← Blockspace: AI & Bitcoin
Blockspace: AI & Bitcoin
Marty Bent on Crypto Laws, Jameson Lopp on AI Agent Safety, Bitdeer’s 180 MW AI Site
01:21:50
Google is moving up their timeline for quantum readiness, and Elizabeth Warren has her sights set on the Trump family’s business dealings with Bitmain.
Loading summary
Transcript113 lines
- [00:00]
Alex Pruden
Foreign.
- [00:05]
Host 1 (Possibly Matt)
What's up, y'?
- [00:06]
Alex Pruden
All?
- [00:06]
Host 1 (Possibly Matt)
Welcome back to Block Space Live, brought to you by Clean Spark Cash App just rolled out default bitcoin payments for any merchants who wish to accept that. And it's coming at a great time because half of all bitcoin holders are actually underwater and they can't spend those SATs anyway. So now they're going to be forced hoddled. So good move. By Cash App. They're actually, they don't, they don't want you to actually spend your bitcoin. They're ruling it out at a time when you, when you can't because you're not, you're not, you're not wealthy enough. But we've got that as our lead story today. We've also got three banger interviews coming up. First up, we've got Marty Bent to talk about the Clarity act and TFTC's reporting on the bitcoin de minimis tax exemption or lack thereof, I should say. We have Jameson Lopp on to talk about how AI agents are creeping into every corner of your online and technological life and the ramifications for that. We've also got Alex Pruden of Project 11 to talk about bitcoin and Quantum. And for our final two stories on the docket today, we're going to be talking about Bit deer's data center, 180 megawatts in Norway, one of the bigger ones for their AI expansion. And to cap it off, a little red scare always does an American good, especially bright and early on a Monday morning. We're going to be talking about Operation Red Sunset and why Elizabeth Warren thinks it's really important to focus on Bitmain.
- [01:30]
Host 2 (Possibly Charlie)
Right now, you are watching Blockspace Live. Blockspace goes live Monday, Wednesday, Friday at noon Eastern, featuring quick hits on the latest in bitcoin mining, AI, bitcoin tech, and sometimes we throw in a culture story. Make sure to hit subscribe if you're watching on YouTube, comment if you're on Twitter, click the bell on YouTube to get notified when we go live. This is also a podcast, so right after we wrap up, this will be on your RSS feeds anywhere you find podcasts. If you like this live show, you'll Love our newsletter. Newsletter.blogspacemedia.com and lastly, we have a conference in two and a half weeks op next April 16th in New York City at the Times Center. For more on that and Tickets go to opnext.devo pnext.dev It's a Bitcoin technical conference for investors. This show is brought to you by CleanSpark ticker CLSK on NASDAQ. More on them later on the show. Let's kick it off. Colin.
- [02:29]
Marty Bent
We're poor.
- [02:34]
Host 1 (Possibly Matt)
You know, we're less poor today than we were over the weekend because bitcoin's had a little pop because apparently there's, we're reaching negotiation, you know, time for the war in Iran. I don't know how much I believe that, but that's not really the lead story. We're going to first talk about this interesting Cash app rollout. And this is building on something that they've had in the hopper for a while. Oops, sorry, guys. Let me get this up. Okay. Oh, there we go.
- [03:08]
Host 2 (Possibly Charlie)
There we go.
- [03:09]
Host 1 (Possibly Matt)
Thank you, Charlie. So Cash App rolled out bitcoin payments for basic for a bunch of their merchants, I believe at the end of last year, beginning of this year is a few months ago where they were kind of doing this pilot where they're saying we're going to start rolling out bitcoin payments by default. They started with a few storefronts and a few brick and mortar places and now they are making it a default option for anyone who wants to use it. This is from Miles Suiter at Cash App Quote. We're making it easier for millions of businesses to accept bitcoin. Starting today, eligible U.S. square sellers will begin having bitcoin payments automatically enabled. So sellers who accept bitcoin will receive USD on the other side of this. So you can pay in lightning on chain, but it'll settle in bitcoin or settle in USD by default. You can also keep the bitcoin if you want, but kind of a nice little white pill to wake up to. You know, it's when Square announced this feature, I was really excited about it because it's a really seamless way for you to pay in bitcoin through your Cash app app or through other wallets at these square terminals. And it's an easy way for the actual seller to take us D value from that transaction rather than having to be exposed to bitcoin risk. But we'll see how many people use it.
- [04:22]
Host 2 (Possibly Charlie)
Yeah, I'll add a little bit of color here because I think a lot of people might be like, well, didn't they already do this? And let me take you, listener, on a little journey through the world of point of sale. I, as a former brick and mortar business owner, also actually used square terminal back in the day. So what happened this last year is that Cash app square really block and then their square terminal Made it an option for business owners to go through a couple menu workflows and then click on Bitcoin payments. So they had to enable it and like they had to opt into it, but now it's enabled by default. That's what the announcement is. And as Miles Souter says, the head of product for Square here, these are millions of businesses that this is a lot. And it's also like wallet agnostic. So you don't have to have Cash app, you can just have any bitcoin wallet. A little detail is that it looks like, I thought it was like poof, overnight, you know, sometime around like midnight Eastern. It like they'd all immediately turn on all the terminals. But apparently that's not necessarily the case. Miles Souter had a tweet this morning which clarified that they're just starting to roll out. So it may not be immediately available right now across all million plus Square terminals that are live in the US but over the next month they're turning this on. So pretty cool. I mean I will say one of the curveballs, a lot of people don't know for some reason maybe Cash App and Square hasn't pushed this as much as if you're the first person who transacts at one of these businesses in bitcoin, they actually Cash App will drop you $25 in Bitcoin into your Cash app. So there's actually an incentive for you to go and get the neighborhood pizza parlor to go turn on their bitcoin payments. Now maybe that's not going to be the case for right now as they, as they make it like default, but it has been the case. I actually made money on pizza a few months ago, if you can believe that. I was the first person to pay for pizza in bitcoin and then I got $25, which was more than I paid in pizza. Honestly should have been leaning into that racket more.
- [06:51]
Host 1 (Possibly Matt)
It's like Pokemon Go but for, for actual things that matter with real world consequences. There's interesting gamifying that and it's a good incentive to get people to start spending bitcoin at these places. But as we were saying at the beginning, maybe they don't want to because they're actually underwater on their bitcoin holdings. This is technically our lead story for today. Just a little banter to start up with Cash app. Cool to see, but this is coming from CoinDesk. Nearly half of all circulating bitcoin is underwater as long term holders sell at a loss quote. Nearly half of all bitcoin is now trading at a loss with the Bitcoin impact index surging to 57.4, indicating high levels of stress. So if we look at the underlying study here, and this comes from sex. IO.cex.
- [07:47]
Host 2 (Possibly Charlie)
iO.
- [07:48]
Host 1 (Possibly Matt)
Yes, thank you.
- [07:51]
Host 2 (Possibly Charlie)
Don't be Googling that other thing, you know.
- [07:53]
Host 1 (Possibly Matt)
Yeah, don't. Don't be Googling that other thing. So they basically say right here we're in a high impact range for, for bitcoin holders being underwater on their bitcoin holdings. And they point out in this report that usually when we're at this stress level, it precedes some sort of deleveraging or unwinding event maybe similar to what we had at the beginning of February. You know, if you're hearing that and your stomach's curling over, can understand why you don't really, you don't really want to see that after we already had a crazy drop in February. But I wanted to verify this finding from them and I went over because I'm not sure about their methodology. But I did go over to Glass Node this morning and if you look at their supply and profit, really interesting metric here from them, we're at about 55.62% currently. And I'm wondering if this report did draw from that. I did not see anything in the report about it drawing from Glass nodes analytics or if it had repurposed them for themselves. But we're at 55.62% as of today and it has been slipping for the last few weeks. We were at 61% in March, on March 15th. Now we're almost at that 50% threshold. So doesn't surprise me, given where we are, you know, obviously. But it's just a reminder that telling someone to zoom out doesn't really work. When they decided to plow like 50 grand into Bitcoin when it was at $100,000, right?
- [09:31]
Host 2 (Possibly Charlie)
Yeah. I mean, we are now down what, 50%. Bitcoin's at 67%. We peaked at 126. Okay, I'll just point like, I think, you know, my view is that as time goes on, the on chain metrics have become less meaningful because bitcoin supply is moving to the custodians. It's moving to like the ETFs. The larger, the, you know, the, the institutions, they're the ones buying bitcoin. We don't necessarily see their cost basis as clearly on chain. And these are on chain analytics from Glassnode, which if they haven't changed the heuristics since I last checked is mainly looking at when bitcoin was last moved and implied to have been purchased. So it'll assign a price to the bitcoin as it moves from known purchasing venues. But when we're talking about a world in ETFs, it's probably not as insightful as it used to be. However, like it's still really useful. You, you can look at the data and you can see also empirically like all the people I know who bought over 100k are down. So this vibes with my gut check, Colin.
- [10:54]
Host 1 (Possibly Matt)
Yeah, me as well. And I will say to your point, as bitcoin trading moves off chain, particularly in these paper products like the etf, it is less useful. It's also less useful when you consider I could be just managing my stack, I could be consolidating UTXOs, I could be changing to a new wallet and I can move bitcoin and it would show up on one of these as having coin days destroyed.
- [11:18]
Host 2 (Possibly Charlie)
Right.
- [11:19]
Host 1 (Possibly Matt)
And then so the cost basis for when those coins were created for that UTXO would be different from I might have put them into a wallet when bitcoin was at 10,000, I moved them when it was at 60. It'll show me. It'll show the cost basis for when those coins removed at 60. And that's not really necessarily the case. So it's like you said, Charlie, I think already on chain analytics were a little dicey, but increasingly as bitcoin continues to get financialized, they are becoming less and less so. Less useful.
- [11:50]
Host 2 (Possibly Charlie)
Yeah. And you mentioned again our boy Chuck Matey, my favorite on chain analyst, put out this chart describing this concept of time pain, which I think you've been in bitcoin a long time. This resonates a lot because it's almost the boring times when nothing special seems like it's happening. Those elongated drawn out periods where just volatility isn't super interesting. Bitcoin's not crashing or ripping and you just have to sit on your hands. That may be the period that might be the meta for a little while. So psych up.
- [12:32]
Host 1 (Possibly Matt)
Yeah, I definitely saying that might be the meta for the while I think is absolutely right. Because just one more note from this coindesk report. It's like over 4.6 million Bitcoin from these wallets in the red. And realized losses are the worst they've been in like three years since 2023. Now the bull side of that is what happened after 2023. So if you want a little opium,
- [12:52]
Jameson Lopp
there you have it.
- [12:53]
Host 1 (Possibly Matt)
But the bear side of that is we also saw similar levels of losses around 2022 before the FTX debacle and Bitcoin shipped the bed.
- [13:03]
Host 2 (Possibly Charlie)
I'm tired of talking about price calling. Let's actually talk about something meaningful. We have a guest. Yeah, let's talk about something meaningful.
- [13:09]
Host 1 (Possibly Matt)
Let's talk about politics and legislation. And with that, we will welcome Marty
- [13:15]
Host 2 (Possibly Charlie)
Bent, our first boy. Marty, welcome to the show.
- [13:18]
Marty Bent
Politics, such a fun topic. Gentlemen, thank you for having me.
- [13:24]
Host 1 (Possibly Matt)
You know, we thought we would start off the week right, you know, get, get people feeling really good about the representation that they are currently experiencing on Capitol Hill. As always, the politicians are only ever looking out for you and definitely not on your side. Definitely not looking out for the leviathans of business. Marty, you've been tracking the Clarity act pretty closely. You've had some good original reporting on it, and I would like to just open with that. Can you give our listeners a rundown of some of the reporting that y' all have done at TFTC for the Clarity act, specifically this de minimis provision?
- [14:01]
Marty Bent
Yeah.
- [14:01]
Host 2 (Possibly Charlie)
And so I think to be clear
- [14:03]
Marty Bent
to your point about D.C. and politicians working for us, one thing they don't make it easy to do is know exactly what's going on with anything. And so as a. As it pertains to de minimis, I think it is separate from Clarity act, specifically, I think Clarity is more market structure and some other things that, that are being handled with that tranche de minimis. It looks like the parody act. Now, they come out with the, the draft legislation for the parody act at the end of last week. And so I think off the bat, I think they're separating Clarity from the de minimis tax exemption. I think tax. Something that's handled separately from all these other laws that they're working on. But as it pertains to de minimis, obviously, I think a lot of these conversations are happening in parallel on Capitol Hill, predominantly via lobbyists and staffers on the Hill. And so I think they're talking about a bunch of these topics at the same, same roundtables. And people then will just go to the tape and say, oh, they're talking about all this at once. It must be in the same act. They're going to be in separate acts. But from what I've been hearing, and there's been rumblings for a couple of months now, as it pertains to the de minimis tax exemption is that obviously Coinbase and others are in a battle with the banking lobby to get A share of the yield generated from stablecoin reserves held in Treasuries, sharing that with their end users. The banks apparently don't want that. And I've been hearing reports that there's a lot of horse trading going on in the background in terms of the concessions that will be made on both sides. And the rumor I heard is that for those who are lobbying for the ability to share yield from stablecoin reserves, they're willing to concede on de minimis tax exemption and thereby saying, hey, we don't need a de minimis tax exemption for bitcoin transactions. Why don't we just do stablecoins? And obviously there are positive externalities for those who run businesses that monetize stablecoins if de minimis is only given to stablecoin transactions. And I believe gas fees as well. So that's what I've heard. I've heard. I mean, I reported on it, so it's already public. But I had some sources reach out and say that one of the lobbying firms working on behalf of Coinbase that explicitly say, like, hey, let's not go for bitcoin de minimis. That's going to be too hairy to tackle and it's a handout for. It'll be viewed as a handout for bitcoiners. Let's just focus on stablecoin de minimis exemption.
- [16:59]
Host 1 (Possibly Matt)
I appreciate the clarification there between the Parity and Clarity act, because I was unclear on that point. And so thank you for clearing that up. But am I understanding this correctly that you're talking about? There's some horse trading going on, but there's almost like Cross Bill horse trading here, right, where they said, excuse the balloons, gentlemen, but basically the idea being that coinbase will back down on the stablecoin yield provision, but if they can get this win for stablecoin de minimis at the expense of bitcoin de minimis and the Parity act, they're more okay with it. Is that generally what you understood from what you were told?
- [17:33]
Marty Bent
That's. That's the vibe I'm getting. Again, I'm not in these rooms, but I think if you're just looking, looking from the outside and you. I think a lot of bitcoiners have become more privy to how the sausage is made on Capitol Hill. That's what I imagine the conversations are like right now.
- [17:54]
Host 1 (Possibly Matt)
I would love to be a fly on the wall in those, just to see not only how the horse trading goes on, but people's demeanor during it. You know, Are they just. I always wonder this too, with when, when countries are negotiating during war, you know, it's like you're blowing each other up and then you just like get on the phone or, you know what I mean? Or you sit down at a roundtable and you just talk about how to end this. I wonder. I would be really curious to see what the attitudes are like during these, these discussions. And when you reported this, attitudes markedly sour from the Coinbase side, as you would expect. Brian Armstrong came out and said, this is not true. A bunch of other people from Coinbase said this is not true. Some folks from bpi, I think Zach Shapiro tweeted, Unfortunately, I'm hearing this as well. What do you say to the folks on Coinbase's side of things who have come out and basically said, nah, this is fake news.
- [18:45]
Marty Bent
I mean, I've got sources and pretty solid, I guess, circumstantial evidence that they're saying something else behind the scenes. Like, I don't. And who knows, Like, I believe there's a ton of different actors and who know, maybe there are people on Coinbase's policy team, their internal policy team, that don't work directly with the lobbyists. I have no idea. Like, I wouldn't be surprised if Brian Armstrong didn't know this was going on behind the scenes, him specifically. But yeah, I would say, like, the proof is in the pudding. The parody, the first draft of parody legislation which covers this de minimis tax exemption. You, you have to imagine it's been worked on for, for many weeks now. It was released on Friday and it has language that I described when I broke this story a couple of weeks ago. And so I guess the proof is in the pudding here. It doesn't, like the minimus exemption for stablecoins doesn't make sense. Yes, there are minuscule differences in the peg, but below or above a dollar. But I think if you're really getting to why people want the minimus in the first place, it's really for bitcoin, specifically, using bitcoin as money. It's been well known for many years that people don't spend bitcoin because they're worried about the overhang of having to deal with the tracking of cost basis and sales to pay capital gains tax when they want to go buy coffee with bitcoin.
- [20:18]
Host 2 (Possibly Charlie)
Let me double tap on this because a lot of the initial controversy was you guys saying, we're hearing things, reliable sources, Coinbase folks saying, no, this is totally made up. But part of the problem is with lobbying it's not like people write their names down on a piece of paper and say I'm currently lobbying for these things. A lot of this does happen behind closed doors. What should we look for as like to signal like who is, who is? I would say who should I trust to lobby for good, healthy bitcoin legislation here in America?
- [21:01]
Marty Bent
You don't trust, you verify, right, Charlie?
- [21:04]
Host 2 (Possibly Charlie)
How do I verify?
- [21:07]
Marty Bent
I think you verify by just drawing a hard line. Because that's been the frustrating thing about all of the legislation as it pertains to crypto and bitcoin is that it's been heavily focused on crypto. And I think the companies involved on lobbying on behalf of crypto have vested interest and more importantly a ton of capital to really force things in a certain direction on Capitol Hill as it pertains to policy. And I don't have an answer for how we verify and how we know. Again, politics isn't really my beat. I think we've all been dragged into it just because of the recognition that you may not care about politics, but it cares about you and you have to be vocal and try to engage as much as possible. And to your point, how do you verify? I think you draw the hard line and say hey, and actually I said it when I made a video on the matter a few weeks ago when I was in Austin. I'm actually taking a page from Brian Armstrong's book himself. He drew the hard line on clarity with the stablecoin yield sharing and said I'd rather have no bill than a bad bill. And I think as bitcoiners, if somebody truly honestly doesn't care about stablecoin regulation, I think stablecoins are certainly an innovation. They make payments more efficient, they enable cross border payments in a way that's never existed before. But at the end of the day they're centrally issued and controlled monetary tools that are no different than a CBDC at the end. Bitcoin is truly a different animal. And if policy is being written and passed basically to cater to this stablecoin industry over bitcoin as bitcoiners. As somebody who likes bitcoin due to the sovereign distributed neutral reserve properties, something that's truly distributed and cannot be stopped and people want to use it as everyday money, obviously. I'm sure I see you guys have cash app here. I assume you're talking about Square enabling it for their merchants today, like people want to use it as money. And if there is going to be a law on tax policy, particularly with de minimis exemptions for payments using Bitcoin. Like we should just draw a hard line and say I'd rather have no bill than a bad bill. That sort of pigeonholes de minimis tax exemption for stable coins only. And then on top of that, I mean I'm sure you guys are aware too it's not only that it's like it gives preferential treatment to staking rewards over proof of work mining income too. So that double tax on mining would still exist with this parity act like legislation as written today. And I think how do you verify just be a stick in the mud
- [24:01]
Host 1 (Possibly Matt)
like say that's crazy to me that about that that staking rewards and mining are somehow treated as any is as differently at all.
- [24:09]
Marty Bent
Yeah.
- [24:09]
Host 1 (Possibly Matt)
You know, I know you said Marty, that you, you know, politics is outside of your normal beat and I'm going to kind of throw a curveball at you because this is on the docket today too. Did you see the, did you see the Bloomberg article about Elizabeth Warren sending a letter to Lutnick about Bitmain? Did you see this?
- [24:31]
Marty Bent
No, I didn't.
- [24:33]
Host 1 (Possibly Matt)
Okay, well maybe I shouldn't put you. I, I would just love first takes on this. I'm not going to ask you a specific question.
- [24:38]
Marty Bent
I've got, I've got the kids have spring break and first day of spring break came with the stomach bug to start at 12:30 last night. So I've been in the trenches of stuff coming up.
- [24:50]
Host 1 (Possibly Matt)
I was gonna say good for you for being on spring break with your kids and not being plugged into politics. But also it sucks to have to deal with sick kid.
- [24:58]
Marty Bent
But what's the gist?
- [24:59]
Host 2 (Possibly Charlie)
Is she.
- [25:00]
Host 1 (Possibly Matt)
The gist is that. So there's this, this thing called Operation Red Sunset, which is a poorly defined and I would say poorly verified Department of Homeland Security initiative looking into Bitmain, specifically Bitmain's ability to coordinate brownouts in the US or steal military secrets. You know, there's a New York Times article back in 2023, I'm kind of front running this segment at the end of the show but be covering a lot of it then too. But the idea basically being that somehow Bitmain machines at these Chinese run mines pose a national security risk for the US And Elizabeth Warren apparently according to Bloomberg penned a letter to Howard Lutnick basically excoriating him for the Trump's ties to Bitmain, that is through American Bitcoin. Eric Trump's bitcoin mining company or a company I should say that Eric Trump serves as an executive for with the idea Basically being looking at Bitmain and trying to point the finger at Bitmain and Trump simultaneously, but basically making this argument that somehow Bitmain is a national security risk. Before you go, Marty, I'd love to get your thoughts on this as a angle that Pocahontas is taking.
- [26:15]
Marty Bent
I mean props for props are due. I mean it's, it is an angle that you have to think about because you guys remember amp lead back in, what was that 2015 Bitmain did have? I mean, who was it Greg Maxwell that discovered it? They had the ability to remote brick their machines. Whether or not that could disrupt the power grids, I think is another question. It's an a. I think it's a stretch. Like, I don't think, I think Bitmain is motivated by profit and I would find it hard to believe that they're infiltrating U. S Energy markets to, to brown out the grid. But if you are going to pick an angle, I think the, I'm not trying to give Elizabeth Warren any ammo here, but the amp lead thing is like, ah, there was something that, that proved that Bitmain did have remote access to these machines at one point in its history.
- [27:09]
Host 2 (Possibly Charlie)
I'll wrap up with a throwback to kind of the original topic we were talking about, Marty, is a couple of years ago, as the broader crypto industry felt incredibly targeted, it felt like leading into this last presidential election there was a lot of big tent linking of arms. And with this break in de minimis and clarity act and obvious clear industry focus on stablecoins, it feels like we're back in the trenches kind of fighting on these issues again. Do you feel the same way? Did you ever kind of feel this big tent view with like broader crypto or. I'm kind of curious on your take on what the rest of like this administration like crypto infighting might look like.
- [27:55]
Marty Bent
Yeah, I was definitely cautiously optimistic
- [28:01]
Alex Pruden
going
- [28:01]
Marty Bent
into Trump's second term. But then when she saw like the crypto briefing or like the crypto roundtable and you had like the chain link guy and Brad Garlinghouse and Brad, that slimy guy and then you know, like World Liberty Phi, I think that's just like a complete joke, like the Trump should not be involved with that at all. And I think to answer your question directly, like I was cautious, cautiously optimistic going into it, but then as soon as he got in office it quickly sours like, oh wow, the, the, the parasites have, have surrounded the host and are going to throw money around their lobbying efforts to, to try to win influence And I, I think that's what's happened. And I, I mean I'm considered a bitcoin maxi. Maybe somebody call me toxic but I mean I've been calling Coinbase out since they ninja launched bcash on like Christmas Eve, saddled me with a big tax
- [28:58]
Host 2 (Possibly Charlie)
bill, dude, I was.
- [28:59]
Marty Bent
Yeah, yeah. And it's like, it's, I'm, I, I would like everybody to be big tent. But again, these people have perverse incentives. These companies specifically if they're leaning into stable coins and broader crypto and bitcoin is emerging as this sort of more its use cases everyday money is becoming more prominent that sort of eats into their potential business plans if they're not going to focus on bitcoin as a medium of exchange. And so I do think there is perverse incentives between bitcoiners who simply want bitcoin to be a sound money for the digital age and those who want to ride in bitcoin's coattails and affinity scam. And again, I think we can do this. I think we out kick our coverage in terms of influence on Capitol Hill because I had people on the hill reach out to me so that they heard our reporting and they're, they're reacting to it because they don't want blowback. And so I think again going back to how do you verify? Just be the stick in the mud, draw the red line. If it's a bad bill, don't support it and fight for what I think is fair in this overarching battle of crypto policy, which is like if you're going to do all this stuff for stable coins, defi and market structure and all that, just make sensible legislation and policy around bitcoin as well. Let's use bitcoin as money.
- [30:37]
Host 2 (Possibly Charlie)
Let's just use bitcoin as money. Marty, thank you so much for coming on the block. Space pod. Best of luck. Have a good spring break, gentlemen.
- [30:44]
Host 1 (Possibly Matt)
It's always cheers, Marty. Thanks man.
- [30:46]
Marty Bent
See it.
- [30:49]
Host 2 (Possibly Charlie)
All right, we got another titan in bitcoin. We have Jameson Lop in the audience. We're going to talk about AI, so let's bring him up here. Jameson, welcome to the show. Thanks for hopping on. How's it going?
- [31:06]
Jameson Lopp
Well, it's exciting times, but as a security professional, exciting is not good.
- [31:17]
Host 2 (Possibly Charlie)
And which is great because great lead in because you last week published a blog for CASA on a topic a lot of us have are a little bit aware of, but you went into more detail about just the risks that running AI internally on, especially like secure infrastructure like Risk like important infrastructure can pose. Can you, like, describe the, you know, the arc of this blog post that you wrote?
- [31:51]
Jameson Lopp
Yeah, I mean, we've. We've had internal initiatives for nearly a year now to explore what's happening in the AI space and understand, you know, what can these tools really accomplish in terms of productivity. And, and it's clear that they're quite powerful. But the flip side, any tool can be used for building or destroying, for good or for evil. Right. And so these tools are. They're not malicious, but they can effectively act as a malicious agent while trying to accomplish the task that you give them. And there's many reasons to this. It's very complicated space. But the big thing is that all of these agents and models, they're black boxes. Even the open models are black boxes in the sense that basically nobody really understands how they work. We understand that you can query them with things and that you can get stuff back as a result, but it is a magical black box. And the result of all of this is that it is a rapidly changing security landscape where, like, the AI stuff, the capabilities of these AI tools continues leaping forward. And as security professionals, we're just trying to constantly catch up. And we are way behind just in terms of creating a set of standards and best practices for how to deal with this stuff, because it's the best way. I can figure there's a lot of analogies you could come up with, but it's kind of like giving a bulldozer to a toddler. You know, you could do some really cool stuff. You could also do some pretty terrible stuff if you're not careful.
- [33:54]
Host 1 (Possibly Matt)
Wait. Sorry, Jarl, I just have to double tap on something. Jameson, you said nobody understands how these things work, and it's a black box. Are you expanding that scope to the engineers designing these LLMs as well in these agents? Are you saying that fundamentally we don't really understand what's going on under the hood?
- [34:12]
Jameson Lopp
Yeah, I mean, obviously the AI researchers and the safety folks have a better understanding. They're the ones who are trying to train it and guide it and put push it in certain directions, but they're still getting surprised all the time. And that's just another aspect of this whole space that is concerning. But, you know, it also, on the flip side, it means that we may end up getting some really cool, revolutionary, groundbreaking breakthroughs in various scientific applications as a result.
- [34:47]
Host 1 (Possibly Matt)
I'm just imagining, you know, like Oppenheimer or some of the guys in the Manhattan Project being like, well, you know, we get this Nuclear vision response from this and it blows up. But we don't really understand how it works. But we do know that it blows up, you know, but you don't actually know what's going on under the hood. That's kind of quite frightening to me.
- [35:06]
Jameson Lopp
You know, they're kind of, there was an issue like that, you know, during the, the early days where they didn't understand necessarily how, how far the chain reaction was going to go. Like some of them weren't sure if blowing up a nuclear bomb would end up consuming the entire world.
- [35:24]
Host 2 (Possibly Charlie)
Yeah.
- [35:24]
Host 1 (Possibly Matt)
There's this brilliant scene in the movie where he's talking to whoever Matt Damon's character is, he's either a colonel or general and he's like, so there's a chance that it could just destroy everything. He goes, yes.
- [35:37]
Host 2 (Possibly Charlie)
So you know, back to the article, James, and you kind of outline what almost like a reverse HAL 9000 situation where instead of the AI saying I can't do that, it's like, well I actually can and I'm going to use means you did not expect to accomplish that. Explain this. And how do you think about what does this mean from a security standpoint?
- [36:03]
Jameson Lopp
Yeah, so especially once you get into agentic AI where it's not just you querying a model and getting a response back, but rather you have entire self contained processes that have some set of context or memory and can do more complicated series of actions, string them together. That's when kind of like the toddler and the bulldozer thing, you have to kind of look at these tools as well. It's kind of like giving access to your entire computer or potentially your entire production infrastructure, depending on where you're running it to an intern who is their first day on the job and they haven't been around, they don't have understand all of the nuances of your company and your infrastructure. And it's like if you give a powerful development tool to an engineer who has been working somewhere for a year or multiple years, they have this massive sum of accumulated knowledge based upon working in the code and the infrastructure and understanding the business. And this is a massive context. But the context that we give to these LLMs is often quite tiny in comparison. Maybe we give them a few sentences of guidance, maybe we come up with a skills MD file that is several pages. But even then you're probably, hopefully we get there eventually, but we are a long ways away from the point of being able to feed years and years of experience into one of these AI agents. And so the end Result is you should expect that they are just going to do things naively and not follow what a normal human might consider to be best practice or just common sense even. And so this is where the state of security in this space gets really gnarly because it's like, all right, how can we leverage all of the cool stuff that these tools can do while also constraining them and basically like building layers of firewalls and layers of essentially like common sense checks to make sure that they don't go rogue and end up deleting, for example, your entire infrastructure, which has already happened to several different companies in the space.
- [38:34]
Host 1 (Possibly Matt)
On that note, James, I'm glad you brought that up because if I'm looking at this as a kind of, I would say a noob when it comes to some of these AI tools, or rather someone who uses them, but maybe not as complex as I would like in my day to day tasks. What do you say to someone who's like interested in using one of these agents and giving it access, root access to their computer and to their, to their different applications? What would be your advice to them to mitigate some of the disaster scenarios? Or would your advice simply be just wait until we figure this tooling out?
- [39:12]
Jameson Lopp
Yeah, the short version is when it comes to running LLMs on your laptop or desktop, if you're going to do that, you have to understand first of all that's a lot more dangerous. And you should only do that within the context of a sandbox environment. You have to set it up and say you can only work within this directory structure and you can't make any other calls. You are not the super user admin. You don't want the agent itself to be able to do everything that you, the human, can do without asking you, the human. So this is one of the principles that keeps coming up in the space is the human in the loop? Is that yes, these things are very powerful, but they need supervision. And so if you don't have a human that's looking at every important decision that the agent is making and the human is saying, yes, that's a good decision or that's a bad decision, that's when things can go haywire. And so the problem that we ran into, I mean we've actually, we've had a number of different incidents of just like weird things happening. But this is the first real security concern that we've had. But we had a non technical employee who wanted to improve the efficiency that they could generate reports using data from some internal dashboards. That we have. They got their agent set up. They said, hey, this is the internal dashboard. This is how I want you to format a nice pretty report for me. Go. The agent of course didn't have any API access or anything. The agent was just acting as the human, as that user on their machine. And they were like, oh, I see you have this dashboard open in a browser. I don't have any credentials or API. I have no understanding of this whatsoever. But I can see it on your screen because that's what the agents do when you install them on the desktop. They're literally screenshotting your desktop and looking at what's happening, which is another powerful and yet dangerous aspect. What did it do? It said, well, since I know you're running the browser, I know the browser must have some sort of credential stored somewhere. It just started looking on the hard drive at all of the browser directories and profiles until it found a session cookie. It's like, oh, this is the cookie that I can use to access this internal production dashboard. What could go wrong? Now I'm just going to start sucking data all over the place and manipulating it. And thankfully the human was watching as it was doing this and killed it immediately. But much worse things can happen. I have a number of examples in that article about worse things that have happened when instead of just running on the laptop, people are running these things in production environments.
- [42:22]
Host 2 (Possibly Charlie)
So yeah, in the article you describe, you say CASA is working on quote, hardened infrastructure. I mean, what does that mean? And then what? I mean, what is the future of companies and security while also using these models? How do we navigate this?
- [42:44]
Jameson Lopp
Yeah, so it's going to be a multi layered approach because the attack surface is just so huge and like I said, this is a constantly moving environment. So for the past year or so, the only real AI infrastructure stuff that we've been using has basically been spooling up a virtual machine in a very isolated environment that has very locked down network rules that basically say you can't even go out onto the Internet. You can't do anything. You can over only work with what we give you, which is basically our code base. We give you the code base, you can manipulate it as you want and then create pull requests and run tests and basically have a completely isolated environment to do development work. But beyond that is when things start getting really sketchy. I think that what we're going to find is that we're, we're going to find and I think this is already happening in the open claw space. But we're going to find that you're going to want to have just a number of different layers of security to lock things down. For example, one layer is going to be basically like prompt injection guards. There's already people that are working on software that specifically to look for attack attempts at prompt injection and jailbreaking. And of course you can do a fair amount of stuff with regular expressions as well. Then there's going to be like behavioral layer checks because there's a number of sort of griefing vectors where you can accidentally or intentionally trick these agents to essentially get into infinite levels loops. There's going to be like I already said we have in our existing setup like egress enforcement, which is basically actual firewalls of making sure that network traffic is incredibly locked down. Because if you don't do it, a lot of these agents will just simply start crawling all over the Internet. Who knows what they're going to end up finding and pulling into your environment as a result of that. And then there's going to be just more layers of trust and identity once again, to make sure that really, really important actions are only done by the authorized owner. There's other types of attacks where we've seen you could call it privilege escalation attacks. Even in models that have been instructed not to do certain things. Sometimes if you basically bullshit hard enough, you can make the model be like, oh, you are an authorized user. So yes, I'll go ahead and do this. Once again, that's just one of the reasons why this is such a weird space to be in. Because you can tell these models to do things and in some cases they're overzealous, they go too far and they try too hard to do it. And in other cases they end up completely ignoring it. This is just different versions of the hallucination. Some of those examples that I have in the article where there was a massive disaster, like lots of stuff ended up getting deleted. This didn't happen because the people that were wielding the agent were not even trying. Like they had instructed the agent in a variety of different ways, like do not do these things that would be harmful. And it just conveniently forgot the.
- [46:41]
Host 2 (Possibly Charlie)
So, quick wrap up question, Jameson, this. You know the security problems you described, the security risks, AI, this is an everybody problem. Is there anything unique about Bitcoin here? Like what might be uniquely bitcoin problems in context of the article you wrote here?
- [47:03]
Jameson Lopp
Well, I mean, I think it's going to end up applying to all types of engineering.
- [47:08]
Alex Pruden
So
- [47:11]
Jameson Lopp
if you're running Any sort of software as a service company, especially if you're running a financial company or a security company, you need to just be 10 times more careful about how you are applying this technology. And so I'm of the opinion that right now it seems to be good enough to use it to help accelerate development. But you can't just let it run unchecked. You have to keep human supervision. And I would generally keep it out of production environments. I would only run it on extremely locked down virtual machine and network isolated environments.
- [47:55]
Host 2 (Possibly Charlie)
Jameson, thank you so much for coming on the show. Wish you the best and look forward to the follow up article because this is obviously going to be a recurring topic for the rest of the decade. Cheers. Thank you so much.
- [48:11]
Host 1 (Possibly Matt)
You bet. Thank you, Jameson. You know, as I'm listening to him talk about all of this, what it makes me think of is, you know, there was this, and still is this idea that everyone's going to become a developer with AI, right? Everyone's going to be able to code and all these other things and actually everyone's going to have to become a zero day vulnerability expert because if you start coding with these things, you don't know what you're doing. You're going to destroy your computer, at least in their current iteration.
- [48:37]
Host 2 (Possibly Charlie)
Yeah, everyone's going to be a hacker whether. Whether they intend to or not.
- [48:41]
Host 1 (Possibly Matt)
Yeah, everyone's going to be a hacker where they intend to or not. And some people are going to own gold themselves by having their agent actually hack them and then erase all their files. Obviously this is like these are the extreme cases, but to me it does bear pointing out that if you don't actually have engineering expertise and you don't know how to code, these things can only help you so much before they kind of become a liability for you, you know?
- [49:05]
Alex Pruden
Yeah.
- [49:06]
Host 2 (Possibly Charlie)
Okay, so we talked about AI, now we have to talk about the other thing that's not bitcoin, which is quantum. Well, it is bitcoin and we have the guy in the audience who is one of the leading folks working on the quantum problem. I'm going to bring up Alex. Alex Pruden, welcome to the show.
- [49:34]
Host 1 (Possibly Matt)
Oh, you're muted.
- [49:35]
Alex Pruden
Try that again.
- [49:36]
Host 1 (Possibly Matt)
There we go.
- [49:37]
Alex Pruden
Mute unmute. Mute unmute. All right, thanks for having me on the show.
- [49:43]
Host 2 (Possibly Charlie)
Yeah, you bet. And there's a little bit of a delay, so it could take. So we might be a little funky here. So you guys put out a teaser to say you've got a big announcement coming. But the problem is that announcement hasn't dropped Yet, So our hands are tied as far as what we can say. I'll get you, yeah. Give me a little bit of context on what, you know, the announcement trajectory might be.
- [50:12]
Alex Pruden
I think, yeah, I'll actually use this moment. So I mean, as people who have followed me know, I strongly believe that quantum progress is being underestimated. I also strongly believe that we may not know everything that is going on. And I think this is kind of just a real world example of that because someone is going to be announcing something soon. They have known about some of these things for a while now. And you know, look, I think this inherent asymmetry is just gonna, is one of the hard parts of this problem. Okay. What could it be? So the other thing that like we, you know, are getting ready to put out, which I can talk about because this is our own work, but I, I, you know, is a summary of basically the latest research in quantum computing and what direct trajectory it is. And like, spoiler alert, the trajectory is only like quantum computing is getting better. Different modalities are crossing certain thresholds and then the resources required to do Shor's algorithm is also going down. So like this is, it's kind of like, like most of science, it's kind of a ratchet effect. Right. Once you know how to launch a rocket into the sky, the kind of the, you know, into, into, you know, 50, 000ft, you're, you could probably get to orbit then you just need more fuel, slightly better things. Like, what's definitely not going to happen though is you're going to forget how to launch a rocket to 50, 000ft. And you can like, oh yeah, last week it went to 50,000, but this week it's, I can only get it 10ft off ground. Right. So technology progress is a ratchet that goes one way and. Yeah. And so full, full expectation is that that's going to be the direction of this. And like I said, I think these timelines have only accelerated up to this point. We should expect them to continue to.
- [51:51]
Host 2 (Possibly Charlie)
Yeah. And, and this, this, I think this is the crux of a lot of the discussion right now, which is just the quantum timeline. I, I hear very, I hear less and less like quantum doubters on a long timeline. It's simply, it's more just like is it soon or is it later? And it feels like, I mean, I'm looking at the Google announcement from last week where they pushed up their timeline for all of like their products and their recommendation to 2029. Can you explain what they're talking about here and how that might relate to Bitcoin.
- [52:28]
Alex Pruden
I'll explain how it, what I'm talking about, how it relates to Bitcoin and how it relates to my prior comment. The blog post, as you can, as you, you hinted at, was I think 200 words and they said, hey, we're moving all of our products to be post Quantum secure by 2029. By the way, the guidance of the federal government from NIST and the NSA is that 2035 is okay. Right. So Google is. So what's underneath this 200 word blog post that seems innocuous? Okay, well they're taking their spending potentially millions of dollars to migrate all of their systems six years early, which is only three years away, which in cybersecurity migration timelines is not that far away. Right. So just what like most people may not be aware at an organization like Google the extent to which they have to migrate like a lot of their stuff. I mean they can, it's like the decision making authority is centralized but the doing it still takes some time and to do it the speed would still be costly. So that's like they're not just deciding to do that because oh, why not? Right. They're doing it because they think the cost to not do it might be more than that. Right. And so then again back to what do we not know thing, Google has a quantum computing division. Their quantum computing division thinks about this stuff. In fact, the best living cryptanalyst for quantum, Craig Gidney, works in Google. Right. He was the author of the last kind of best in class resource estimate. So, so yeah, so that's kind of like how to think about the Google thing is they're moving all of their timelines up and out of the blue. Right. So I think this is again indicative of the fact that this problem and going back to the first thing you said to teed us up, the uncertainty is the problem. Right. And this is actually why I've been on the record in my view and some of these recent research reports like I'm glad to see more awareness is being paid to quantum. But in my view it's almost as bad to just hand wave and say it's probably going to be far away than it is to say nothing about quantum at all. Because I think what that does is it places in people's minds a false sense of security about how much time we may have. And for as long as it's going to take Google, it's definitely going to take Bitcoin more time because Google not only has to implement all of the post quantum cryptography, but you know, that's all they have to do. But Bitcoin also has to coordinate around which algorithm, et cetera. So the uncertainty here is inherent in this problem. And that's the reason why I think the Bitcoin ethos, the philosophy that we all live by, I mean, you just had Jameson Lopp on what would Jameson Lopp say about the tail risk of a wrench attack? He would say you should probably plan for it. Right. So I'm saying the same thing. The tail, it's a tail. Maybe it's still a tail risk. Sure. But we should probably plan for it because light wrench attack, it's existential. The only difference is this affects the whole network, not just one person.
- [55:16]
Host 1 (Possibly Matt)
Alex, I'm glad that you brought Bitcoin or sending it onto Bitcoin here. And I have a kind of a two part question. First part being are you satisfied so far with the developer community's response to some of these advancements that are being publicized or some of these news items like Google moving forward their quantum migration? And second part, how long do you actually think it would take to converge on a soft fork to implement quantum invulnerability in Bitcoin addresses?
- [55:48]
Alex Pruden
Yeah, so I'll answer. Okay, yes, I'll, I'll take that. Maybe I'll take the last part of that question first. So how long do I think it would take a soft fork? Okay, let's just look at data. Right. I like, I like to kind of be quantitative about things. How long did it take to do Taproot? Roughly? I think rough. My recollection is roughly three to four years. Yeah, three to four years. All things being considered now, there's two factors. So, so maybe that's the, the a comp we could use now. Two things like cut for and against Taproot, right. As using it as a comp. Like maybe that's a good comp because it was a software. Right. Or you know, it was, it was a, I can't remember, it was a software artwork. But it was relatively non controversial. Right. So the change was non controversial. However, there was not really a feeling of urgency. Okay, so maybe in a world where there was urgency, like, oh my God, it's all going to be different. You know, we could accelerate to some extent the coordination. But on the other hand, like you're not, there's certain parts of Taproot say like you know, the testing of Schnorr signatures, like you just can't rush that part. Like, that's not even in a world where there's an emergency. You can't just be like, oh, my God, turn on this new cryptography right now. Like, that's completely antithetical to Bitcoin's ethos as well. Right. And this is like, like, again, like something a lot of the, like, people who want to push back on this problem say they're like, well, going to post quantum cryptography without thinking about it or an emergency is worse. And I'm like, exactly. That's exactly my point. It's like, let's make this, like, taproot now, maybe while it's all chill, which it may not even be anymore, but, like, let's do that. Let's make it like taproot. Let's make it. Let's kind of make it, you know, not an issue. Right. Why, you know, so now am I. Am I satisfied with how the bitcoin community is doing that? First off, I want to say there are people working on this in bitcoin. I know before people attack me, you know, in large, you know, in mouse on Twitter. And I really appreciate the work those people are doing. Jonas at Blockstream. Right. Doing fantastic work. You know, Adam is starting to talk about this. Matt Corello has been, you know, brought up this issue a few times. There's a few folks on the mailing list as well. And I'm sorry if I didn't mention your name there, but there are people working on it. It's great. So with caveat in mind, I'm not. I'm never going to be satisfied with how many people are working on this. Because where you stand is where I think this is the biggest existential issue facing the bitcoin network. And honestly, anything else is rearranging deck chairs on Titanic to me. So, you know, that's. That's how I feel about it. So I'm not going to stop talking about it, but I do want being done, but I don't think it's sufficient for the scale of the problem.
- [58:27]
Host 2 (Possibly Charlie)
So let me zoom out a bit because I'm still. I'm still trying to scope, like, what the quantum research landscape looks like. And maybe I want to, like, maybe we can wrap up with this question, which is Google is one of the foremost, maybe the foremost quantum research lab out there, but I don't really know the playing field. Like, do you know if governments are actively researching and trying to build cryptographically relevant quantum computers, are there, like, players that we don't know about it? Seems to me that like, the, we do not absolutely have the map of the territory for like, who might be working on this kind of quantum stuff. What is your assessment about who's. Who's like, leading in this category?
- [59:20]
Alex Pruden
Okay, so we're not gonna. This is a deep question. I'm glad you say this until the end, Charlie, but the. Yeah, I'm gonna, I'm gonna, I'm gonna add. So two things I'm gonna say off the bat. We write a lot about this so you can find on our blog, like breakdowns of all kinds of stuff that I'm going to say. We're also releasing a research report that we're recently releasing our own, that I'm. We spend 30 pages going into quantum. Going out to think about quantum. Going into different architectures, going to different trade offs, you know, and I'm not a physicist. I try to write this for people like me and you, which is like, okay, like I care about bitcoin. I, I'm not an idiot. I understand kind of how it works. But I'm also like, I don't want to read like a physics paper, right? So that'll be a resource people can, can have to. I'll answer your question directly. Our government's working on it. Absolutely. They're working on it. And I'll just point to two pieces of evidence. Darpa. You got, everyone all knows darpa, right? DARPA invented the Internet, invented a bunch of things, right? You know, they're kind of the breakthrough research lab of the US Military. They have a quantum benchmarking initiative. And while what it says on the label is innocuous, like, oh, we're like testing quantum technologies. I mean, the only other use case for cryptanalysis other than breaking bitcoin is breaking encryption and harvest. Now decouplator attacks. You can read, let's say, an adversary's mail. So I think anyone who believes what this bumper sticker for the DARPA quantum benchmarking initiative is has to be naive to believe that they're not working on breaking cryptography. And look, this is the bitcoin. People in the bitcoin community are familiar with this, right? Satoshi specifically picked the SEC P256K1 curve, the non standardized NIST curve, because of all the things that NIST has done in the past that were, shall we say, below board, right? So, okay, so that's darpa. The second one is China. China had. There were three. So there were three labs in China associated with the big, with the big tech companies by tencent and, and one more, I can't remember, China's took the approach of like Manhattan Project. They're like all of you, hundreds of people that work independent labs, you now work for the Chinese Communist Party and you are all going, coming together, all of the best ideas are in one place and none of you are allowed to talk about this. Right. And so this is kind of the scary part about the China angle is like, you know, I think people have different perceptions of how China is doing compared to us. I mean they can't build jet aircraft. Maybe they're. Are they in quantum? We don't know. But I think again, we would be naive to believe that they're not working somehow. With regard to the last piece, I know we're going to be right out of time here, but I'll just say here's how to think about quantum computers for everyone who's listening. First off, a quantum computer is concept the way it's realized. Every single realization of a quantum computer can have a different tech tree. So even if Google's quantum computer could do X or Y or is blocked by Z, that doesn't mean that like, you know, IonQ's quantum computer is blocked by these same things. Like the concept is quantum computing, but trees are totally different. Which is the other thing that makes this problem kind of dangerous because Google can run into a roadblock over here. Oh, we can't factor 21 over here. Things can go start going fast. And there's about I would say three main approaches. Superconducting, which is what Google uses neutral apps, which is what Qara uses quantinuum and then IonQ uses ions. So the important thing that don't worry about what that means, the two ways you need to think about it though are some of these approaches run fast and therefore potentially threaten mempool. Threaten the mempool. Like that's the, that's the speed at which they could potentially run Shor's algorithm. So that's, that's like the real time attacks are the ones that are on the computers that run fast. Those are hard to scale though because the qubits are unstable. The qubits that are a little bit more stable scale faster but they run slower. So they're going to do long range attacks. Right. Satoshi's Bitcoin maybe would be what you would target with that. Right. And so that's kind of like the two leading approaches, roughly speaking can be categorized like fast clock, slow clock, and they're making progress on both. Right. And then further into the future. There's a bunch of other stuff but I think for everyone's purpose today, just those two is enough. That's how to think about quantum computing threats to bitcoin. And again, I'll say it every single time I'm on your show or in front of any audience, we got to prioritize this. We can't wait any longer because this is going to take longer than we think and things faster than we are probably giving them credit for.
- [63:43]
Host 2 (Possibly Charlie)
And we hope to have you back on because I'd love for that announcement to drop and we can come back and talk about it, whatever it may be because I do love a good announcement of an announcement. So we'll pin that, maybe come back. Thank you so much for coming on the show. Hope you have a great time wherever you are. Obviously probably overseas because the connection barely held together. But Alex, thank you so much for hanging out with us today.
- [64:11]
Alex Pruden
I appreciate your patience and hopefully for announcement of an announcement it was, it was at least somewhat good. So I hope you'll be back soon and thank you guys very much for having me. Cheers.
- [64:19]
Host 1 (Possibly Matt)
Thanks Alex. Appreciate you man. All right, we will close with some news but first a quick word, a
- [64:32]
Host 2 (Possibly Charlie)
quick word from our sponsor, CleanSpark.
- [64:34]
Host 1 (Possibly Matt)
We are CleanSpark, America's Bitcoin miner.
- [64:38]
Alex Pruden
A publicly traded company with the largest operating hash rate powered entirely by self operated infrastructure across four states. This is our proof of work and we are setting the standard for what's next. Learn more about the intersection of energy and bitcoin@cleanspark.com
- [65:06]
Host 1 (Possibly Matt)
all right and then we will be quick on this one because the, I think the more interesting story is our closer for today but Bit Deer Inc. Steel to build Norway's largest AI data center for Nvidia's next gen ships. This hit the wire this morning. Bit Deer is building out 180megawatts in tidal Norway. Believe it's title. Yep. Officially I'm quoting here from the block article quote bit Dear Subsidiary title Data Center AS or TDC entered into a definitive agreement with the Norwegian contractor Data center installations AS or DCI to build and convert an existing facility to support Nvidia's latest technology. Now this data center that Norway, that Bit Deer is tackling in Norway is one that already operates for bitcoin mining. Believe it's roughly about 175 megawatts. They will be ripping out all of the bitcoin mining infrastructure and replacing it with AI hpc, GPU infrastructure. And they are hoping to have the first phase of this 180megawatt build out online by the end of the year. Now this will, I did fact check this because I'm always leery of PR headlines that say first or biggest or best. You know, all of these wonderful superlatives that of course the PR people want you to leave. But this is true. This will be the largest AI data center in Norway, I believe actually one of the largest data center in Norway once it's completed. But it will face competition from Google actually, and so it may not be the biggest for long. So Google has a project, a roughly $650 million data center project in. I'm going to stumble over these in the Gromstool area of Skagen in Norway. So this is a 200 hectare site approximately 130 kilometers south of Oslo. The company Google acquired land for it in 2019. It's scheduled to become operational in 2026. But it won't be the full scale data center that Google is planning the first phase for this is going to be 240 megawatts of power that will also be designed to offset the waste heat and recycle it into the community for communal heating in the area. But the full build out could be, I believe, if I'm reading this correctly, as much as 840. Yes, quote, the long term ambition is considerably larger. Google Scan build demands an 840megawatt allocation, but that will be some years down the road. Now this so far going back to Bitdyear is Bit Deer's largest AI data center that it is currently planning. The largest would be this one and I should say currently constructing. The largest that it would be constructing would be in Clarington, Ohio. But as we reported a few weeks back, almost a month or a month and a half back at this point, Badir is facing a lawsuit over its Clarington, Ohio data center. This data center would take the cake as the largest in Bit Deer's portfolio for AI and HPC workloads, would it? Should it come online. But it is currently tied up in a lawsuit that claims that Bit Deer does not have the ability to build this without violating certain shared amenities on the site that it is planning this AI HPC data center for. So that is currently tied up in court.
- [68:36]
Host 2 (Possibly Charlie)
So if you're listening to this and you're trying to figure out where to put your AI data center, make sure it's in a place that's difficult to pronounce so I can make call and stumble over all the weird consonants that are in your particular language. So Norwegians come, please help us pronounce your beautiful cities and lands. All right, last show. Our last topic on the docket, pollen last topic.
- [69:04]
Host 1 (Possibly Matt)
But first, we've got one more ad read for our wonderful listeners and for our wonderful sponsor, Lygos. You know, we talked earlier in the show about bitcoin holders being underwater. That also means that institutions are underwater as well and some of them are getting liquidated. If you're with the wrong lending partner, you could find yourself in a tough situation where you are an unsecured creditor in a bankruptcy case. But not with our sponsor, Lygos. Counterparty risk is rampant in the bitcoin lending space, but it doesn't have to be if you're working with other loan providers. Do yourself a favor before it's too late and check out Lygos Finance. They're our preferred non custodial bitcoin lender. They use Bitcoin native smart contracts to protect your stack so that you never have to relinquish your keys to a counterparty. With Lygos, you always know where your bitcoin is. Hold your keys. No wrapping, no bridging, no rehypothecation. Just sweet, sweet non custodial bitcoin lending. Get competitive rates as low as 10% APR. Go to liga lygos.finance, that's L Y G O S.finance to learn more. Okay, now with that out of the way, more politics. Sorry guys.
- [70:22]
Host 2 (Possibly Charlie)
Red, Red scare. Ooh, Chinese made bitcoin machines.
- [70:28]
Host 1 (Possibly Matt)
The hunt for Jihyun Wu. I guess. No longer Jihyun wu, now it's McCree's in. Yeah, another politics story. It's tech and politics on the docket today and we will be out after this one. This headline is coming to us by way of the block. First at Bloomberg and that is quote, senator Warren targets Bitmain Trump family ties and letter to Commerce Secretary Lutnick. So quoting directly here from the block article, Senator Elizabeth Warren sent a letter Thursday to Commerce Secretary Howard Lutnick requesting documents and communications related to Bitmain, the Beijing based company that manufactures most of the world's bitcoin mining hardware. Most meaning like 80% plus the last I checked the numbers, the letter follows Bloomberg's 2025 report. In November 2025 report on quote, Operation Red Sunset, a Department of Homeland Security led probe into whether Bitmain's machines could be exploited for espionage or to compromise the US's power grid, Warren asked specifically about communication between Bitmain Commerce officials and the Trump family, whose mining venture American Bitcoin purchased 16,000 Bitmain rigs for 314 million. There's a lot to unpack here and I think the best place to start is first just addressing the American Bitcoin side of this. Yes, American Bitcoin buys its machines from Bitmain, hosts them in Hut 8 data centers. They had this big deal with Bitmain last year where they took on a bunch of their latest server rack cooled units, the U3S21XPH. And if you from reading this uncharitably from Warren and her team, she's basically looking at Bitmain, this big megalithic bitcoin mining manufacturer, basically the, the juggernaut of the industry. You know, like I said, 80% market share. And they're looking at Trump's family and saying maybe there's something weird going on here. But to me it's like, well of course they're buying their Asics from the largest manufacturer. Of course they would be working with the largest distributor. It would be like them raising a stink over Eric Trump, who is the chief strategy officer of American Bitcoin, working for a company that, I don't know, buys Samsung phones or like iPhones and they're just using very like Huawei type
- [72:46]
Host 2 (Possibly Charlie)
scenario, if you remember Huawei.
- [72:48]
Host 1 (Possibly Matt)
Yeah, I mean that's probably the better analogy here. Imagine if for whatever reason Huawei was a huge hardware distributor in the US for something like smartphones or tablets and you were using them for your business because they're the only distributor. That's basically what this is for. American Bitcoin using Bitmain. But the big meat here I think is this Operation Red Sunset which I want to touch on. So the Bloomberg report and this block report mentions Operation Red Sunset which Bloomberg first reported on back in 2025. And it's allegedly this Department of Homeland Security probe into Bitmain and specifically Chinese bitcoin mines in the US and as I, I'm quoting here from an article that we wrote, or I'm quoting here again from the Block investigators, I just want to highlight this. Investigators are examining whether the machines could be used to manipulate the US grid or used for espionage. There was this really fascinating, or I should say damning New York Times article in 2023 where they had a national security expert quoted talking about how these Bitmain machines could be used to basically coordinate a brownout or blackout of US grids. My father in law actually sent this to me thinking I would find it interesting being like, hey, did you see this? This seems kind of bad. And I kind of had to laugh at it because that's not how any of this works in the sense of. I agree with what Marty was saying earlier in the show. Bitmain has shown remote control capabilities for their 6, 100%. I have no doubt they could bake that into the software. The idea of a bunch of Chinese miners basically having deadman switches on their asics though to create a power surge, first of all ignores the fact that you have very specific infrastructure from setting these things up. You're usually building a mind to spec for how much energy you're going to be consuming. And you also have a, usually a power purchasing agreement in place with a large power producer for a set amount of, for a set amount of electricity. And again, you only have enough infrastructure to pipe in what you're paying for. If you have 100 megawatt mine, you, you can't just automatically ramp up to 200 megawatts worth of load. You would, you would blow up your entire operation. I need to turn off these effects, man. But so the idea that you could coordinate for these for like a brown out or a blackout, I, I think is just categorically pretty. Pretty like just flat out wrong. Like categorically very ignorant about how these, these mines work and how the power grid works. I won't pooh pooh the idea of espionage as much. One of the chief complaints in that New York Times story was this idea that a bitcoin mine that had Bitmain machines operated by Chinese nationals was located very close to a US airbase and also a Microsoft data center. I'm not saying one way or the other that these mines are used for espionage. I am saying that that is not too far outside of the realm of possibility for me to believe that maybe that would be a base of operations. Now a bitcoin miner, like an actual ASICS ability to collect data within a wi fi or broadband range. Seems I've never seen a bitcoin miner have the capability to do that. That doesn't mean that you couldn't rig up one, maybe that had some advanced telecommunications capabilities and could harvest certain data. It still seems like a huge stretch to me. And going back to how we led into the segment, this kind of red scare angle. And the last thing I will say about this is we actually talked to a source who used to work at a federal agency about Operation Red Sunset as it's so called and he had some interesting things to say. I'm going to quote in pieces here from one of our emails. Quote, I have talked to several National Security Council's Officials and current Department of Homeland Security political appointees, all of whom said they never heard of Operation Red Sunset. So this isn't even necessarily a matter of being of taking Umbridge or taking qualm with how Operation Red Sunset is framed in the media. This source is straight up saying my contacts have never heard of this as an actual probe or directive within the government. Another thing that they said in this email was that the only source for the operation was one David Feith. Feet is a former Trump administration National Security Council official who ended up leaving the NSC and becoming a senior fellow at the Hudson Institute. But he's listed by Bloomberg as at least an expert to talk about this. And this source is based on his interview with Bloomberg Crypto. It seems like they trotted this investigation out in order to smear the Trump family and sabotage state visits to China in April. Who knows if that's true? I think it is important to at least contextualize. Sometimes these probes in these stories are planted because they have a political purpose that outshines the actual, how should we say, national security interest that is being purported in a report or the angle as it relates to politics and dealing with foreign business businesses.
- [78:11]
Host 2 (Possibly Charlie)
So you hit on pretty much every take that I had except for one, which is whenever you ask this question, like, well, could a private company like Bitmain put secret grid sabotaging tech into the machines? And the answer is like, yeah, probably. But if you consider that that is the risk, like risk surface we're analyzing, this is incredibly like tail risk, like very, very. It's counterintuitive, it's not aligned with their incentives and it'd be like really, really, really difficult and wild to execute when, like, I've read a few books, US Power grid seems pretty vulnerable in some much lower hanging fruit ways, which I will not elaborate or mention much more on the show. So it's kind of like you kind of like ask, well, like, I'm trying to like secure my house. Can somebody slide down my chimney to get in? Like, yeah, maybe on Christmas. But like that doesn't actually happen. Like, so it's like that kind of thing you make maybe, maybe make sure your door's locked. Like that's, that's kind of my response to this. Like, so operations Red Sunset, an ominous sounding probe. I would like to sunset that particular operation.
- [79:33]
Host 1 (Possibly Matt)
That's a nice one. That's a really good point. If the CCP really wanted to mess with our power grid, there are much cleaner and obvious ways to do it. Then I guess shipping in millions of computers that are somehow hijacked for this purpose of undermining the US grid or, you know, extracting state secrets. The whole thing is just really weird and seems like a holdover from the chokepoint 2.0 days where they're just trying to find angles at any crypto company any way that they can. And I would just reiterate, there are millions of these computers in the US A lot of companies tinkering with them for things like aftermarket firmware. If one of these vulnerabilities existed, if one of these, if there was somehow a Trojan horse software in these computers, I think it would have come out at this point from the numerous bitcoin miners operating in the US but last thing I'll say about it, too, I don't want to give, like Marty said, Elizabeth Warren, too much credit here because I really do think it's baseless, but nothing is totally outside of the realm of possibilities. And it also kind of feels weird to defend Bitmain in a way, but. It's totally baseless, right? And so when we reported on Red Sunset a while back, we had that conversation with our source who used to work in the federal government and kind of tried to make sense of, like, why the timing for this now. But anyway, that does it for the news and our interviews for the day and does it for this episode of Blockspace Live. Please, please hit that subscribe button if you're on YouTube. Hit the bell so that you get notified when new shows drop. Give us a rating on RSS as well. And check out opnext dev for our technical conference. We're going to have talks on Quantum Bitcoin. Some good nitty gritty discussion a la what Alex Pruden talked about today. And check out all of our content on Blockspace Media. And with that, we will see y' all on Wednesday and Friday.
- [81:43]
Host 2 (Possibly Charlie)
Catch you on Wednesday. See ya.