Stephen Lacey. This is The Carbon Copy.
â
