How One Company Is Using AI To Unlock This Promising Renewable Energy

Forbes

One of the biggest expenses in geothermal power is finding the right spot to dig. Utah-based Zanskar thinks its big data and machine learning platform can help.  0:00 Introduction 0:29 Unlocking The Power Of Geothermal Energy 1:58 How is Zanskar Using AI? 4:43 Why Millions Of Dollars Pouring Into Geothermal Power 8:41 How Much Money AI Could Save Producers 9:28 What's Next For Zanskar?  Read the full story on Forbes: https://www.forbes.com/sites/alexknapp/2024/05/06/how-ai-could-unlock-a-ton-of-this-promising-renewable-energy/?sh=510e077e6bb9  Subscribe to FORBES: https://www.youtube.com/user/Forbes?sub_confirmation=1  Fuel your success with Forbes. Gain unlimited access to premium journalism, including breaking news, groundbreaking in-depth reported stories, daily digests and more. Plus, members get a front-row seat at members-only events with leading thinkers and doers, access to premium video that can help you get ahead, an ad-light experience, early access to select products including NFT drops and more:  https://account.forbes.com/membership/?utm_source=youtube&utm_medium=display&utm_campaign=growth_non-sub_paid_subscribe_ytdescript  Stay Connected Forbes newsletters: https://newsletters.editorial.forbes.com Forbes on Facebook: http://fb.com/forbes Forbes Video on Twitter: http://www.twitter.com/forbes Forbes Video on Instagram: http://instagram.com/forbes More From Forbes:  http://forbes.com  Forbes covers the intersection of entrepreneurship, wealth, technology, business and lifestyle with a focus on people and success.

Transcript

00:00 The recent boom in generative AI models relies on intense computational resources built on power hungry chips and server farms around the world.

00:08 That demand could double by 2026 compounding the increasing need for electricity that's already taxing the grid.

00:14 Hi everybody. I'm Brittany Lewis with Forbes breaking news. Joining me now is my Forbes colleague senior editor Alex Knapp. Alex thanks so much for joining me.

00:26 Happy to be here as always. We have been talking about generative AI for a while now. And for those who might or might not know it

00:34 relies on chips which in turn uses a great amount of electricity. You're reporting that one startup is using a to tap into a form of

00:42 renewable energy. Can you talk to us about what this company is doing and what problem they're trying to solve.

00:49 Sure. So one of the most abundant sources of renewable energy is called geothermal energy which is literally the ability to tap into

00:58 the heat that is naturally generated below the earth's surface. And this is used in a large number of applications. But it's actually

01:09 more expensive than a lot of other kinds of renewable energy. And so it accounts for less than about 1 percent of the total

01:17 electricity generated in the United States right now. A big reason for that is that it costs a lot to explore for the right spots to

01:26 dig and drill. If you have a geothermal project and you manage to drill in the right spot the first time that's still about half of

01:35 the total cost of the project. So you can imagine that if you have a few misses even one or two you suddenly massively increase

01:45 the amount of money you're spending on a single geothermal power plant. And that in turn makes it less incentivized for utilities

01:53 and other potential investors to go for it. And so what exactly is Zanskar the startup that you are reporting on. What are they

02:03 trying to solve here that the different places that you can drill. Yeah that's exactly right. So what Zanskar is doing is they're

02:12 using geological knowledge from all sorts of different sources from satellites to actual fieldwork that the company is doing

02:22 to better identify through machine learning where is the right spot to drill for geothermal energy. Where can we maximize the

02:31 return and minimize the cost of digging into the ground and getting the electricity that we need. And so that's the problem that

02:39 they think that they have a handle on solving. And they just raised 30 million from investors to solve that problem. They just

02:47 raised 30 million from investors. But how much did they have they raised overall.

02:51 They've raised 45 million overall so far which they've used to develop their models and even get you know a few dozen employees

03:01 out in the field to validate those models by by testing the sites and making sure that the models are actually sending people to

03:09 the right spot before they do any large term building drilling for geothermal energy without this technology. As you said it's

03:17 risky. So how accurate is this new technology.

03:21 Well it depends in part but at the region you're talking about. So in areas where there's a lot of data they've been able

03:31 to you know flip the script as it were and make it accurate somewhere in the 90 percent range according to one of their

03:40 investors who talked to me in other regions it may be less accurate than that. But that's really because there's not as much

03:47 data available. But even in those places they're seeing significant increases in accuracy. And as they train their models and

03:55 as they're able to gather more data in the field I imagine those numbers would hopefully go up as well.

04:00 What regions can they be drilling. Where is this drilling going to be taking place.

04:05 That's a great question. So the best places are obviously near places where heat is already coming out from the earth's surface.

04:14 So if you don't have a lot of data available but you want to dig a geothermal plant near geysers and hot springs are you know

04:23 obvious places. Those are spots that the heat is already coming up from the earth. But really a significant portions of the

04:30 West along fault lines near volcanoes things of that nature really enable you to hone in and get the heat where it's closer to

04:41 the earth's surface. And is there competition in this space yet or does Dan's car almost have the monopoly on it.

04:48 No there's definitely a competition in this space. We've seen a lot of investment in geothermal over the past couple of years.

04:58 This is in large part because there are startups such as Fervo which is a competitor have developed new techniques for drilling that

05:06 are more cost effective and also enable geothermal drilling to happen in places where 10 years ago we didn't have the technology to

05:15 exploit properly. And so with new technologies for drilling they're able to hopefully compete on that basis. Whereas Zanskar

05:25 their CEO told me they believe they have the advantage is still in the exploration space and being able to identify the right

05:32 spots and compete that way. We have seen over the past couple of years a I encroach into every industry education media

05:42 technology everything in between. And I'm curious what's your take on venture capitalists going into renewable energy via

05:49 A.I. Is this going to be something that we see more of. I definitely think you'll see more of it. So one of the things about

05:57 A.I. and particularly when you're talking about some of the newer generations of models generative A.I. and other models that

06:04 rely on the transformers and that signature Google paper from 2017 is it's really important to train these models on very good

06:14 and very accurate data. One of the challenges you might see if you use you know a chat bot and I've looked up myself and my

06:24 colleagues and discovered you know fake books fake novels things like that. And that's because they're trained on the Internet

06:30 which is not always a reliable source of information. And they're working on different things. But when you're looking at

06:36 fields like energy where you can validate the results and feed the new data back into the machines and train them better. We

06:44 see this in biotech and health care. There are numbers coming out in drug discovery that using machine learning and drug

06:51 discovery has improved the ability of pharmaceutical companies to go through early stage drug testing in the clinic. And

07:03 also in areas like engineering and even engineering solar cells. A.I. is used to develop new ways to engineer solar cells so

07:13 that they're more efficient. I think the key for making A.I. work is the ability to quickly validate the models and retrain

07:22 them with the new information that you're verifying in the real world. And where I think A.I. is challenging is when you don't

07:29 get that immediate validation when you're not weeding out the bad data and not keeping your models trained and constantly

07:38 improving. As you know there have been many conversations about A.I. the pros and the cons. You listed some cons some

07:45 misinformation some bad data. Also people saying hey is this going to steal my job. Do you think A.I. being used in this way

07:54 in the renewable energy space. Do you think this largely goes in the pros column. I would definitely say it largely goes in

08:01 the pros column. I mean whose job are you stealing. This is an area where less than 1 percent of electricity is generated

08:07 already. Any new fields you're discovering and if you're able to make them more cheaply you're only bolstering the industry as a

08:14 whole. I mean the company is raising money and hiring right now. And I think you know as with any new technology some people

08:23 are going to lose jobs. They're going to be new jobs that are created. That really I think as a society we just have to make

08:31 sure that these technologies are being employed to good purposes and we're doing the right thing. And for those who need to

08:37 transition that they get the support they need to transition to new fields. You mentioned at the beginning of the conversation

08:43 that geothermal energy is so expensive because of the drilling process. And even it's so tough to find that exact point. Can

08:51 you quantify how much A.I. is going to save money wise in this space.

08:56 It's difficult to quantify right now without you know the plans with this technology starting to get built which is still a

09:05 couple of years off. But if you look at it this way you know on average it takes you know two to three tries to find the right

09:14 spot. And that and just the first drilling alone if you only drill once that's half the cost of the project. So if you're able

09:22 to get one drill rather than three drills you've immediately half the cost of a thermal geothermal plant. And is there any

09:29 indication I know as you said that we're in the early stages now of what's next for Zanskar.

09:35 What's next for them is what they told me is that they're working to develop their first projects. They hope to get their

09:44 first geothermal plants online within the next three to four years. They told me they've already identified a large number of

09:52 spots and they're working with both utility companies and some potential customers like in the tech industry to build plants

10:00 where the industry may be interested in co-locating data centers or other logistical operations in a place where they have

10:08 reliable 24/7 energy. Alex Knapp per usual. Thank you so much for your reporting. You're welcome back anytime. Great. Thank you

10:16 Brittany.

10:17 Thank you.

10:18 Thank you.

10:19 Thank you.

10:20 Thank you.

10:21 Thank you.

10:22 Thank you.

10:22 [BLANK_AUDIO]

Category

Transcript

Recommended