Unchained - Why ASICs May Be Better Than GPUs Even if They Tend Toward Mining Centralization - Ep.67

Episode Date: June 19, 2018

David Vorick, CEO of Nebulous, which operates Siacoin and a new subsidiary, Obelisk, describes what he's discovered in his exploration of the mining industry, including why he's decided that specialized chips like ASICs are preferable to generalized chips like GPUs, even if they tend to bring out mining centralization. He also describes Obelisk's new service, Launchpad,which secures exclusive deals to create a custom proof-of-work algorithm for a coin and also the mining equipment for it, in exchange for being the only miner on the market at launch or for a period. He explains why he thinks proof of work is the only really viable type of consensus algorithm and doesn't even see any other potentially workable ones on the horizon. Plus, he has some sharp criticisms of Bitmain and claims that several groups were secretly mining Monero, though he can't reveal how he knows.   David Vorick: https://twitter.com/DavidVorick https://medium.com/@davidvorick Siacoin: http://sia.tech Blog.sia.tech Obelisk: https://obelisk.tech Blog.obelisk.tech David on why Siacoin chose ASICs: https://medium.com/obelisk-blog/choosing-asics-for-sia-4b11695df051 David's long post about the state of cryptocurrency mining: https://blog.sia.tech/the-state-of-cryptocurrency-mining-538004a37f9b Announcement about Obelisk: https://medium.com/obelisk-blog/introducing-obelisk-launchpad-b78756eaa74c Zooko's recap of his conversation with Jihan Wu, CEO of Bitmain: https://forum.z.cash/t/so-i-had-a-videochat-with-jihan-wu/29379 Bitmain's tweets being transparent (account currently disabled): https://twitter.com/BITMAINtech/status/1001376036985028608 Bitmain's blog post about being transparent: https://blog.bitmain.com/en/antiminer-z9-mini-shipments-experiment-radical-transparency/ Thank you to our sponsors! Preciate: https://preciate.org/recognize/ https://www.blockchainwarehouse.com Learn more about your ad choices. Visit megaphone.fm/adchoices

Hi, everyone. Welcome to Unchained, your no-hyp resource for all things crypto. I'm your host, Laura Shin. If you've been enjoying Unchained, pop into iTunes to give us a top rating or review. That helps other listeners find the show. Blockchain Warehouse is an international blockchain accelerator, offering a wide range of token sale advisory services to promising blockchain-based ventures. With the leading advisor network, BCW is at the forefront of building landscape-changing blockchain companies. and hosting successful token sales with more than $20 million raised so far. Let's raise the bar together with Appreciate, launching this summer. As a sponsor of Unchained, Appreciate has recognized amazing people because we believe in the strength of recognition and relationships, and the strength of community.

Who will be recognized today? Stay tuned. My guest today is David Vorak, CEO of Cyacoyne and now Obelisk. Welcome, David. Hi, great to be here. I want to spend the majority of the time talking about mining and your new ventures, but let's first get your bio and talk about your first company. How did you get into Bitcoin? I was introduced to Bitcoin by a friend of mine in college back in 2011. I remember the price being at about a dollar and a half. And unfortunately, I did not buy, nor did I have much money to my name. But I was super fascinated by the concept and just got sucked. into it. And then I sort of fell into the trap where I'm like, oh, the system looks so bad and looks like so easy to improve. So I actually spent much of like my first two or three years

trying to come up with ways to do Bitcoin better and sort of eventually like conceding defeat and realizing that it's actually a very well done system and that all the inefficiencies are actually like necessary to how it works. And then I went on from there to make Sciacoyne, which is a decentralized cloud storage platform. The idea was that we could take a lot of the same things that Bitcoin had done for money and do it for the cloud. So we wanted to make a way to put data onto the cloud without having to give up control to say like a third-party service. We wanted you to be able to store your data on the cloud and also be fully in control of what's happening to that data. And when did you have that idea? So that started. That started

like late 2013. We incorporated nebulous, the company behind SIA, in 2014. We basically worked on it feverishly for about a year, and we released the first version, the sort of alpha release of the SIA network in 2015. And how did you come up with that idea? And also now that we have these competitors like FileCoyne and Storage, how do you differentiate from those?

Yeah, I think that I was just really interested in data and data sharing and file sharing. And I had a lot of files in my own that I wanted to store. And it felt suboptimal or like uncomfortable to use something like Dropbox or Amazon or Google Drive because they can see all your data. And it just didn't seem like the right way to do things. And there didn't seem to be a good alternative. So I thought like why not, you know, why not make the alternative? of, I told myself it doesn't seem that hard.

In retrospect, it is very difficult. But yeah, now that we have competition, I think that our key distinguisher has always been our determination to release things and to, I guess, two things. The first is that we've been completely uncompromising on the decentralization. So we've never had a hybrid, you know, partially centralized, partially centralized, partially decentralized platform. It's always been a fully decentralized platform. And that's been super important to us because we didn't want to depend on,

you know, some feature that would be difficult to replace or upgrade later. And then the other thing is that we've been really determined always to push, to push features out early and start getting user feedback early and then, you know, improve from there. And so we were the first decentralized cloud storage platform to launch. I think today we're the only fully launched, fully decentralized cloud storage platform. That's been the case for three years. And it's just been a like a journey of continuously improving it, adding features that people view is critical and trying to make it, you know, a more full platform.

And you kind of maybe alluded to this earlier, but I was looking at the stats on the Cyacoyne website and saw that you don't have quite 800 storage providers yet. and there are 175 terabytes of use storage. That didn't seem like a big number to me. Why hasn't Cyacoin seen more adoption yet? Yeah, that's a good question. So I think the biggest reason is that we don't support full backups. And so while you can put your data in the cloud without losing control, today the uploading machine is a single point of failure.

And so if the machine that uploaded the data fails, then you lose all your data and there's no easy way to recover it. But that's what we are working on through the summer. We've already begun work and we have like an eight-ish week roadmap development plan to remove that. When that happens, we think we'll see more adoption. And then sort of right alongside it, we're developing a file sharing feature, which we also think will really stimulate adoption. Last year you announced you would be manufacturing mining equipment

for SIA through a new company called Avalisk. Why did you decide to start manufacturing your own mining equipment? Yeah. So the biggest reason was that we saw Bitmain had cornered the market for Bitcoin, cornered the market for light coin and dash. And they seemed to be on an unstoppable rampage. And we didn't want the same thing to happen to SIA. We wanted there to be multiple manufacturers that were competitive. And most of all, we wanted to make sure that nobody in particular had cornered the market. So at that point, we were super naive to the hardware industry. It was our first time doing anything like that. And we just sort of took the leap and went for it. We certainly did an enormous amount of research. We talked to many companies who had done it before,

including like K&C, Butterfly Labs, and Spondulis to help get guidance and understand what's necessary to put out a hardware product. But the big goal is really to learn more about the mining space and to make sure that Bitmain wasn't the only player in the SIA ASIC world. And at that point, what kind of equipment was being used to mine Sia coin? At that point, everything was completely GPU mined. And so we figured, you know, if we can be first to market, any ASIC is going to be a big deal. And it's going to be, you know, minimally 100x better than GPUs. We, from the beginning, designed SIA to be an ASIC-friendly network. We've always believed that A6 are inevitable. And we wanted to make sure that Cycoins transitioned to A6

was smooth and friendly and open and easy for many manufacturers to get involved. Yeah. For listeners who don't know the difference between A6 and GPO's, can you explain that and then also explain why you chose A6 and also then why the community gave you some heat for that decision? Yeah. So a GPU is a general purpose, computational, machine. Most people who have GPUs use them for video games, at least before cryptocurrency. And so what it means is that, you know, every cryptocurrency or most cryptocurrencies have different proof of work algorithms. And so you have a GPU can typically do any of those different algorithms, but it's inefficient because that flexibility to switch between algorithms

just means that it, you know, it has some extra features that it doesn't need. In ASIC, is highly specific. And so all it can do is this one thing. And so in the case of the Sion network, all it can do is Blake 2B, which is our proof of work function. And so if we even make a tiny tweak to that proof of work function, the ASIC is completely broken and completely useless. Well, a GPU, you just, you know, you can tweak the GPU,

and the GPU can stay on the network. But when you make something that's that incredibly focused on one task, it's enormously more efficient. And in the case of Blake 2B, which is our proof of work function, it's between 250 and a thousand times as efficient. So, you know, $1,000 of ASIC hardware can outperform, say, you know, $500,000 of GPU hardware on the SIA network. And so the reason that we chose, so some algorithms, some coin developers and coin communities prefer to choose algorithms that are difficult to make ASICs for. they want to be GPU friendly. And the reason is because anyone can go to the store and buy a GPU.

You know, every computer shop is going to have GPUs that they can sell. It's very easy for someone to pick one up. In ASIC, however, is made by, you know, a single manufacturer or a small number of manufacturers. They don't, you know, they don't have shops on every street. You have to order it online from a specialty online store. And a lot of times these manufacturers are choosy with who they sell to or they put limits on it for the average person. but they may go and sell 10,000 to their friends. And so this creates essentialization pressure and really the world of A6 doesn't match the ideal vision for decentralized cryptocurrency mining.

But the unfortunate reality and the reason that we chose A6 for SIA is that A6 are inevitable. For any algorithm that you pick, someone is going to be able to create highly specialized hardware that's dedicated to that algorithm, and it's going to be substantially faster than a GPU. So we believe that there's no way to escape ASIC manufacturers in a proof-of-work coin and in a proof-of-work ecosystem. And so we wanted to pick an algorithm that's very easy to develop ASICS4 that had a very low barrier to entry. That way there's a better chance that we have multiple manufacturers as opposed to just one.

And even at the GPU level, you know, we see there are really only two major GPU developers in the world, or GPU companies in the world that's AMD and Nvidia. And that's because GPUs are enormously complex and highly specialized. And so there are only two companies that are at the scale to be able to do so competitively versus, you know, like Bitcoin Asix, you have, you know, Silicane and Asic Minor and BitMane and Binfury. And so you have a lot more manufacturers. And while there is one who's clearly ahead, at least they're like five or six total versus like the GPU market where there are only two total. Interesting. And so I earlier did mention also that there was kind of a little bit of an uproar about this decision.

And I also wanted to note that at the beginning when I asked you about what distinguished Zyakloin from your competitors, you did say your commitment to decentralization. However, I believe some of your users objected to this choice of ASICs because choosing ASICs does lend itself to some mining centralization. Is that correct? Why did you make that trade-off? Yep. We saw it as a long-term decentralization choice versus a short-term decentralization choice. Like I said, we think that no matter what algorithm we pick and no matter what strategy we pick, they're going to be specialized miners. And so the best thing that we can do in the long term is pick an algorithm that enables as many manufacturers as possible.

Now, I think Monero has taken an interesting alternative approach, which is that they believe more or less the same thing. They see A6 as inevitable, but they are trying to bide time. And they think that they can use a bunch of temporary means to thwart A6 and try and preserve short-term decentralization before eventually, embracing ASICs in the long run. And so we don't like that approach. We think it's too heavy-handed of governance. However, it is an example of a coin that believes the same things we do, taking a different approach to trying to decentralize their mining.

Some coins also try to make their hashing algorithms ASIC resistant, but you are skeptical of this as a strategy. Why? Because having worked closely with hardware developers, we've learned that the number of tricks hardware developers have available to them is, you know, far more, they're far more rich and far more advanced than anybody in the software world or most people in the cryptocurrency world realize. And so most ASIC-resistant algorithms have been made by software developers who are trying to, I guess, kind of fool or, or, uh, stumble hardware developers, but the software developers don't know the hardware world very well. They don't understand what techniques hardware developers have, and so they don't realize that most of these ASIC-resistant algorithms actually have sometimes trivial workarounds or

simple hardware implementations that are going to be extremely, extremely outperform what a GPU can do. And we actually saw this with Equahash. So Zcash was originally designed to be, acyc resistant, they used the Equahash algorithm. And while it worked for a while, we saw Bitmain come out with a miner that I think is 10x more energy efficient than GPUs and something like, if I remember correctly, something like 50 times as fast. And then Inosilicon, just a month later, released a minor that was even 5x more energy efficient than the Bitmain miner and was even faster than the Bitmain miner.

And so what's happened is this sort of prophecy that you can't truly ever be ASIC resistant, someone will always get there, played out very well in the Zcash case. Now, we've also seen in the past other attempts to be ASIC resistant, light coin was originally meant to be ASIC resistant. They used some tricks that they thought would make it difficult for hardware developers to make ASICs, but of course, that didn't work. Like coin has A6 today and has had them for a long time. Dash is another example of a coin that was originally ASIC resistant, but now has, is now an ASIC mined coin because hardware developers figured out how to make specialized hardware

for the dash mining algorithm. And so this is a trend that if you talk to hardware developers, they basically confirm, you know, no matter what you do, there's always, some way, some trade-off we can make, that when we make specific hardware for an algorithm, it will be substantially faster. And so the hardware devs that I've talked to and that I've worked with have said, they don't believe that ASIC resistance is possible,

just based on the nature of the work. They don't see how you could ever make an algorithm that is equally, you know, equally performant on a general purpose hardware, like a GPU or a CPU, as it is on an ASIC. You did write, however, that ETH hash, which I guess is the Ethereum-Hashing algorithm, was the most ASIC-resistant algorithm you've seen.

How did they do that? Yeah. So the tricky thing about F-Hash from the hardware development perspective is that it's basically a bunch of random reads to memory. And as it were, memory is actually already pretty optimized around doing low latency, random reads. So the Equahash algorithm ended up being pretty close to what specialized hardware, in this case, DRAM, is already doing, and DRAM is widely available.

And so that made it more difficult to break away and make specialized hardware. But as we've seen from the Bitmain miner, you can do it. You can make specialized hardware for attach. and it will outperform GPUs. There's another algorithm that got released recently called PraguePOW. It's still under development, but they basically took a new approach to making an ASIC-resistant algorithm. So these were hardware people who took a GPU,

and they basically disassembled all the parts of it, and they made an algorithm that requires you to use as many parts of the GPU as possible. And so these hardware doves still openly admit that there are improvements, that an ASIC could make over a GPU on this algorithm. They said, you know, we can't do it perfectly. But this is going to be the absolute best attempt that you can get. And we think it's going to be close enough that it's not going to matter, you know, within like a factor of two or a factor of five, as opposed to being within a factor of

a thousand. I want to go back to the blog post that you wrote announcing obelisk. You mentioned that you discovered that mining centralization in Ethereum was even worse than it was in Bitcoin. What is you uncovering there? And do you know if that situation persists today? So I don't know if that's the exact wording. But mining centralization in GPU world, basically just like with A6, you have these large GPU farms that set up. They have access to hardware manufacturers. They can buy GPUs at rates much cheaper than a consumer can afford. And from what we can tell, the vast majority of Ethereum mining is done on these giant farms, not at users' homes, right?

And so it's still not decentralized. And so even if you get to go with general purpose hardware instead of ASICs, you can't defeat the economies of scale that just govern the mining industry, people who run bigger farms and run bigger operations and have better connections to the hardware world. are going to get lower prices and are going to have, you know, more efficient mining. And they're going to be able to push the difficulty up to a point that it doesn't make sense for consumers or home users to mine these algorithms. And so we, I don't know, I don't know if I could say that it's worse than Bitcoin. But I can definitely say that it's, you know, it's the same progression that we saw in Bitcoin. And then even beyond that, we don't, I don't know if we've seen it yet.

But if someone like Invidia or AMD were to take their own hardware and start mining on Ethereum, they would be able to do so at cost. And the margins on GPUs are very high, especially compared to like the Bitcoin ASIC industry, where most ASIC manufacturers are selling the miners at, as far as hardware is concerned, relatively low margins. You know, an Nvidia GPU or an AMD GPU is held at high margins. If Nvidia were to decide to mine, they could get those GPUs for much cheaper than anyone can, and they would have a substantial advantage over the entire rest of the ecosystem in mining.

And so you would see heavy centralization around the manufacturer that is way we already have. The thing is, I don't think at this point GPU mined cryptocurrencies are at the point where they've really gotten Nvidia's attention. It's still a relatively small part of Nvidia's revenue. And so they, you know, they aren't making moves that bold. But I think that if it were to continue to grow and prove to be a stable source of revenue, we would see an AMD or an Intel or an Nvidia move in and have a substantial advantage over everyone else. That seems so counterintuitive to me that the margins are higher on the GPUs than on the specialized.

Betmoot activated. The scorebed app here with trusted stats and real-time sports news. Yeah, hey, who should I take in the Boston game? Well, statistically speaking. Nah, no more statistically speaking. I want hot takes. I want knee-jerk reactions. That's not really what I do. Is that because you don't have any knees? The score bet.

Trusted sports content, seamless sports betting. Download today. 19 plus, Ontario only. If you have questions or concerns about your gambling or the gambling of someone close to you, please go to conicsonterio.ca. Local news is in decline across Canada, and this is bad news for all of us.

With less local news, noise, rumors, and misinformation fill the void. And it gets harder to separate truth from fiction. That's why CBC News is putting more journal analysts in more places across Canada, reporting on the ground from where you live, telling the stories that matter to all of us, because local news is big news. Choose news, not noise. CBC News. Hardware. Why is that? That's because it is very difficult to make a competitive or a highly efficient GPU. You're talking like billions of dollars of development effort and research. And because it's so difficult to make a competitive GPU, it's basically, you know,

invidia and AMD in a price war or dueling each other. And so, you know, they can distinguish on features instead of on price. And so there are ways that they can continue to sell chips without undercutting each other on price heavily. Now, in something like Bitcoin mining with A6, it takes a lot less starting effort. You know, we estimate on the order of if you're going with T's, TSM on the order of $25 to $30 million to start an ASIC company. And while you're not going to be at the efficiencies of BitMain,

you're going to be much closer. Your shot 256 ASIC is going to be much closer to what BitMain can do than what a GPU startup is going to be to what Nvidia can do. And so because that gap is narrower, if BitMain prices their hardware too high, startups like Obelisk can come in and make hardware and sort of, you know, chew with their margins. And so because the barrier to entry to the industry is a lot lower, we see that the margins are also like just necessarily a lot lower.

Interesting. I also want to ask you about what happened this past winter when Bitmain announced the Antminor A3, which was a Siacoin miner, and that caused some consternation in your community and talk of a fork that would render the A3s unable to mine Cyacorn efficiently, you ultimately decided against that kind of fork, and you even wrote this blog post that announced your decision, and it ended by saying that you welcomed Vitmain's customers. So why did you go that route? Yeah, so I think that initially we saw the A3 as basically an attack on the network.

We had a big community effort to produce A3. for the network and sort of protect the network as a community. And so many, a large fraction of the most involved people in this high ecosystem were heavily financially invested into Obelisk. And so to basically when Bitmain came to market and threatened that, it basically soured and disrupted and turned off a large portion of the SIA ecosystem. And we did lose a lot of users to the Bitmain announcement. A lot of people have been very unhappy ever since.

And so I think if we had premeditated, if we had planned ahead a little bit more and said, you know, what happens if Bitmain shows up in January, we would have, you know, as a community, been coherently focused around the idea of forking to get rid of Bitmain and to protect Obelisk since so much of the community was so invested into Oblisk. Unfortunately, we didn't have this planning in foresight. And so when Bitmain came to market, there was this big confusion about whether, you know, what should we do? What's the right thing to do? Is it greedy to fork and protect Obelisk?

Is it, you know, if we do fork and put Obelisk, you know, give Obelisk a software-based moat or a fork-based, you know, exclusivity, does that mean that SIA is centralized under Nebula? and I think there were plenty of reasons to argue that is absolutely the right reason, the right thing to do. And I think, you know, that's, if we had been together as a community and making that decision, I think it was the right decision to make. But we had several prominent community members come forward and say, if you fork, you know, we will leave and we'll lose faith in sign, we'll never come back. And that spoke, you know, volumes to us.

And we realized that we were stuck with this, you know, if, if you don't fork, you're going to lose a lot of community members who are angry and upset that they lost money. If you do fork, sort of the ethical polish that SIA has always had is threatened. And people may think that SIA really is just a centralized coin and that the decentralization motto and everything is just for show. And it's just marketing. It's not how SIA really is.

And so we decided in the moment that because the community was split and because, you know, it was difficult to fight this argument of greed and difficult to, you know, dismayed the accusations or, you know, disprove the accusations that we were acting purely out of greed for ourselves. We decided not to go through with it. We decided we would rather deal with Bitmain and have the A3s on the market and just sort of embrace the situation. We're going to discuss more about Obelisk and also its new service launch pad, but first I'd like to take a quick break to tell you about our fabulous sponsors. Blockchain Warehouse is an international blockchain accelerator, offering a wide range of token sale advisory services to promising blockchain-based ventures. With access to heavyweight technology leaders, the accelerator is heavily involved in crafting the blockchain technology, token sale, and regulatory landscape. Blockchain warehouse will launch the first ever Crypto Shark Tank in June. This week's episode features Block 66, an innovative blockchain mortgage platform that streamlines and organizes the facilitation of residential and commercial financing. Block 66 reduces the amount of time, persons involved, and money that it takes to complete a full mortgage cycle.

Previously, real estate transactions on the blockchain required 100% funding up front, but that's a thing of the past with Block 66. Find out more at www.com.66.io. Now it's time to recognize someone sponsored by Appreciate. If you've been wondering when a truly new consumer app would be launched on a blockchain-based protocol, at scale, your weight is over. Appreciate is building the future you deserve, a trustworthy and transparent one by design, powered by the Goodwill Composite Protocol. Today, Appreciate recognizes Manuel Stegars, the acclaimed documentary filmmaker. In his award-winning short film, The Blockchain and Us, Manuel, Manuel brought understanding of the impacts of blockchain by showcasing the enthusiasm of

crypto-pioners. Way to lead the charge, Manuel. Go toprosate.org to learn more about the community and recognize someone. Let's raise the bar, together with Appreciate, launching this summer. I'm speaking with David Vorick of Cyacoyne and Obelisk. It sounds like this controversy that happened in the winter when Bitmain announced its A3 miners might have been. have informed your new service launch pad and the way it's designed. And I'm going to describe it briefly, but you can feel free to correct me or explain it further. It seems like you guys design

both the custom proof-of-work algorithm as well as the ASIC hardware that works in that algorithm. Is that correct? And was it influenced by what happened with the A3s? Yes. So you're correct on all accounts. What happened with the A-3s was that, you know, you had this community feel, this decentralized coin, this thing that people were very proud of, then Bitmain just kind of came in and soured everything by being, you know, the dominant manufacturer. We've seen this, you know, this happened to Dash, it happened with light coin. We even seen it repeated, you know, I know Silicans done it to Zcash and to Decred. And so, you know, we, we think that there's a, today you're basically, proof of work coins are basically

stuck with a situation where whether or not you embrace A6, at some point, A6 are going to be on your network. And then basically, whoever gets there first and whoever's first to market with an ASIC gets to call the shots for a while and gets, you know, gets to control the hash rate, gets all the coin issuance. If they want can perform certain types of attacks. And so the transition, as most coins have seen, has been very brutal to go from GPUs to A6. And so with Launchpad, if we are developing a algorithm behind the scenes and ASIC hardware concurrently, we have a strong guarantee because of the secrecy that Obelisk is going to be first to market. And then, of course, beyond just, you know, Obelisk being first to market,

we have legal agreements with the developers and sort of agreements and contracts with the community of what we're going to do with this first to market power, especially because it's been granted to us sort of exclusively, we can make sure that when the transition from GPUs to ASICs happens or when the coin launches, we can just launch the coin with ASICs out of the gate. We can make sure that those ASICs are owned by the community and those ASICs are decentralized as opposed to being centralized

under a single manufacturer. And then other things that we can do is we can make these, the ASICs or the algorithm ASIC friendly, we can make it easy for other manufacturers to get started. then we can even open source the chip designs so that all really another manufacturer has to do is take the chip designs and do another tape out. Wait, I actually just want to back up because you just said that you can help ensure that the coin is not centralized under a single manufacturer,

but it sounds like that's exactly what will happen. Yes, so the key difference there between what, say, Obelisk is going to do and what, say, Inosilicine or Bitnain might do, is that Obelisk is under a legal contract to distribute the hardware a certain way and will be very transparent about how the hardware is being owned and how it's being distributed. And then as a second step, Obelisk will also make sure that when other manufacturers want to get started, they have, you know, a lot of guides to lower the barriers

and make it easy for other manufacturers to come and compete with Obelisk. So yes, it is still a centralized launch, but it is a much smoother and more community-oriented launch. And it is a launch that is geared around getting other ASIC manufacturers ramped up in the future, as opposed to being a launch centered around owning the market and making sure there's only one ASIC manufacturer ever on that coin. Is it a risk to have one company do, create both a proof of work algorithm as well as the ASIC hardware? Like, could a rogue employee game that in some fashion? And in general, what are the risks that people would run in trusting one company to do both things? Yeah.

So I would argue that the risks are, at the very worst, they're equal to not using the launch pad service. Because if you don't have, if you don't have the launch pad service, you basically, have this whoever gets first to market, whoever makes the best chip, whoever plays most aggressively, is the person who owns the coin. And so you have this completely, like entirely rogue element. So either you go in a situation where you're going to get a first to market and have no control at all, or you have this trusted situation where you are trusting a manufacturer. At the worst case, they could go rogue, which is what you have without it. And in the best case, this entity with a reputation can help you get off to a smooth start and help you get off to a

coach and start. And also with Obelisk, you're focusing on proof of work coins. What do you think of other non-proof of work consensus algorithms like proof of stake or hybrid proof of work, proof of stake models, or even something newer like threshold relay and affinity? Yeah. So we both at SIA and at Obelisk believe pretty strong. that proof of work is really the only really strong method of achieving decentralization and achieving decentralized consensus specifically. And so hybrid models like, say, the decred model, we are in support of, but that's because the hybrid model works because it has this like fundamental tether using proof of work

that gets it off the ground. So other models like proof of stake, and I'm actually not familiar with the DFINI, method you brought up with. But generally speaking, when getting scrutinized or when going under peer review, these non-proof and fork models have been shown to have a lot of issues. I would be interested to know what you think of it when you learn more about it because they use a random number generator, apparently, to decide who will add the next block. And it's guaranteed to be truly random, not like, because I said to, um, the founder, I was like, wait, that seems like something we've had for forever. And he was like,

oh, no, no, no, no, for, you know, like anything where you think you're getting random numbers, like a lottery or whatever, there's an algorithm. It's not truly random. So in the end, with threshold relay, what you end up with is something similar to proof of work in that the amount of mining equipment that you have on the network or the percentage you have, I guess you would say, of the total network, that, you know, over time, you would end up with a similar proportional amount of coins, but, you know, it's not, it's not as electricity intensive. There are so many coins that are moving away from proof of work to things like proof of stake. So do you worry about a broader industry shift and how that might affect the business

prospects for Obelisk? Having been on the forefront of consensus research for a while, especially starting 2013. I feel fairly confident that it's not a big risk. And I feel confident that proof of work is going to be around for a long time. I think Ethereum, for example, has experienced a lot more trouble getting proof of stake off the ground than they were expecting initially. They had to set back their power time bomb or whatever. And I think that these alternate methods that have come through, And in 2013, sort of the Bitcoin core community got a core understanding of the things that really

don't work in these different models. And nothing that we've seen in the past four years or five years has challenged or has made it look like there's something different than these core issues. And so I don't think that there's anything on the near horizon and probably not even the distant horizon, which would supplant proof of work or offers similar security properties to prove of work. I think all alternative systems make different assumptions, assumptions that I would challenge as being less secure or weaker than what proof of work can do. Let's talk about Bitmain. This is, I guess, the largest manufacturer of cryptocurrency mining

equipment in the world, and it runs some of the biggest mining pools. And last year, it had $2.5 billion in revenue, I believe. How do you compete against Bitmain? Yeah, so that's a great question. And I think that it's something that's not easy. As I pointed out in my blog post, there's these economies of scale where if you're throwing, you know, a billion dollars at a hardware problem versus somewhere, versus someone throwing, say, even $100 million at a hardware problem, the billion dollars is going to have a lot more room for optimization and a lot more room for economies of scale. And they're going to be able to produce substantially more than 10 times the hardware

for 10 times the price. And that's just how the hardware industry works. And that's just how hardware efficiencies works. So I think that you're certainly right to call out Bitmain as a very challenging opponent. And I think one of the key advantages that we may have is right now the industry, the silicon industry, is supply shocked because the seven nanometers have been late. And so companies like Apple and

Nvidia and AMD are buying up all the seven nanometer supply. It's Bitmain's big, but they're not as big as Apple. And so if Bitmain and Apple get into a bidding war, Apple wins. And so at least in the near term, I think that that provides one gap. But another gap that really helps is that the Shot 256 algorithm or the Bitcoin algorithm specifically is so simple that all the optimizations that Bitmain figures out, a lot of them end up being, you know, easy to replicate from a lower scale. And so, so I think that it will be an uphill battle for sure. But I also think

that there are, you know, I think it's a battle that can be won, that we can see multiple players in the space, you know, seven, eight, even, you know, 10 plus players in the space who make hardware that is approaching what Bitmain is capable of and is able to remain competitive. You've written a few times that you think Bitmain is a bad actor. Describe what happened when you try to manufacture your equipment in China. Yeah. So this, I want to be very clear, is something we don't have any proof on. And so there could be any number of reasons that this happened.

And so this is not a accusation at Bitmain. However, we were warned before we. started manufacturing not to do anything in China because Bitmain would mess with our supply chain. This is a warning that I received from multiple different advisors telling us to stay out of China. And a lot of our manufacturers are a lot of our engineers who helped us pull, you know, bring the chip together, bring the unit together and the mining rig together, thought this seemed kind of silly and so opted to go to China anyway. So originally we were using two services in China.

One was SMIC, which is a Chinese foundry. So we're going to use a Chinese foundry to make our chips. The other was called DataEd was actually an American manufacturer that did the bulk of its manufacturing in Chenzen in China. And both of these supply chain components actually fell through, and they fell through late. So when SMIC told us that they couldn't work with us, we lost probably two months of development time because the chip was already almost complete when we got this news. And so we had to shift to a new foundry that cost us a lot of development money and it cost us a lot of development time. And then same thing. When data ed told us that they were unwilling to work with us anymore, we had to basically scramble at the last minute to find,

new manufacturers. And since we lost out on the China pricing, this also cost us several million dollars. We ended up having to pay the price of American manufacturing, which is substantially higher. We didn't have time to work with, say, Singapore or Mexico just because they dropped us without giving us enough lead time to find a replacement except for people, you know, that we could go drive to every week to get things moving quickly. And so, you know, this is just an anecdote.

There's never any sort of trail to indicate that Bitmain was involved in any way. But we do have two places where Chinese manufacturers dropped us just like we were warned would happen. And do they think of any reason why they were dropping you? So we don't have a good reason from either of them. Admittedly, I was not directly involved in the conversation with SMIC. our chip devs were handling that. So I don't know the full reason that this happened with our manufacturers. I was more directly involved.

And our engineers described their excuse as unprofessional. And it didn't make sense. And it wasn't reasonable. And it's not what you would expect from a manufacturer. But other than that, we don't have a very strong reason. Some people think that Bitmain was secretly mining coins. you referenced this a little bit that they were doing that with new equipment before selling the equipment to customers. However, Zuko had this conversation with Gihon and then he, and Ji Han

Wu is the CEO of Bitmain. And he published a recap of the conversation. Apparently he asked Jehan whether or not Bitmain had been, you know, doing this secretly mining the coins before selling the equipment and he said no. So have you looked into that at all? And did you read their conversation and what were your thoughts on what they said? I did read the conversation. It doesn't match what I know. The things that Ji Han claimed don't match the things that I've heard from other people associated with Bitmain. And wait, what specific things did he claim that don't match what you know?

Yeah. So one of the things that he claimed was that the secret Monero miners were not Bitmain and that he didn't know anything about it. he hadn't had time to look into it. I can confirm that we believe that the secret Monero miners were not Pitmane or someone else. However, we also believe that Bitmain was heavily involved and that Jehan absolutely knew what was going on and could have named the people who were doing the secret monaeromining.

So this is one thing where, again, we're not 100% sure, but we're very confident that Bitmain as a company, if not Jihan himself, was involved in some. way with the secret one narrow miners. The other thing is that... And do you have any sense of how they were involved? I don't think I'm able to share. Okay. So keep going. The other thing? The other thing that we were told from people close to Bitmain, and again, this isn't, you know, this is not confirmed, but we were told that Bitmain had been mining Cyacoin since November.

and of course they didn't announce their minor to the site public until or to get to the public until January, middle of January. However, as soon as they announced, the machines were shipping, you know, within like 10 days, it was clear that they had been making them at least for several months. And we were informed, again, by people close to Bitmain that Bitmain had been mining since November. And this is apparently, at least as far as rumors go, something that's very common within Bitmain where Bitmain will mine for several months. The rumor goes three to four months, usually,

before releasing hardware. A complicating factor is that we know that Bitmain far and away is not the only person involved with secret mining, and we know that there are several efforts and several groups that participate in secret mining. And so though we can point at certain trends and say we're pretty confident there's secret mining happening here,

it's difficult to figure out which group. And in many of the cases, it's not Bitmain. It's someone else who's doing the secret mining. And are those groups also mining their own, manufacturing their own equipment, or how can they be secret mining? We believe that some of them are manufacturing their own equipment or using help from groups like GUC. We believe that some of them may be contracting, like say,

either Bitmain or Bicol. We believe Inno Silicon does a lot of their own mining, although, again, we don't have any definitive proof. I mean, it is very difficult, like, secretive space to navigate. But we think that most secret mining rigs are manufactured by known names, even if the groups funding the secret mining and the manufacturing process are not the known manufacturers. Interesting. You did write earlier that mining manufacturers sell mining equipment instead of keeping it for themselves only if they think they can get more money for it by selling it than by basically printing their own money with a machine themselves. Is that what you meant when you said that? No, that's actually not what I meant. I was more pointing at the public mining and say like Bitcoin. For example, why would a company like Bitmain ever sell a money printing machine?

for less money than it's going to print. And I think that the clean answer is that they wouldn't and that they don't. They will only ever price a mining rig that they are selling at a price that is above what they believe they could make by mining it themselves. Oh, right. Okay. So, but then to draw that conclusion, what do you think this means for whether or not the average person can make money off of mining?

Like, can they? Or, you know, what does this mean for retail? buyers. Yep. So I think that it means that retail buyers cannot make money off of mining. And I think that that's a luxury which is going to disappear. Now, it may make sense for Bitmain to sell hardware to a professional mining farm. If that mining farm has access to, say, one cent electricity, and the best bit main is able to do at scale is, say, three or four cents, because really cheap electricity is difficult to do at scale, then there may be some reason there where that specialized mining facility can make more money than Bitmain can. But you're not going to see that in a retail environment.

You don't see retail people with one cent electricity. Typically, it's closer to 10 or 15 cents, which Bitmain can do way better than that. And in GPUs, I think in the long run, any GPU mining is going to go the same way because eventually you're going to catch the attention. of Nvidia, if the ecosystem grows enough, Nvidia is going to start mining themselves. And of course, it's the same thing. Why would Nvidia sell a GPU for $800 if they knew they could make $1,000 by keeping it? And they won't.

They do it today because cryptocurrency isn't on their radar, or at least they aren't willing to take a risk on holding that type of inventory and mining themselves. But if the industry is stable and proves to be a long-term, like, stable investment, I think you will see companies like Nvidia getting involved and mining personally. And again, that means that retail miners aren't going to be able to make a profit. To go back to this conversation with Zucco, after Jihon and Zucco spoke, Bipman wrote a blog post and also some tweets saying that it would now try to conduct

this business in a more transparent manner. and so they tweeted out updates on the shipment of the Zcash miners. What did you think of this move? So at this time, I think we should view it as purely a PR stunt. I actually haven't seen the tweets they put out, so I'm not sure. I think that the information is useless unless they also disclose the exact number that they have queued for manufacturing. I don't know if they disclosed that number or not,

But I think that if they didn't, you can point to that as being non-transparent and as being, you know, an illegitimate attempt at transparency. Even if they did provide that information, I would question whether that's a long-term commitment that they're making or it's something that they're doing to clean up their image in the short term, especially if they're dying, say, like an IPO and they want to have a good reputation going into the IPO. is that something you can trust in the long term? Because Bitmain's DNA really has seemed to be about short-term profit and about not caring about, you know, screwing up the rest of the ecosystem or really, like, digging into the rest of the ecosystem. If that's what makes the money, that's what they're going to do. And so I wouldn't trust that they've changed that DNA without seeing substantial,

like, more long-term and, like, heavy-handed commitments to transatlantic. So maybe they have changed, but I would definitely wait longer to believe it. Have you ever spoken with Jihon or met him in person? I have not. Okay, so you didn't reach out to ask any questions about anything? We did reach out when they announced the Siam liner. And we got put in touch with a lower level marketing person. And basically they more or less stonewalled us trying to get basically all of our

customer information. They wanted to know the name and address of every single one of our customers. They claimed so that they could give our customers coupons, but I don't think under American law it's even legal for us to do that. So we at least had one interaction that felt half-hearted or just abusive even. Beyond that, we haven't made too many attempts since January to reach out to Bitmain. Hi, guys. Laura cutting in here with some notes recorded separately from my interview with David. As I mentioned, I asked Gihon Wu, the CEO of Bipmain, about the various allegations David made on the show. About whether or not Bipman was involved in Obelisk's foundry and manufacturer, SMIC and Data Ed,

halting their work with Obelisk late in the process, Jihon said they didn't influence their manufacturers to work or not work with Obelisk, that the manufacturing business in Shenzhen is very competitive with numerous manufacturers, and so how can BipMain control all of them? He also said Obelisk was blaming Bipm. for Obelisk's supply chain management failures and that this was irresponsible. About the allegation that Bitmain was involved in secret mineral mining, Jihan said, no, that Bitmain was not. Quote, David accuses Bitmain without any evidence.

About manufacturing Sia coin miners, Jihan says that it was Bitmain who approached David through a consultant and that they had actually delayed the release to have a face-to-face discussion with him. Jehan said after it confirmed that David had a hostile intention to change the proof-of-work algorithm of Sia coin. Jehan said Bitmain decided to release the A3 minor. About asking for the names and addresses of the Obelisk customers, Jihan responded, quote,

Oh, this is so manipulative. We provided several options when we said that we can help. We thought we should take responsibility to help their pre-order customers. We gave them several options. Here I'll paraphrase the options. One, Bitmain issues coupons to Obelisk's customers. Two, Bitmain ships its A3 miners to Obelisk customers with those customers consent. Three, Bitmain ships all the miners to Obelisk, and Obelisk can reshift the miners so their

customers' information can be protected. And then Jehan said, quote, however, they just refused. They refused to work with us in any way. About whether Bitmain was mining Sia coin before announcing the miners, the A3 miners, He said that they didn't do large-scale industrial mining on Siacoin, only testing. He also said stealth mining wouldn't work at a large company like Bitmain because it would require more than 100 employees, whereas a smaller operation like obelisks could more easily do it.

Finally, about David Singh, he didn't believe transparency from Bitmain would be useful unless Bitmain released the number of units it had queue to manufacture. Jehan responded that they wouldn't release such numbers unless every competitor released the same number. Overall, Jehan said, quote, we felt he was in a seriously manipulated mindset against us, and we decided to release minors without discussion with him anymore. He needs to verify the information he gets and make educated decisions. Now, back to the rest of my interview with David. Speaking of minors, we have been talking a lot about Vim because they're the Big Fish,

but you also wrote a blog post that talked about another minor, and I don't know how to pronounce this, how long, I think is the name. And they created a Decred miner that they sold out of. But then later you apparently discovered that 50% of the mining rewards in Decred were going to an address associated with How Long.

So how do you think that happened? And what do you think it says about how we can potentially curb the power that mining manufacturers have? Yeah. So I, following the blog post, Halong reached out to me, and they insisted stubbornly that that wasn't their address and that they weren't entity mining 60% of the decredit hash rate, which after further investigation, I now believe that it was actually Inosilican who owned the hash rate. We did see 60% of the hash rate going to a single address. At this time, we believe that address is Inosilicon. However, we're not 100% certain. So I may have made that claim to boldly on my my blog post. And I've been meaning to go back and edit it. However, it's, it's nonetheless the

case that it, whether it's Haylon or Inosilican or it's a third party, someone seemingly had 60% of the hash rate. And that this is, this is an issue. And it's one of the big, like, reasons that we wanted to create Launchpad was because it's like, if you, if you don't have a controlled breakout with A6, you end up in a situation where you are vulnerable to whoever getting there first deciding to own 60% on hash rate. If they start to enforce, say, like a soft fork that only allows their mining hardware to be effective on the network, that could permanently block all of their hash rate from the network unless the network does a hard fork. And so Launchpad is the best that we know how to do in solving the,

the first miner to market problem. But we've seen consistently that whoever gets first to market for mining hardware, for cryptocurrency, the cryptocurrency community tends to be incredibly unhappy about that situation. And there tends to be a lot of negative side effects. And so launch pad is our best attempt at remedying that, though, as you pointed out, it does have its own, like, trust issues and it is sort of a centralized beginning. And so it's an incomplete solution, but it's the best we know how to do. And so far, it's going to be Siacoin and DECRE that you are creating mining equipment for.

What else is on the roadmap? Yeah. So we have two projects that have been commissioned that we can't talk about, but Oblisk is working on two additional projects. And then beyond that, we are currently fundraising. for to make a Bitcoin miner. The amount of money that we need is fairly large. So it's currently looking like we're not going to get there. But we do have our site set on, if not in 2019, then maybe in 2020, making a Bitcoin miner.

And will you be continuing to manufacture all your mining equipment in the U.S.? At least for the time being, it looks like we will be staying in the U.S. for manufacturing or maybe going to other parts of North America, such as Mexico. How low would the Bitcoin price have to go for us to see a significant shutdown of facilities? Based on our recent research, I think that if the Bitcoin price stayed below $5,000 for a substantial amount of time, you would start to see, you would see the hash rate stop growing. And so you'd stop seeing new mining hardware being purchased.

If it went below $3,500, I think you'd start to see, you know, big shutdowns of mining equipment. Interesting. So we've spent the majority of the episode talking about mining, but I also wanted to ask you about something that you mentioned recently, which is that you're working on a scaling solution called microchains. Can you give us a brief description of what this is and how it works? It's more or less a generalization of just what we already have with altcoins and the lightning network. So what I've observed is that, you know, Bitcoin stuck at, say, three transactions per second, light coin stuck at maybe five or eight.

And, you know, each individual blockchain has a pretty, like, limited throughput for what they can achieve. But with the Lightning Network, you can allow people, you know, people on the Lightcoin blockchain can transact trustlessly with people on Bitcoin blockchain over Lightning. and neither user needs to have the other chain. And so when you expand this to a very broad ecosystem, you can get an ecosystem of say 100 or why not blow it up even more, make it 100,000 or 100 million blockchains where a user only needs to be on this one tiny chain that's very easy to run a full node on,

but through things like the Lightning Network, they're able to transact trustlessly with any other user on the network. And so when you do, when you take that approach and you take the approach of making a ton of really tiny chains as opposed to a small number of really big chains, a lot of interesting game theory happens. And so microchains when I release the blog post that I've been working on will explore a lot of the game theory that happens. But it's something that I think currently think might be a viable way to bring heavy scalability to the ecosystem and also bring about more mining decentralization. I think there are some mining advantages to an ecosystem like this. Well, I look forward to reading it when it comes out. Where can people learn more about you, Siacoin, and Obelisk?

I think the best place is probably blog.Sia.com. And then Obelisk has a parallel blog at blog. Great. Well, thanks for coming on Unchained. Thanks for having me. Thanks so much for joining us today. So learn more about David. Check out the show notes inside your podcast episode. new episodes of Unchained come out every Tuesday.

If you haven't already, rate review and subscribe on Apple Podcasts. If you like this episode, share it with your friends on Facebook, Twitter, or LinkedIn. Unchained is produced by me, Laura Shin, with help from Elaine Zelby, fractal recording, Jenny Josephson, Rahul Singh-Ready, and Daniel Ness. Thanks for listening.