Bitcoin Mining as a Catalyst for Green Energy Innovation | Fred Thiel of Marathon Digital Holdings

Transcripts are autogenerated. May contain typos.

00:00

[Music] and welcome back welcome back to the merge TV uh the North American blockchain Summit today we have uh with us an esteemed uh goat of cryptocurrency today we have Fred teal the uh CEO of marathon digital Holdings uh thank you so much for coming on the show great happy to be here yeah so you've got such a story pass HBS gr uh public private fxy how did you you have to do the Fireside shot with the RFK yep that was great were you here for VC as well uh I was here I heard vix speak yeah yeah yeah it's just fascinating to see such a

00:50

a a a plethora of talent here in the Dallas area talking about something that everyone thought was all but debt for quite a while and you your company has made it through everyone seems to be hopefully hopeful hopefully hopeful yeah it's it's exciting times you know a year ago July so year and a half ago basically we had zero hash rate running and here we are largest Miner of the publicly traded miners from a self- mining perspective uh We've now gone Global we have facilities in UAE and Paraguay and

01:21

continue to expand we're focused Not Just In traditional kind of utility scale grid stability sites but we're also now doing energy Harvest in with methane from landfill gas we have a bunch of other projects uh and kind of taking uh ens bio waste and and methane turning that into energy which we then use to create heat from Bitcoin mining which we then feed back into the process as low-grade industrial heat so heating buildings heating green houses uh supporting uh shrimp Farms um and all sorts of other production processes so

01:57

we view Bitcoin mining as really having an industrial function which is heat generation sure yeah and heat sink and and as you said stability and stabilizing the power grid y absolutely yeah and and so how you mentioned that you're you're Global basically right uh how was that transition from a uh mostly like real estate here in the United States to to worldwide well it all comes down to working with the right Partners so whenever you're going to do something new you ideally want to work with the

02:30

best people possible in the location where you're going uh in the case of UAE we were going into a country with a very specific set of rules uh a specific set of environmental constraints regarding climatology you know it's extremely hot extremely humid you can't use aircooled miners you have to use immersion technology and then uh you know the only real way to be successful was to partner with the Sovereign who obviously has an ownership interest in the power company the energy distribution business and

03:02

everything else and so we were able to um do a pilot to prove to The Sovereign that we could operate in the environment and we did a One megawatt Pilot uh late last year that ran for 100 days before an engineer had to go back on site and they said okay this is the most reliable mining system we've seen and then we were able to um uh finalize our agreement with them in February of this year and here we are barely you know eight months later nine months later and we've gone live on our 250 megawatts capacity there so a you know fascinating

03:39

kind of deal when you look at in the US that deal would have taken two years to do um and then Paraguay again kind of a big hydroelectric Dam using stranded energy and we're doing a pilot there and if that works out well then we'll grow that to also kind of larger scale great so in the I I've been in the industry a long time like you and I've under kind of seen these conversations where like you could have a having a conversation with a sovereign government like that is is wildly speculative theoretical where

04:11

did it you've been in this a while where did it break towards like reality versus fantasy like you know you know what I'm saying like the where did it feel like there was a breakthrough like people these these large uh organizations took us seriously enough as an industry to even have these conversations about uh Bitcoin as a battery and these types of things yeah great question so um I wish I could say we were able to convince them of the benefit of doing Bitcoin mining but they already were orange built before we even

04:41

talked to them so they had done two other prior attempts at doing this and um for technological reasons those um felds didn't work out so they were already orange built it was just they already got it you know they 4 gwatt of power consumption in the summer one gwatt in the winter they still need to run all that power because the Water Dell used the heat so they all of the arguments around grid stability and Stranded energy they were sold on already okay right um so this is not a this is a project which has very defined

05:16

social benefits very defined economic benefits uh and very defined strategic benefits right and so they were looking for the right partner to solve this they had failed twice M and it was kind of they had one last chance one last at bat and so it went from hypothetical to real when we did the pilot and when they saw the performance of the pilot it was okay now we not just have faith that you could do this but we have trust because we've seen the data right and then it was all about having a very transparent and very

05:50

um you know uh equal level dialogue with them about everything we were doing so we were sharing about our technology advancements our investment in or which is a us-based minor technology company which um we were very involved in getting off the ground I just spoke to him he was he was your son of my show yeah we just chatted interesting a yeah yeah I mean it it's you know this industry is really all about how you leverage technology to advance it right because you can only do so many real estate deals around buying power now

06:22

it's advancing to the stage where you have to be have a vertically integrated technology stack you have to be able to operate not just that you utility scale with super high efficiency but also be able to operate at sub utility scale so one megawatt 2 megawatt sit so stuff we're doing with methane gas capture on landfills um the stuff we're doing around energy harvesting where we're collecting uh using for example methane or biogas you feed that into a system that generates heat the heat generator

06:54

is a Bitcoin minor yeah and you use that heat and feed it back into an industrial process that's fast right and so now you're getting to a place where you're not using parasitic load on the grid so you're using you're generating your own electricity and that energy is not costing you anything because you're getting paid to process the material and you're getting paid for the heat so I I've only heard this level of complex you know recapture design like I I've covered Eder brise Tech going back

07:25

to 2009 or so and uh only have I heard this like from HP and IBM in their their like their their experimental data Labs you know the HP Research Center in paloalto uh So within even Enterprise is kind of rare to attempt this level of like complex systems even so much more so rare in mining what has been like the driving motivating factor to kind of go for these complex systems is it by necessity or by like a a mission driven Focus that you guys have to really make it as green as possible or something well we made a

08:05

commitment um two years ago that we were going to be fully carbon neutral in our operations and we believe that Bitcoin mining can not only be carbon neutral with carbon negative and we also believe that Bitcoin mining can be energy neutral meaning we generate our own energy and we don't have to take load off of the grit we think that solves for all sorts of issues um also I've been in tech for 40 years and some of the first work I did in Tech was in the supercomputing world uh and the company I worked with at the time

08:37

was focused on deploying the industrial control systems for CERN in Switzerland you know uh singer link flight simulator for the F-16 so we were doing very very sophisticated control technology and as you look at how technology can change Industries um it's pretty amazing I've also spent a lot of time in the Industrial Automation and process control world through my efforts in iot Internet of things and you know this energy harvesting is the perfect example of that um maybe it's my Swedish background we're very focused on the

09:11

environment and Engineering stuff but um you know we believe that by Leading with technology and by Leading with a very big focus on green and sustainability that we're able to differentiate ourselves from everybody else kind of in the industry who's really focused on being big utility scaleman uh one of the emerging uh narratives that I I it's been it's been ruminating around I think for a little bit now but like we we've talked about it about three or four times in the show this week is the idea that uh the AI model

09:45

training system uh is following a similar track of evolution to bitcoin data M uh data centers uh what are your thoughts on that like is that is that something you guys are looking at is that something you're thinking about uh or do you disagree with that narrative so maybe chalk this up to having all sorts of battle scars from being an early mover in Industries and just before I kind of answer the question um I took a company public back 23 years ago which was one of the first companies to build embedded communication

10:20

Technologies for iot systems so you could essentially connect anything to the internet right right we built some of that first technology company still exists today it's name is lonic um and they still do exactly that um we felt certain in the year 2000 that by 2005 a billion devices would be connected to the internet remember that everybody was going to have it right here we are 2023 almost 2024 and I think people are now saying 50 billion devices and we're still not there so these things take time M the technology goes into waves you've

11:00

heard of the theory of scurve where you know you do an initial so if you look at AI data centers and you look at Bitcoin mining data centers what do they have in common there's only one thing power density right right Bitcoin mining data centers interruptable load you can shut them off at any time right if you're running an llm you can't shut it off time you need Network ping of 35 n milliseconds or less so you need to be close to a point of presence the design of your data center in the Bitcoin world we have tier zero

11:38

meaning it's very simple low complexity we use containers we spread containers out over a field we use immersion technology right sophisticated but low cap X you're talking about something that's you know half a million dollars a megawatt of infrastructure an AI data center is 8 to 12 times more expensive to build just just the infrastructure now let's start adding h100s by some people's estimates you're talking 49 million dollars a megawatt w when you're all in now the h100 is really a first gen machine the

12:16

h200 already exists the H300 will be announced soon if you build these data centers and you're stocking them with h100s expecting to have a threeyear or four-year contract life when you're a customer you've got two things working against you the guy using the h100 today is going to want to run an h200 tomorrow and they're going to want to run an H300 the year after that so what's the likelihood they're going to stick around until you get paid back on all the infrastructure you're invested in the

12:48

next thing is the technology stack in llms is changing so you're now putting Vector databases below it that increases the performance of these llms by factors of magnit and when you then stack these things these generative models and you essentially have them fighting against each other like ganss work right right now all of a sudden you're speeding up this even more and while the complexity of the models will continue to grow right GPT 4 was I think four factors bigger than gpt3 GPT 5 is about to be

13:25

announced and you know GPT is effectively writing all the future gener ations of GPT um the hardware curve has to follow so as we look at it we think you know initial technology wave hype cycle trough of disillusionment then we go back up and to the right hopefully uh we're still climbing that high pill right we haven't gone down the crash so we're not chasing this Market at all okay other than we are building cooling technology for the industry leveraging our immersion technology from our Bitcoin data centers which is applicable

14:01

to the AI space and so we're addressing the market that way can I pivot the conversation back a little bit to um money supply to monetary policy to cryptocurrency as a a whole I I know that you've been uh pretty active in the trying to get like you know some basis on institutional adoption and uh as it as c as well I see that that's also a slow moving I had a conversation with a client of mine who was high level executive at a large insurance company and it was free SPF blow up and they were this close to

14:42

saying yes on being able to accept payment for specific policy so on and so forth and the problem they had was the comfort of having it on the balance sheet right so it's a