00:19:35 · Alan Chang
[00:19:35] S0: Yeah, 100%. So, maybe I can first give some background on the state of the grid before and off previously and today, right, and how that's going to evolve. So if you look back maybe more than twenty five years ago, right, of how grid operated, it's actually very simple, right? You have one large very, very large coal plant that can serve millions of homes or hundreds of thousands of businesses, and you send power along this very, very long cable. And if there's any fluctuating demand from temperature, for example, you can rev up the engine on the coal plant or you can rev it down, right? Very, very simple to manage. In fact, you can actually manage on a spreadsheet, right? Now fast forward to today, we've decommissioned a lot of coal, replaced it with solar and wind, which is very intermittent, very weather driven, right? So in order to understand solar and wind production, you really, really need to understand weather forecasting. And this is like a data science and meteorology problem. And also, this is very disputed, right? So coal plants are very concentrated, and usually build it co located with coal mines, right? Whereas, solar and wind, very, very distributed at very micro scale to utility scale. So your energy generation and storage is hyper local, distributed around the grid. And then you have on the demand side, you have electrification AI, right? So electrification, for example, the biggest one is EVs, right? So an EV consumes roughly the same amount of energy as one small house, So if you think about, let's say the domestic market is roughly 40% of the power economy, and if everyone gets an EV, you're basically increasing to 80%. Then problem is that everyone plugs in the EV after work, right? So you have 40% up to 40% spike in energy demand right after work. And then you have AI, right? So AI, for example, during pre training or inference, right, there is intermittent energy demand as well. Right? So for example, GPU is not always doing work. So you have this energy load variance, which is also distributed, right? So you're not building data centers concentrated in just around cities, right? You're also building data centers in the middle of nowhere, right? So you have more distributed generation, more distributed demand, more volatile generation, more volatile demand. So it becomes very, very hard to manage. Right?
00:22:25 · Alan Chang, James Wise
Because the basic objective for the grid is to manage supply and demand, But you have to do it on a local level, not on a national level, on a local level. It's very, very hard to do it and requires very, very high level of sophistication on software, data science, meteorology, and etcetera. Right? So... And you have this AI, obviously, you know, demand skyrocketing. Right? So right now, it's low single digit percentage of the energy economy. You know, we can... There's... You know, depending on who you ask, it could be, you know, more than a 100% of the energy economy. Right? And There are some 20 [00:23:08] S1: forecasts out there. [00:23:09] S0: Yeah. Like, the the... Also, the forecast range is, like, massive. Right? So reality is no one really knows what is the true future, true demand of AI. But I think everyone can agree on it's gonna be much higher. We just don't know how much higher. [00:23:21] S1: I mean, Grok have just opened the first gigawatt scale data center. [00:23:24] S0: Right. [00:23:25] S1: And they're just getting going.
00:23:25 · James Wise, Alan Chang
[00:23:25] S1: And they're just getting going. [00:23:26] S0: Yeah. Like, if you go back to four years ago, right, like, the first data center, talking about, like, it's in the megawatts. And now, years later, we're in the gigawatts, right? And maybe, like, a couple of years down the line, it's like tens or hundreds of gigawatts, right? So yeah, like ultimately, I'd say the problem, like the power cost is only going to go up, right? The power price volatility is already very high, we expect to go higher. And unless someone solves this problem, what I worry is there's going be a lot more you know, not in my backyard type objections to AI data centers. Right? So for example, like, right now, if you look at The US, like, a lot of the data centers are in the Midwest, and we're already seeing power prices in those regions going up, right? This is happening today already. And people are starting to complain, right, because they're competing for the same energy resource, right? And why worry is unless we solve this problem, you're going to have an uprising of riots of people objecting to new data centers, which is obviously not good. But unless this problem is solved, I think there's a very high chance that will happen. Yep. [00:24:52] S1: And obviously, the teams you've got at Fuse are working on solutions to hopefully tackle that imbalance using [00:25:00] S0: software. Software and how are you? Yeah.
00:25:00 · Alan Chang, James Wise
[00:25:00] S0: software. Software and how are you? Yeah. [00:25:03] S1: And obviously right now, Fuse is gonna solve a little part of that problem with home energy balancing, and as you say, smarter software. But to solve it at a macro scale probably does require government intervention and changes in policy. Is that fair? [00:25:19] S0: I think policy will help, but ultimately what matters is the companies that run it, right? And my view is probably like 80% of this solve comes from companies, and probably like 10%, 20% comes from policy. And reality is I don't think the companies today in the energy markets are very are the sophistication required to lower the energy prices. I just don't think we're there. [00:25:50] S1: Do you think we need a change in our generation mix as well? Obviously, fused energy is very focused on renewables, and rightly so. But how do you think about other potentially new aspects of generation, whether that's nuclear, small modular, other ways of bringing energy into the grid?
00:25:50 · James Wise, Alan Chang
[00:25:50] S1: Do you think we need a change in our generation mix as well? Obviously, fused energy is very focused on renewables, and rightly so. But how do you think about other potentially new aspects of generation, whether that's nuclear, small modular, other ways of bringing energy into the grid? [00:26:06] S0: Yeah. I think we advocate for any type of energy more, right? I think it's dogma, right, to advocate for one specific energy technology than the other, right? It's like imagine if you're building a company, right? And let's say you're choosing a database, right? You shouldn't be married to a specific database technology. You're like, Okay, for this application, I want to choose this technology, right? In a similar way for NG Mix, right? There's pros and cons to each energy technology, right? And you can debate for a very, very long time which one is better, but reality is they each have pros and cons, right? So we should have more of everything, right? And my view is we should allow the free market to decide if it's cost effective to build it, the market will build it. The reality is right now, there are more money chasing energy projects than energy projects being built. The market will decide which energy technology is better because they'll make the most returns that way. [00:27:17] S1: Yeah. I like the idea of not being pro nuclear, pro renewables, pro oil and gas, but just being pro more. [00:27:23] S0: Yeah, pro more, yeah. [00:27:24] S1: I think if we can do it in a way which is cheaper and responsible, I think we should be pro more, right?
00:27:24 · James Wise, Alan Chang
[00:27:24] S1: I think if we can do it in a way which is cheaper and responsible, I think we should be pro more, right? [00:27:30] S0: Is And [00:27:32] S1: obviously you've worked in two regulated industries now, which is a rod for your own back. You're very unusual as someone who wanted to do that, both in financial services, a revolution now in the energy industry, a fuse. What role, if any, you think politicians should be playing, do play in working with startups at your scale? Do you find people are responsive and understanding and supportive? Or, you know, is this an area which actually, as an entrepreneur, you think, look, I'm just gonna build, I'll let the politicians do the politics? [00:28:08] S0: Yeah, I think that's a good question, right? Look, reality is, I think trying to influence policy is very, very hard, right? I mean, regardless of whether you're startup or a big company, right? But I think it's even more difficult than as a startup. Right now, what we're focused on is basically within constraints, do the best job we can within constraints. And I think there's actually a lot of things we can do, right, to make it much better. Now would you... If you ask me, would I like to remove the constraints? Of course, would like to remove the constraints, right? But worthwhile constraints, I think we can still do a very good job. But if if there were, you know, let's say, better policy, then we could even do a even more amazing job.