00:11Hi everyone, thank you for being here Yeah, so let's be frank it took a little bit more than just the math so a lot of the blood sweat and tears of Incredible people in this room Got us here So hi everybody. I'm Omri. I'm the co-founder and CEO of Pearl and

00:33I wanted to deeply thank each and every one of you For being here and for your time and interest in the Pearl project for centuries Money's been a social proxy for human labor and innovation managed through social contracts credit tax and monetary policies Well, it's hard to predict where AI is headed or what are exactly its limitations

01:03One thing that's becoming clearer every day now is that our future economy will be denominated in compute cycles more than it is in in human labor and in this reality where the production of knowledge is Becoming more mechanical in some of the frontier labs in this audience are spending 100 million 130 million dollars a day on inference if I'm correct

01:28One could argue that the real currency is not money It's energy and data the two what we collectively call Compute so the two scarce resources whose fusion creates intelligence is is data and energy data and compute We believe that this dramatic shift in the production of knowledge compels us to rethink the fundamental Properties purpose in the creation process of money what money actually is

01:57The basic scientific question that started This the pearl project and we Elon and I asked ourselves About a year and a half ago was whether we can build a monetary currency Directly from these two key scarce resources data and compute This is the motivation for this question is not merely mathematical or philosophical It is to address at least three core incompatibilities of the fiat system

02:29With emerging AI economy The first the first one is that the unit economics of AI today is simply non-fungible not all LLM tokens are born equal producing a Deep-seek R1 LLM token on NVIDIA hopper has a very different physical cost production cost then producing a reasoning token of chain of thought of Quinn 14 B on AMD MI 3's right so these are just two very different physical costs

03:02The current crude market pricing of you know X dollars per million LLM tokens is at best a crude proxy for that this actual energetic cost But it it hardly serves as as a proxy for the data dimension, which is much more elusive and harder To articulate The second drawback or a limitation is that in the current economy AI consumers are

03:29Completely left out of the upside of the wealth generated by AI so AI users like myself who are you know Just using Claude for for coding or or chatbots. We don't get an anthropic or NVIDIA share Well, most of us at least But by doing so yet consumers are the ones driving demand and model improvement through prompting RL data labeling RLGF and so forth, but we're we don't really have a stake in

04:01The in the wealth generated by the AI revolution, right? So that's that's another part the third one and final one is that in this in this era where Autonomous AI systems are slowly penetrating our you know economy and driving a lot of the value We must have an economic system that allows machines to participate alongside Humans in the same ecosystem

04:28Without you know onboarding KYC and identification, right? That's we just have to allow machines to somehow manage the two only resources we deposit in their hands or in their You know in their in their system, which is compute and data, right? So AI agents just receive two researchers the data the prompts in the in the compute budget And we believe AI agents should be able to transact seamlessly and directly using these resources and to manage them without any permissioned or

05:03Regulatory or social Restrictions with their obviously not subject to Right. So this is what we often call Among ourselves that a money for machines, right? So it's money for machines and humans And we will see how to get all of these three properties and more and then some we will see actually additional Functional properties that were not possible for AI systems that are

05:31In some sense follow from the technology will present in the tool will use to create this AI native currency is proof-of-work blockchains All right, so a Three-minute crash course on on blockchains. What are blockchains? So blockchains are a technology that tries to answer the following challenging question of how do we all collectively?

05:57maintain a Decentralized maintain an update a decentralized peer-to-peer Transaction ledger or network Okay in a way that is consistent that each and every one of us Despite

06:12Having hits its owner her own local view. How can we all maintain and update a consistent state? in a way that's Without any trusted third-party. Okay, so even if 49% of us Are failing or are abstaining from reporting anything? We want the system to remain live right for this to happen each and every one of us should be able to replicate the state

06:40of this data structure of this ledger Right and another feature is that we ideally we would like this system to be permissionless So users can come and go enter or leave the system without any onboarding process So we don't actually know how much compute or how much Users are participating in the system All right, so the first thing you might think is you know, how should we go about it like?

07:05There is no central bank. There is no single point of failure. How should we all agree on the next state, right? This system is useless if we can't if we just maintain it We have to be able to agree on the update rule how do we update and collectively agree on the next state of this chain and The most natural thing you can essentially the only thing reasonable thing to do Assuming you have honest majority which is the weakest thing we can ever hope for right if there's no honest majority

07:33All bets are off. Anyways The most natural thing we can do is just take a cast a majority vote, right? That's the most natural thing to do if there's an honest majority. I will just Well gives a higher chance and then you can maybe repeat it and ensure that an honest The honest majority determines dictates the next State the problem is that you know besides being inefficient. I can't go to each and every one of you and just solicit a majority vote

08:01this rule is actually not secure because by Abstaining from voting some of you who don't think the next transaction is favorable to you Can just collude and abstain from voting that could actually bias the answer to something different, right? So that's actually that's the majority vote is simply not secure The second idea is maybe we can do something more efficient

08:26Which is kind of an unbiased estimator for a majority vote, which is just let's just sample a random leader and let her Or him approve the next state. Okay, so this is kind of an expectation. This is simulating a majority vote But then again, remember, there's no trusted party. There's no single blackboard written in the sky Who draws the randomness? How do we agree on the support of the distribution? Right? It's not remember It's a permissionless system. We don't even know What is the sample space of this distribution, right? So these are, you know, two

09:02You know unsuccessful trials, but actually if you think about it the biggest problem with the majority vote is The fact that if voting is cheap no matter how you implement it if voting is free Then Alice could just replicate herself 10,000 times and just Pretend to be 10,000 voters and she could just bias the result however she wants because you know Zack would communicate with her and he would be led to believe that Alice is acting like 10,000 voters

09:31Does that make sense? And if you think about it, this is you know, there are many ways to view Bitcoin as technology, you know You know, it was a breakthrough in many senses. This is the core innovation of Bitcoin The core conceptual innovation was the guiding principle That making voting expensive is just designating who's allowed to vote and Bitcoin's rule was

09:58voting entails demonstration or investment of a scarce but abundant resource so an expensive but abundant but widely accessible resource Okay, so that was and if you think about it, this was you know, blockchains have evolved in the past 20 years To many other consensus mechanisms, which some of you might know All of them stick to this guiding principle where users or voters or minors will

10:26invest resources Expensive but abundant resources to secure the system in order to incentivize them to invest those resources The system will issue rewards coins or money or however you call it in that will incentivize the the the users to keep securing the system Does that make sense? Bitcoins or Nakamoto's

10:51Observation was that electricity is a resource that satisfies these two properties. It's abundant enough. You know, it's widely accessible in the world But it's definitely not free at least in most countries so Bitcoin's idea was to enforce this voting system or Or, you know Consensus protocol through what's called proof-of-work protocol. So in two minutes, what is proof of work?

11:22So in a blockchain we have Transactions grouped into blocks, right? So in this block we have user ER 20 Transferring 10 coins to user 0 and 5 for example And then each block has a unique ID a unique identifier or a hash code that uniquely identifies it and If Alice wants to add a transaction to this ledger to just approve

11:48She wants to transfer Bob maybe $20 20 coins then What will happen is that Alice would need To update this and the only update operation in the system is the operation of extending a block so You can extend any block in this chain. So it's actually a tree. It's not really a chain But what defines honest users is that honest users Only believe the longest chain the longest path from root to leaf, okay?

12:18That's very it's probably it's going to be intuitive, but it's going to be clear later Okay, so Anything else than believing the longest chain is considered, you know, all bets are off I'm not assuming anything all I'm assuming is that at least 51% of the users are Following the longest chain. All right, and now If Alice wants to add

12:40This transaction to extend let's say Alice is honest So she's trying to extend the the longest chain. So it's like plugging a socket into the To the wall or whatever and she would need to solve a computationally hard problem so in order to add this transaction to the public ledger she would need to

13:02Take the idea of the block. She's trying to plug into to extend She would treat this string as just as a bit string and then Alice's Task in order to approve this block this transaction is to find a completing string this black puzzle piece That will cause a random hash function to evaluate to a very small number Okay, if you know anything about, you know, random hash functions, you would know that any

13:29Guests for the black puzzle piece, which is a bit string has equal chance of evaluating to this number Right and importantly in Bitcoin the number D, which is called the difficulty of the system is dynamically adjusted So that in expectation No matter how much compute there's on the network exactly one Winning puzzle piece arrives every ten minutes. Okay, so it doesn't matter if there are one Hexa flops or 23 tera hash flops in the system the clock the clock keeps

14:06Taking every ten minutes. Okay, so if you think abstractly, what is proof of work? It's just a distributed clock That clicks once every ten minutes every ten minutes. There's exactly One block That's being opened does that make sense? Yeah, and for for them mathematicians in The crowd what characterizes proof of work is it's just a Poisson process where the arrival of the next lottery ticket Is just a Poisson random variable

14:43Why am I telling you all this? So you might think so is the process clear so this is the only operation. This is how the system works Notice that it avoids the problem of soliciting Votes it's completely asynchronous right because I don't care if you know, you don't Avoid voting. It's his loss. The system keeps going right So you might think okay, that's just a random arbitrary

15:11Hard problem random hatching just random number guessing right that's just arbitrary Why not do anything else? But if you think about of it? There are two key properties There are crucial about random hashing which makes it work the first one is that the proof of work puzzle this puzzle piece that Alice needs to solve that miners need to compute it better be hard to compute But very cheap to verify right so if Elon finds this puzzle piece You know finding it is exhaustive search, right? It's just a random number guessing just every ticket is has the same winning chance

15:46But if Elon just shouts out this number I immediately know it's right Yep, the second thing is that in order for the system to be open truly open and fair There cannot be easy instances for Alice right there cannot be like low-hanging fruits because then a dishonest miner Will just pick the easy instance and it will mine it will just scoop rewards much faster And then we just lose control over the clicking of the clock right every economy where you cannot control the pace of

16:20Minting new coins will collapse right so that that that won't work. Is this clear? so these are two very crucial properties and Sorry, so yes a more mathematically. I would say that any miner or user that Spent holds in his hands P fraction of the total hash rate or the total compute in the network

16:46Should have a chance of winning which is essentially P So if you have P fraction of the compute it's only fair It's the only fair thing right if you have a P Fraction of the compute in the network You're winning you're gonna scoop a P fraction of the rewards right and that makes us completely fair in Symmetric, so the the hardest is just uniform for everyone

17:08Okay, and turns out random hashing satisfies this and You know it's been what like almost 20 years Bitcoin is past the test of time whether you know if you ignore The bumpy road of the price it it is a pretty substantial human achievement that such a system actually It actually works right it works and it's slowly getting gaining more and more Approval higher up But the obvious problem is that Bitcoin is just

17:40Consuming as you saw in the in the video a ton of energy just on random hashing right so between 1 and 2 percent of the world Global electricity more than I think Argentina and Norway is spent just on these random number guessing Right, which of course are securing the chain, but beyond that it's just garbage, right? So there's no there's no extra additional utility and if you think about it this problem is even worse today because Bitcoin and AI are actually competing over the same electricity their mind

18:11They're being operational by very different hardware, but they are competing at the core of it for the same energy Right, which is why a lot of the hyperscalers Are replacing today their Bitcoin racks with GPUs and TPUs because the profit per gigawatt is higher in AI today as it is This is commonly referred as Bitcoin's declining security budget where aside from AI as the chain is Becomes more mature and less and less reward are issued

18:41The incentive to keep mining and participating decreases So which that brings us to the you know extremely natural idea Which we were far from the first to conceive of proofs of useful work, which is that you know Most natural idea is why can't we replace random hashing with something with a tacit actually useful, right? that with electricity that's already being spent to do I know drug discovery or Quantum simulation or scientific computing and

19:11This challenge was acknowledged by many thought leaders in the past 10 or 15 years In fact Elon's advisor in the the Weitzman Institute was the first to acknowledge this problem the very first proof of work paper way before Bitcoin one of the So there's been a long line of work on this each and every proposal for useful proof of work was shown to fall short of

19:37satisfying one of those core Definitions or properties of useful work and indeed none of those attempts have succeeded One of the most vocal votes was Voices was by Vitalik the co-founder of Ethereum who said that if we can find some useful Computation which is easy to verify then crypto mining Could be could actually become a huge boon to society not only removing

20:05Subjection that bitcoins wastes energy, but also being societally Beneficial and then a few lines later Vitalik ads this problem is is probably infeasible in a few months later A few years later a theorem switches from proof of work the proof of stake and the rest is history all right, so this brings us to the

20:27the pearl kind of Area where What we're trying to do is essentially a following so consider any real-world problem a protein folding DNA sequencing quantum simulations Vector databases anything you're you're working on and it's being deployed in the real world regardless of blockchains and Suppose this task as f of x you're trying to compute has worst-case complexity t

20:55So worst-case complexity is just the running time of the algorithm that solves any instance of this problem So for sorting n numbers He is roughly n log n For multiplying two n-bit integers He is roughly n squared If you went to Stanford, you probably know that you can multiply numbers faster using FFT, but we'll get into this

21:17So this your n squared is roughly The best Practically like the baseline will use here and what we're trying to do here is We're trying to piggyback on the same Native algorithm or the native computation that computes f of x that's gonna happen regardless of the blockchains, right? People are gonna keep multiplying numbers and sorting things regardless of anything

21:44We would like to have with very little extra overhead Beyond the worst-case complexity. We would like to have two outputs. So two for one Right, we would like to have to produce the output the output f of x which the algorithm does anyway And it produces a lottery ticket a certificate exactly like Bitcoin which has this Poisson behavior Does that make sense? great and

22:11If you think about it The reason proof of useful work was so challenging was unsolved for the past 20 years is That unlike random hashing Real-world problems just they simply don't have the structure that or the verifiable structure that random hashing does right? For the system to be truly useful. I better allow the miner itself to pick its own her own favorite instance x right the input That's very different than Bitcoin Bitcoin you get the assigned the problem by the network and proof of useful work

22:46You're truly useful Proof-of-work chain the miner itself is picking the input right so Alice can choose x at her own discretion And if you have some external gains from it like training inference quantum simulations and good for her She'll piggyback and just on those rewards, but it's crucial to allow Alice to choose her own input Of course, you know real-world problems just don't have this uniform hardest They vary a lot in their difficulty let alone for dishonest players who don't care about the useful word and they can just fabricate

23:20Inputs right so in the multiplication example Alice could just choose a and b to be the zeroes right multiplying Zero by zero is not doesn't take n squared time. It's just trivial Does that make sense so this lack of uniformity? Means the system is going to be very gameable and that's the core challenge of proof-of-use for work Does that make sense? so in the in the paper that you line and I published

23:50About a year ago what we studied is the is the problem of Implemented proof in Nakamoto's consensus on top of a specific task, which is matrix multiplication Why matrix multiplication? I guess I was at the time in Nvidia, but probably The reason was was deeper than that and you'll see in a moment So Matt Maul is just Under the underlying technology of not just training an inference an RL and post-training anything you can imagine

24:20It's probably the underlying Operation probably between 70 and 90 percent of any industry scale operation you can imagine You know self-driving cars robots compression quantum simulations navigation everything in the world is is Matt Maul is the most optimized operation in the In the the history of humanity

24:44So what is matrix multiplication, I'm not gonna get too deep here basically we have two arrays of n by n numbers to simplify Right and we're required to output another array Which contains the product of every row in the left matrix a with any column in the right matrix B Okay, so we have n choices for the rows and n choices for the columns of B Each product has n numbers so it requires n time if you if you multiply all these you get n cubed complexity So this is what's called naive matrix multiplication. That's the that's the algorithm that in video AMD cerebrus edge everyone uses today

25:25There are theoretical faster algorithms, which is part of my academic research, but that's that's not That's not Practical anywhere soon and even if it will be pro will be able to adapt to this but this is something we can take offline so the way GPUs multiply matrices right in

25:46in Normal LLMs the matrices could have like 40k by 40k dimensions, right? And so the GPU would actually tile the matrices So you will split into blocks like r by r tiles It would just load the tiles and multiply talk. That's that atomic unit of operation okay, so you can do those things just by

26:07Multiplying instead of a single an entire row in a column you would just multiply a tile by a tile Which is a smaller mathematical right and that fits into memory and that that you can do All right, so in this paper that we we wrote We showed how to generate a mathematical proof which it has exactly mathematically the same Distribution of Nakamoto consensus but

26:33With a tiny Overhead beyond the worst-case complexity of math mode you can actually Produce those the same security mechanism of Bitcoin What's the math behind it again? It's just two minutes and I'll I'll try to take it easy on you so

26:52What's the naive attempt to do this right so now instead of there's this added dimension of data that Alice gets to choose She picks Alice gets to pick a and b if Alex is Alice is running I know Claude or you know or or or Quen or Kimi K2 she would probably choose a to be the weight matrix of some layer and B to be the activation matrix, okay? and

27:20But she can choose anything she wants right remember no one's you know, it's at completely our own discretion Alice could choose Matrices she pre-computed at home a week ago. No one will know right. There's it's a permissionless system. There's no way to know and The most natural idea is to just do the following instead of hashing in order to extend the block as we saw a few slides ago Instead of hashing the idea of the previous block Why don't we just hash the result of some useful computation of the mathematical right so we just as will compute the mathematical and

27:58I'll just apply the hash to the result of this useful computation. That's negligible compared to the cost of mathematical The problem with this is again if a and b are Unpredictable or random let's say then this actually works works perfectly So if a and b are like ran a hard matrices to multiply this would work the problem of course is that again Alice is free to choose her favorite a and b so she could choose the all-zeros or diagonal matrices or Whatever she wants and the problem is that this will allow Alice to reverse engineer

28:33The the winning criteria because everything is deterministic right there's no randomness in the system So Alice could just reverse engineer everything that you could just keep scooping rewards very fast and game the system by choosing easy Or predictable and I hope this makes sense So the idea is that we must somehow bake in or embed randomness into the native algorithm that computes math mode Because if there's nothing unpredictable in the math mode algorithm Alice can just pre-compute it at home and just game the system does that make sense and

29:10The natural way to do this is to just add a little bit of noise to a and b right that's a very natural thing to do and Indeed This is what we would okay So this is what we're trying to seek here for is what theoreticians call a worst-case to average case Reduction which has a very succinct overhead parameter So we would like to take any a and b that Alice produces and we'd like to compute to reduce it to the task of

29:37Random a times B without paying much more overhead And the first trial is to just maybe we just add Random noise matrices so now no matter what Alice chose even if she's trying to game the system and choose the all-zero matrix as a and B If we add random noise matrices Alice is doomed. That's gonna take n cubed Operations to multiply because random matrix and random Gaussian not moles is n cubed hard. We can take this offline

30:11So now we do the proof of work on the perturbed matrices that's hard right so great That's that has bitcoins like security thing But of course the what's the problem the problem is if Alice is an honest user running AI models She doesn't care about the perturbed product. She wants the original a times B the product of the white matrices So what should Alice do in order to decode the white matrices? She should just peel off the really just open the brackets

30:39She just needs to peel off three extra math moles right right to get the original result She would need to do extra post-processing operation. Just so what did what did we do here? Like Alice paid four math moles to get just one so the overhead here is like 400 percent. It's like x4 We're looking for Little of one overhead like very very small right no one would use this this system No one would pay like four math moles. You can actually squeeze. It's a nice exercise. You can squeeze out

31:08Well, it's definitely gonna take at least two or three math moles. So that doesn't work. Okay so The real Protocol the eventual protocol which we call pearl gem. I think this is the probably pearls biggest innovation Is to follow the same idea, but in order to make the peeling of the noise the decoding cheap Maybe you can just add structure to the noise. So instead of just using

31:39brute force random Gaussian matrices What we can do is we can just add the noise matrices which have They look like kind of kind of margin marginally random, but they actually have structures So if we add what's called low rank random Gaussian matrices low rank matrices are still end-by-end matrices if you open them up But they have linear correlations. So actually you can decompose them into the outer product of two like

32:10You know strip matrices and The thing you should know is that this completely solved the peeling problem because now if we ask out if now Alice Needs to peel off those extra three products if you do the math You'll see that all those peeling products are gonna be low rank rank our matrix multiplication That's gonna cost our n squared which is negligible compared to our cube. Okay, so in practice Elon and and Vipple will talk about this. We'll see what's the actual overhead in practice. That's great because now peeling is just negligible, right?

32:48But I claim this this screws up another thing This protocol is no longer secure because if Alice now chooses the all-zeros matrix matrices Alice is left with a much easier problem than an honest player that's multiplying dense matrices Alice now Just needs to multiply low-ranked matrices rank our matrices Alice can do this operation in our n squared time Whereas an honest minor running actual LLM math moles is gonna pay n cubed

33:20That's completely breaks the security. The system is no longer fair, right? There are easy instances. Does that make sense? It looks like we're trying to hold the string from both ends We tried to make the system secure So we added we added 300 percent overhead and we tried to reduce the overhead We screwed up the security, right? It seems like fundamentally this you know this trade-off seems inherent And I think pearls, you know the key innovation here the key conceptual idea was the following that

33:50It's you know for an honest minor for an honest honest Alice that Wants to run LLMs all Alice cares about is the output is a product of a and b right? She doesn't care about the intermediate computations But no one said we need to do the proof-of-work on the output we can we're free to do this on Intermediate computations as well, and I think this is the key idea. So we would like to prevent Alice From exploiting the low-rank structure of the noise in the intermediate space, right?

34:19And this is how we do it so instead of just verifying or hashing the output of A time of a prime times B prime which is gameable. We just saw this is easy Instead we're going to verify we're going to do the proof-of-work Not on the output, but on the intermediate computations that the GPU already does so when the GPU multiplies those to our by our tiles We will just for every such our by our tile that the GPU is going to multiply anyway. We're gonna

34:51Force Alice to hash the product the the result of those intermediate tiles and Check if it starts with let's say 19-0. This is exactly bitcoins adjustable difficulty have a puzzle problem So so and the nice thing is that about low-rank random matrices is That marginally every our by our tiles that you see here is going to be uniformly random This is just a our wise independent distribution. So that's a standard property of course, there's tons of correlations between those styles, but

35:27That's that's the hardest conjecture and then that that's the protocol. That's it. So It was a little quick, but that's it. So Alice just submit commits to their matrices No matter what she does We're gonna add low-rank noise and then we're gonna verify it by having every GPU cycle that multiplies those Intermediate tiles add an extra hashing operation and check so think of it at every our by our tiles Gives you an independent lottery ticket to open a block

35:57This is what Elon is going to talk about later in practice And the hardest assumption is that indeed in order to open all those blocks you need and Cube time which is the naive thing. So we got both peeling which is very cheap and we got security assuming this this Assumption here which has been tested by a lot of world-class

36:22Mathematicians some of which are in this crowd All right, so just to finish up. I have like three minutes What we did in the past year is take this protocol from white paper from the math all the way to production What it means that we actually spent most of not we like most of the pearl team here spent most of his time in the past year just Coding up those kernel those kernels and CUDA and soon AMD and you know

36:50we're slowly expanding Elon will say more about this and what we have today is a is a software package on GitHub which probably some of you have seen which a lot which is you know, which has a VLM plug-in and allows you to run those two-for-one LLM Modules and Really the implication of this system

37:15Which is you know, we're slowing trying to understand this this will be more apparent in Raphael's talk Later this this morning, but the obvious application of pro is just funding the AI build that right So they're reducing the price of training and inference because now using those two-for-one CUDA kernels every GPU cycle that you know

37:42Every new lab is gonna or hyperscalers gonna run can with a very small overhead Which you learn will discuss soon Can just do two-for-one so you have a parallel revenue stream for every GPU cycle on earth It's very important to realize this is not a verticalized system, right? It's not only for training or only for inference or only for closed source or only for open source data centers edge Transformers diffusion models world model doesn't matter everything is not bulls. So the I think the strength of pearl is that

38:15The fact it's not verticalized. It's just an index on all an AI and all of its forms, right as long as Matt most is still around Pearl should be part of that the total addressable market of pearl is According to Eric Smith is like 90% of the electricity in 10 years. So it's I think that's the The strength of pearl and if you think about it this thing, you know after we crack the technology It creates a completely new economic model

38:46Which people just economists haven't thought about because now you have this proof of work incentive but you have also external incentives from being paid for training and inference in the equilibrium dynamics of this system and Implications the market cap of inference and training in AI are completely new things some deep Mathematical questions, which we'll hear from in Rafael's talk in an hour or so or like half an hour And that's it. I think

39:16you know You know Presumably the most native miners or users of the pro network are going to be autonomous AI systems and If you think about it, this is going to be the only currency that AI agents insist in AI systems can natively produce without any KYC any on-boarding process any Regulation is they just take their computing data budget. They just manage it completely seamlessly

39:42So that's I think a unique feature of the system Of course, it has more unique applications that I won't get into like essentially what pearl is doing It's implementing a decentralized voting system for AI for all of AI for all of AI agents systems in the world Because you know if you just go beyond transaction ledgers and decentralized banks what we did here is a completely generic way to Implement a majority vote. So that's that's something that we're also exploring. That's it. Thank you so much

00:05Hi, everyone, my name is Rafael Paz, and I will be talking about the economics of proof use for work. So this started some time ago when Elon and Omri came to me and told me about this amazing protocol, and I loved it, and they said they've been talking to a lot of bunch of people about it, and everybody that hears it loves it, but they often get the complaint or concern that if there is a useful proof of work, and you build a cryptocurrency around it, then the value

00:41of the currency should be zero, because as everybody knows, the value of a proof of work currency should be proportional to the effort and the cost it takes to mine it, but if I could also reuse the work for something else, if the work is useful, then I can sell that work, and that effectively makes the cost of mining zero, and therefore the value of the currency zero. So that seems like a pretty bizarre paradox that the better the technology becomes, the less

01:14the currency should be worth. That's sometimes called the proof of use work dilemma, and it seems like a pretty fundamental paradox. I like paradoxes, so this intrigued me a lot, and as with all the best paradoxes, I hope to convince you that once you look at it the right way, the paradox disappears, so that's what I will talk about, but before telling you how the paradox disappears, let's just briefly

01:43recall the problem we're trying to solve, that of building a blockchain, a secure blockchain. So as you heard from Milan and from Omri, a blockchain is just a way to implement a trusted public ledger, and this ledger should be robust to a single point of failure. So how do we achieve something that's robust to a single point of failure? Well, there is a standard approach, if you want to be robust, just replicate all your data, right, trivially just replicate your data, a bunch of different servers, and then

02:15if someone can go down, we still have the data. Of course, once some of them go down or some of them broken into, we don't really know which is the right one anymore, that's always the problem with replication, right, how do we find out which of the replicas are actually correct, and of course, there are standard methods we just vote, right, and we say that if a majority of them are still fine, then the majority vote will be correct, and we can recover the right result, right, very easy.

02:47So we'd like to have a system that is robust to faults and attacks as long as 50% of the replicas are okay. Now as Omri beautifully explained, there is a problem with voting. It seems trivially, you know, correct, but if you want to have an open system, a fully permissionless system, then an attacker can just spawn lots of different nodes, and votes are really not a very reliable approach, right, and that's why in normal election,

03:18once we take off everybody, right, when you come and vote, we take off your name, we check your ID to make sure people are only voting once, but in a permissionless system, we can't do that, and that's where Nakamoto's brilliant idea behind Bitcoin came in, and Nakamoto said, look, let's not count votes from people, let's count votes from computations, right, so each time you spend some computation, you get a vote, and this is what's referred to as a proof of work.

03:48So you're expanding some computation, every computation gives you a vote, and now if we assume that 50% of the computation in the world is being honest and not broken into, we can actually implement this majority of voting in a correct way, okay. Now this proof of work is often called mining because now people are in order to vote, need to mine, do this computation in order to get votes, okay, and this mining is implemented by, as we heard, just having people trying to solve useless puzzles, so they're trying

04:23to find useless puzzles, and once you've found a solution to a puzzle, that's going to be your proof of work, you're proving to the world that you have expanded a certain amount of computation, okay. Brilliant idea, and in such a scenario, we can now get a notion of security, which hasn't been said to break the system, you need to control more than 50% of the computational power in the network, okay.

04:52This intuitive sense correct, it can also be proven, it's actually quite non-trivial, and in a result from roughly 10 years ago, we managed to prove that, indeed, Nakamoto's protocol is secure as long as the attacker controls less than 50% of the computational effort, okay. Great. Now, you might get to the question, why is it the case that people actually doing this computation?

05:17There's a lot of computation being spent to make the network secure, but why would the honest people want to do it, the attacker maybe wants to do it in order to break the security, but why do good people spend all this computation, and that comes from the fact that this brilliant protocol is designed so that every time that you have expanded a certain amount of computation, you actually get some block reward, so you're getting paid for doing the computation, okay. And that's really the key economic insight behind, like modus billion protocol.

05:49So you're getting some bitcoins, every time you solve this proof of work, okay, today we get three bitcoins roughly, and that's called the block reward, and I'm going to refer to that as all. R is three, in parallel R is around 2.7 thousand at the moment, okay. And the block reward is how many bitcoins you get, and then if you want to translate into the dollars, you need to multiply by the price of the, of the Bitcoin, right.

06:15So the block reward in dollars is P times R, and now for this to be meaningful, for honest people to want to actually run this protocol, we need to make sure that the block reward P times R should be at least as high as the cost of mining, otherwise nobody wants to do it, right, because mine cost a lot of electricity in the GPUs, so I need to be able to get rewards that are higher than that, make sense? Okay, so these two things together give you a very nice notion of economic security.

06:49In order to break the protocol, you need to control more than 50% of the computation spent to keep it secure, okay, and that in particular means that in order to break the amount of money you need to spend needs to be at least the block reward divided by 2, because with the block reward needs to be at least as much as the cost of keeping it, of running those computers. All right, to summarize, in Bitcoin, mining requires massive amounts of useless work, we're

07:21just solving these crazy puzzles, okay. But the intention actually is that the uselessness of this work is that makes the system secure. To break the system, you need to spend 50% of that computation, okay, and that in turn needs to be very expensive, okay, so you need to break, you need to wait a lot of resources. Okay, enters proof for useful work. So can we create proofs of work that are not just being completely useless, but are actually

07:55doing something meaningful? The pro break-through that you heard Omri and Ilan talk about shows that actually surprisingly it is possible, and it is possible to actually get proofs of work for not just crazy hashing, but for matrix multiplication, the very computation that really fuels the whole evolution. So now there's no need for wasted computation, we just take those proofs of useful work, plug it into Nakamoto's blockchain protocol, and we get a secure blockchain.

08:27Without wasting computation, awesome, but does it work? In fact, if you go back to the previous slide and recall why did I say that this was secure, the argument was that to break security, you need to waste a lot of resources. But if the resources actually are free because they're being used for something else, then it might not be costly, and security is gone, right, and indeed this is a fault-clear argument, it's been used in the literature.

09:00That says that proof-of-use work is fantastic as a conceptual, philosophical idea, but maybe it's not the right tool for security and blockchain. So let me go over this argument and to just spoil it already, we'll see that this argument actually is flawed, okay? But let me go over it first and we'll see if you can find the mistake in it. So recall, in Bitcoin, in order to break security, this attacker needs to spend a lot

09:27of computation, more than 50% of the computation is secure in the network. And recall, that computation needs to be expensive, it needs to be at least the block reward divided by two, that's the cost of it, or rent in those computers, okay? So to break security, you need to spend a lot of money. Now, let's consider a scenario where we have a proof-of-use work instead. This attacker still needs to rent all his GPUs in order to break security, right, it needs

09:52to still control 50% of the network. But attacker's not just breaking the network, he does all his useful work, he goes out to the market and sells it, right, this useful work can now be done to, can now also provide AI inference and he can go and just sell it to the market. What's the price he gets? Well, he was going to get the price that is like the current market price for compute

10:18and that's exactly what he's spending to rent those GPUs. In total, what's the cost of this attack? Anyone? Zero, right, he runs computers and he uses those computers in order to provide AI workloads. So at least if you have a proof-of-use work with no overhead, right, the optimal proof-use work, it gives you a fully insecure system, pretty convincing, right?

10:49So we'll see today that this argument actually is flawed, okay? So to explain why, let me start by an analogy and then we'll jump into why this analogy applies in this context also, okay? And this analogy is going to be, we're going in plain chess. So let's say I'm going to go and play chess against a chess master. I'm okay at chess, but I can't compete with a chess master.

11:15So I go and play one guy in the morning, yeah, I lose, okay? I feel emboldened and I decide let me go and play another guy in the evening, I will lose again, okay? It doesn't work. Next day, I decide I'm going to play them at the same time, okay? I'm not going to play them isolation.

11:38I'm going to let these two games be happening at the same time. That seems to make it harder for me to win, but actually not. When this guy makes a move, I go to this game and I do the same move. Then when this guy makes a move, I go here and I make the same move. And I'm affecting making these two guys play against one another. And what happens, I win at least one of them or get a draw in both.

12:06So suddenly, I managed to do something that was impossible in isolation, right? So what's the lesson learned here? The lesson learned is that actually we cannot be treating economic systems in isolation. That's what the flawed strawman argument did. In fact, when you have two economic systems and you have some action that can operate them both at the same time, which is indeed what this two for one action at polls has created

12:36is the way. That creates some interactions between both these markets and that leads to amazing things to happen. We cannot just think of them separately. And indeed, let's look at what happens once you think about these two markets, not separately, but allowing some interaction between them.

12:58Let's go back to this attacker. We said he's spending all his computation, he's generating a computer to break the network, and he's going to sell it. Now this analysis thinks about the two markets being completed, he's doing his stuff here, and then there's an inference market that is pricing inference, but that's wrong. If he's making use of it, then so should the good guys producing inference be doing.

13:30Inference providers should also be using a two for one operation, right? Just as he's doing it, so should they. So they should actually be mining in an efficient economy, and they are getting now pearls to subsidize their inference. What that means is that, in fact, the price of inference should decrease. How much will it decrease by?

13:55It will decrease by exactly the block reward in equilibrium. So what happens here is that when he's going out there to sell the useful work that he created, he's not going to get the full production cost of that inference, he's going to sell it cheaper, he's going to sell it for under-production cost, because these guys are actually receiving pearls that are bringing down the price of inference. So in order for him to actually do this attack and sell it, he can only sell it for the discounted

14:31price, so he still needs to pay the rebate. And it turns out that if you do the math, the rebate is going to be exactly the block reward. So in summary, in fact, the price of attack turns out to be exactly the same as in Bitcoin. It's going to be the block reward divided by two, exactly the same numbers in Bitcoin. It turns out that actually, as you'll see in the full economic model, the block reward is going to be at least as high as that in Bitcoin, and sometimes even higher, so the cost

15:01of attack will actually be even higher, in some cases, in Bitcoin, but never lower. So to summarize, AI providers are now getting tokens whenever they provide AI compute. Those tokens enable them to now undercut the current price, and in a free market, that means that the price of inference is going to go down. And that means that an attacker now needs to still pay this discount on inference to attack the system, and that discount is going to be exactly the same as, in fact, the block reward

15:32before. All right, so to forenose all of this, we need to have an economic model of proof of user to work, and that's something that we've been working on. And in roughly a week or so, we will release a paper on this. And in essence, the very formalized thing is by considering two different markets, a security market, and this is a market for, a mining market for maintaining a blockchain.

15:58And a second one, an AI market, or an inference market, because we have these two markets. And then we consider a bunch of players that are just compute operators, GPUs. A player can choose between three actions. They can choose to just mine, to just provide their compute in the, in the blockchain. This is just like going in mining Bitcoin today. Or they can do the max in L for learn.

16:27They can just provide user GPU for product inference. That's a classic world, or they can do this thing referred to as a duplex operation. The duplex operation is a joint, a two for one operation. It participates in both of these two markets at the same time. It co-producers both mining and inference, but maybe with some overhead. So it doesn't get as much as if you just do this, and they're separately, but you can

16:55do them at the same time. And the question is what happens in such a scenario when considered these two joint markets with this action that connects them. And the main results of the paper are going to be, number one, the cost of attack is actually at least as high as in a Bitcoin style world that we refer to as Bitcoinia. And the main, second main result is that we can actually get a complete characterization

17:20of the equilibria in such a scenario as a function of the duplex overhead. So without going to the formulas, let me just kind of highlight what happens in this scenario. In a scenario where you have very good low overhead, what the economic effects of a two for one or duplex operation that enables doing both of these two mining and inference at the same time with low overhead. In that scenario, the blocker words that are being created in a proof-of-work blockchain

17:56are subsidizing AI workloads, and that is always going to lead to lower prices for inference. Now under so-called elastic demand for inference, or something called the Jevons effect, that will actually lead to not just more inference, but even a larger inference market. Let me just mention a word about the so-called Jevons effect. Many of you have probably heard about it. The Jevons effect, in essence, says that once you improve the efficiency of a resource,

18:31that leads to not just, it actually leads to a larger total consumption of the resource. So this is something Jevons already noticed in 1865, that once steam engines became more efficient, people thought, now the price of oil is going to go down. And instead, the opposite happened. There were more need for oil prices going up. Why would that happen?

18:55Well, the conclusion was that once engines became more efficient, suddenly that unlocked new use cases for steam engines, and in fact, we needed even more oil. And the same is expected to be true for the AI market. Once the price of inference becomes lower, that will actually make the size of the market grow because it unlocks new use cases. Okay.

19:19So let's jump in for a few minutes. Let's spend roughly five minutes and start to give you some details of the model. So excuse me a little bit for, I'm going to give you some equations here on the board, but hopefully, you'll be with me for a few minutes. To formalize that the full model, let's first study a pre-poil world. A world where there is no two for one yet, no duplex operations.

19:44So what does the world like that look like? Okay. A bit-constant market where people can do what are referred to as solo mining, they're just doing mining, okay? And that is going to produce some crypto tokens. When I say tokens today, that refers to crypto tokens.

20:01And let us the note of the number of people that are actually doing that. And then we have a separate machine learning market, an AI market where people are playing this action L, and that is producing some useful work, okay, just inference or AI. These operations both have a cost, it's the same cost E, E refers to the cost of running a GPU for 10 minutes, okay? So we're thinking of block times of 10 minutes, E is the cost of running a GPU for 10 minutes.

20:31What are these markets? What do these markets look like? So this is now pre-poil, this is, you know, we have a Bitcoin market and we have an inference market. These are well-established and well-studied markets in the ecosystem. Let's look at the Bitcoin market, okay?

20:45The Bitcoin market is something called a Talok market or a cake cutting market, okay? In Bitcoin, we're getting a block reward every 10 minutes, so we're getting these P times R dollars, let's say, whatever it was, $3 times 80,000, $20,000, $40,000, 10 minutes is being created, okay? And we're doing a cake cutting contest. So that means people can enter, any GPUs in the world can come and enter.

21:12And the more people are entering, the smaller part of the pie I'm getting. So that's what this picture shows here. So the reward I'm getting by entering this is going to be P times R, the block reward, divided by S, the number of GPUs we're entering, minus the cost for me to run a GPU, okay? And now we might ask, what's going to happen in such a market? What's going to happen is that if this number is positive, then more people are going to enter,

21:40right? If I'm making more money than the cost of producing it, more people are entering, and that is eventually driving up the number of people here to become exactly so that P R over S equals E. And that's what's going to happen in the equilibrium, great. So very easy way to solve what happens in a Bitcoin world. Everyone with me?

22:02Yeah, great. Now we move on to the inference market. This was even easier. That's something called a classic betron market, okay? This may be one of the most classic markets in the literature. So we have a demand curve that specifies for a certain price, what is the amount of compute

22:23that the market demands if the compute is being sold at a certain price P, okay? And it's very easy to see here that in such a market, the price eventually is going to converge down to exactly E, the cost of producing it. Because if I'm selling it for $10, but the actual production cost is $5, then it will be another guy that will sell it for $9. And then I will sell for $8, and we go down until we're actually going to end up at exactly

22:53the price of production, okay? So the reward that GPU is getting is the price we're selling it for minus E, the cost of producing it. And in equilibrium, P is going to be equal to E. And in that case, I know exactly how much inference has been produced, that's just this demand curve of E. Very easy, okay? This is the people world, okay?

23:17I want you to think of this mining operation as a fork. This fork is, I'm using it to dig into the ground, to dig for some crypto tokens, okay? And I want you to think of learning as a spoon that allows me to scoop inference workloads, okay? And now enters this duplex operation, this two for one. And that's like a spork, okay?

23:40This allows me to dig in the ground for coins, and allows me to scoop AI workloads also. But it's not as efficient during either this task. There is some overhead here, okay? I'm going to refer to this overhead as alpha and gamma. So alpha is the overhead with respect to how much worse duplexes are doing machine learning. And gamma is how much worse it is at mining.

24:02Right? And in the construction that Omre and Ilan showed, notice that you had to add some, in order to do machine learning loads, you need to add some extra operations. That you could strip away a little bit of overhead, okay? And the same way, if you only wanted to do mining, you could have started off with zero old zero matrices, and that would make the computation a little bit easier.

24:23So there are some overheads involved in doing this, okay? And that's what alpha and gamma are correspond to. So mathematically, using this duplex operation, it's going to correspond to doing a one over alpha of a learning operation, and a one over gamma of a mining operation, okay? So I'm getting that as same output as doing that, but at the price of only one compute. So for example, if you think of alpha and gamma, these are just examples to make the numbers

24:57easy. If you think of them as being 1.3, so you have 30% overhead. That means that doing this duplex operation is going to give you 80% of what you want to do just pure machine learning, and 80% of doing pure mining, okay? So you're effectively getting 1.6 for one, makes sense? You refer to as a proof of use of work as being non-trivial, if the sum of these things

25:24is more than one. You're getting more than one for one, all right? You're getting even 1.1 would be 1.1 for one, that's only again, you're doing two more than one at the same time, okay? And now we can set up these reward equations taking into account these duplex. So in this, the fork here, the mining operation, as I said, the reward I'm getting from mining

25:48is just PR over S minus E, that's from before the reward of selling inferences, the price of the inference minus E, okay, as before. And now we have duplex, and duplex is going to, whenever you use a duplex, I'm going to get some of the inference that I'm producing, so I'm going to get the price, but divided by alpha because there's overhead. And then I'm going to get some of this thing, the PR over S, but divided by gamma, okay, and

26:17then I only have to pay E, okay? So now I can just set up, what are the prices, what are the rewards that people are getting? And then in equilibrium, as we said, nobody should be able to make any money because if people are able to make money, then more people are entering to drive down the prices. In equilibrium, none of these things should be more than zero, and furthermore, if some action is active, if I'm actually doing something, I'm not making negative profit, nobody wants

26:44to do something that you lose money. So those are the equilibrium conditions, these are classic equilibrium conditions. And other questions, we have these conditions, let's just solve what equilibrium is. And that's typically a non-trivial thing, because now, you know, typically people analyze things by saying let's consider them separately, but we can't do that now, I mean, that was the flaw for the previous argument that people said like, you know, let's assume that this

27:08mark is operating independently, it's not, we need to actually, in fact, consider the con joint production, that we're producing things in two markets at the same time. So the main result that we show in the paper is that there exists, in fact, a perfect characterization of equilibrium. There is a one-dimensional characterization, and there's a single parameter that we refer to as the TI ratio, token inference ratio, and here token means crypto token.

27:36This theta, think of it as the, or think of it, it is, it's the ratio between the size of the cryptocurrency market versus the size of the inverse market. Once you tell me the ratio between these two sizes, that perfectly determines the equilibrium, as a function of the, overheads of the, of the primitive. So without going into the math, let me show through a, through a graph, what happens. In fact, we get a few different interesting phase transitions.

28:10So here's a graph that shows how compute changes as a function of the efficiency or the overhead of the duplex operation, the two-front operation. So here we're starting with something that has low efficiency. In that case, we're in a world where we refer to as bitconia. In bitconia, that's the pre-poly world. People are doing pure mining, here, this is wasted work, they're just doing mining.

28:39And here people are doing inference, okay, everything is completely separate. Now once the efficiency of the two for one becomes, so nobody's using this two for one. As if it wasn't there. Now once the efficiency becomes better, we enter fortasia. In fortasia, nobody's doing any, any pure mining anymore, pure mining disappeared. And instead of the impure mining, people using the pearl two for one operation instead, starting

29:08to use duplex. So duplex still has overhead, it's not a no overhead here. And here we see that in fortasia, this light blue area here corresponds to the amount of compute that is actually not useful, because of the overhead. So the overhead here is actually the same size as a bitcon. And indeed, the price of the coin is going to be the same.

29:35And the amount of inference delivered is actually also the same. So in terms of an economic market, it's not changing much. The price is the same as a bitcon, the amount of inference produced in the world is the same as a bitcon. But there is some changes, you see this graph going up here. So what is this?

29:52This is the amount of computers, amount of GPUs that are entering now. And the amount of GPUs that are participating is much higher. So what that means is that in order to break security, you need to actually break into many, many more GPUs. That's what we call this fortasia. Security has been fortified.

30:12So the economically doesn't change much, but in terms of security, it's much stronger. You need to break into many, many more computers to break into. Things become really interesting once the efficiency becomes even better when you have low overhead. So once the overhead is sufficiently low, we enter duplexia. And in duplexia, crazy things start happening. In duplexia, we see here that the amount of inference provided in the world actually increases.

30:40The token price increases, but even more so. So this is the amount of inference being provided, how much it increases. But the green here is the value of the inference. And that's even higher than the amount of inference increase. And the reason for that is because we're providing a discount on inference. So in fact, we're actually producing even more value than just the increase of inference.

31:07So to summarize, despite this common strawman approach, proof-of-use work does not make attacks cheaper. In fact, they're at least as expensive as before, and sometimes even more expensive. Second of all, on the good conditions, so if you have a two-for-one operation like Paul that has very low overhead, in fact, you're going to get a blockchain that increases the size of inference in the market.

31:35And this follows from the fact that tokens are subsidizing AI workloads. And that, on the Jevons effect, leads to even bigger market size. And that, in turn, leads to a cryptocurrency that's worth even more. And we get into this virtuous cycle where the blockchain actually is subsidizing AI, and at least a higher valuation, and a bigger substance and so on and so forth. So in essence, a good proof-of-use work blockchain leads to more social-evaluable computation in

32:08the world. That's a pretty bizarre statement, right? So we're adding this blockchain to the world, and suddenly we have more useful computation happening. Who's paying for that? Somebody's to be paying for this extra computation.

32:25And if you think about it for a minute, you realize that, actually, it's not so bizarre. This is, in fact, something called senior rush. Since we're releasing new tokens, inflationary emission tokens, those tokens are actually paying for all these new AI workloads. And this is not a new phenomenon. That happens already in Bitcoin, in fact, even the US government releases new money.

32:47But in Bitcoin, this release of new tokens, what is it used for? What is it paying for? But what is it paying for? It's paying for burning computation. Nothing. It's just paying for the burning computation.

33:05Here, it's paying for new AI workloads. And that's the amazing aspect here. So in essence, a proof-of-use work chain becomes a financing vehicle for AI workloads. As a consequence, that means that if you get a blockchain that has higher adoption or better efficiency, that leads to a bigger AI market. That means that AI provides actually incentivized to ensure higher token adoption better technologies,

33:38because that leads to bigger subsidies and, indeed, a bigger AI market and ultimately to more social useful AI computations. And that's it. Thank you.

00:00implemented on top of mining pools and it's also connected to this question around like smaller cheaper models. Our whole idea is sort of to be like have all the work be like game theoretically useful in the sense that obviously you could run a cheap model or a less useful model and you'll be able to mine a lot but you won't be making any money from AI so like long-term economic wise you won't be able to be profitable and that's sort of the idea with the protocol. Yeah then this is also where the overhead biggest challenge and useful proof of work

00:36this is where it really enters the picture because you know no one would be you know you have to make the gap of efficiency the co-production of pearls for or useful miners should be close to zero or basically or you know much cheaper no one would opt in if they knew that multiplying diagonal matrices or all zero matrices is 10 times as fast as or even three times as fast as multiplying real dense math moles which is what you typically have in you know in transformers or you know in most layers.

01:27And on the latency in overhead so I think the overhead and 10 on those JAMA models served on together AI and soon on the more hyperscalers that's going to be roughly five percent and to end overhead between five and 10 percent. I think that's something we insinuated also on the white paper quite explicitly that we are going to release the current protocol supports only into eight precision which is something very few models actually run on today that was on purpose partly in order to prevent you know all of the AI workloads

02:12you know hovering over the network at once so I think that was probably instead a smart move in the next you know as we run the white paper in the next few months we're going to release blackwell kernels basically you know kernels that are going to support floating point for fb4 workloads and that scheme is going to be actually it's going to be a hard fork so it's going to be a new scheme like it's a new protocol it's very similar to the current one but it actually has some it removes it slices the overhead it supports floating point and it's going to be

02:50much easier to like plug in kind of plug in replacement for for purlifying AI models and there the overhead is going to be literally on like negligible so you're not going to feel it and that adds to Don's like point on like the reason why of course you can multiply garbage matrices on pearl which we obviously know that you can fill bubbles and do whatever like most you know most people would think but of course the cost of production is the best for people who already have useful matrices which are

03:28matrices that someone some external entity paid for their multiplication right and that's that's pearls whole thesis that the GPU cycles were already paid for by those hyperscalers model companies and consumers running you know chatbots robots AI agents and so forth and and now you can bootstrap and and and get a cheaper cheaper cost per you know cost per lm token crazy bird is asking what does an AI company gain from my pearl aside from reducing operational costs so I think that at this point there are a few obvious things that an AI company could gain I don't know what

04:19will happen in the future as the coin and the ecosystem evolves but right now one thing is a reducing operational cost which is an already a non-trivial thing to do at almost zero extra cost for lift that's one thing the other thing is they get to draw more customers because if an AI company imagine a scenario where an AI company mines prl pearl on their balance sheet while doing useful work and it doesn't keep everything to itself it also distributes a portion to the users wouldn't all of us want to be customers of such an AI company where we we supply the

05:04prompts we help their models train we help their models become better and we also get some kind of reward for that in the form of of the token so so I think in the in the optimistic scenario where this is adopted by all the companies this is something that's bound to happen and then AI companies would just users which is the most important thing for them even more than reducing operational cost yeah it's another point to mention that block like if you think of prl as not just as AI money which we unapologetically try to create right so that's why we

05:49for Bitcoin Bitcoin we're not going over the hyper tokens per second like you know payments perps bridges performance like we're not we're not Solana we're not here in right we're going for the like the Bitcoin of the AI era so we're trying to create AI native money and money that machines can can use most notably like you know AI systems and agents there are like if you believe you know enterprise scale adoption of this chain which is the real unique distinguisher of pearl from pretty much any other crypto project in the world right

06:31like open AI and anthropic can conceivably mine you know on their balance sheet any other AI crypto project you have in mind so I think that's the unique angle of pearl but if you think of and believe enterprise scale adoption then no pearls just you can view it as just a non sovereign like decentralized database that could potentially contain an inference trace and zk of course without privacy leakage but it can it can contain like encrypted inference trace of all the AI agents or all the AI systems in the world and

07:14you know it's it's you can imagine unique applications that can emerge only on such you know model tracing which is the problem of like you know you know you know proving that you know on December 14th 2025 and you know your AI surgeon ran deep seek R1 and not when 14b right that's these are compliance or like settlement queries that are getting more and more popular and important so I think that's a non sovereign you know open voting system for AI agents so my blockchains are ultimately a voting system and

08:04think of it that now AI agents can just have a native voting system which is permissionless and open to to um any um to any company to any system I don't see um I I see a question about incentivizing actual AI workloads and then random from tom random matrices are almost 10 six to 10x more hash rate the incentive is pretty clear like if you're doing an if you're an AI company and you're doing inference anyway the additional cost of of doing this proof of useful work and earning coins is is going to be negligible for you versus someone that is doing use you know

09:00random matrices they have to fund the entire operation themselves from the from and hope that multiplying the random matrices is is is is going to make them enough money which which won't be possible as as long as compared to someone doing useful work will just be way way way more profitable um so that's that's the idea there yeah uh I see a question by um on the plans for the compute marketplace uh which is a compute platform that we're running um I think it's it's it's pretty clear that these things are gonna diminish once um

09:40um you know public listing will will occur and and we're actually we wanted the compute marketplace to exist uh in order to maximize distribution and give unlike bitcoin where you had to be this like you know mit nerd with an hbc in the basement of some you know uh hbc lab um to get early exposure here for the ai era you know we we we believe that we should give people um you know just normies uh access to to uh this asset uh as early as possible uh that's why we fought very hard to have credit card swipes uh available I think it's the only proof of work project that currently

10:29has integration to credit card swiping with with stripe um but the end game for the compute platform is really just revamping to start serving real inference workloads so pearl compute or the pearl cloud is gonna basically become like a essentially an inference provider like a neocloud that's gonna serve whose claim to fame is hopefully going to be serving uh state of art llms with the cheapest um right doing the two for one and that's that's very natural right because a lot of the gpu's that um are incentivized to mine the network they don't have the useful work they

11:12don't know how source demand for useful inference and you know um the natural thing to do is to um build like a either a marketplace or a native just a cloud inference service which serves our own prolified models so we will take um you know we're working now on glm which is like the one base model for which is trying to compete uh with appropriate harnesses trying to compete with with claude for her and friends so we're gonna take those open source models we're gonna prolify them which means that we're we're gonna replace all the linear layers with the pearl protocol

11:53so running those models is gonna do the you know it's gonna do the two for one and um and we're gonna of course uh um redistribute uh uh either discounts or pearls or a combination of both back to the users and again the claim to fame of this neocloud will be just being the um um providing the the the cheapest inference and ultimately will also pearls uh for compute right currently what we know how to do is turning compute into pearls this is what mining the chain is doing the end game of pearl is is in the real fly will it will become

12:36where hopefully we can do the other way around when we can turn not just compute to pearls but pearls into compute so having a inference service that accepts pearls for inference that that's going to be the real flywheel and what really we're trying to create which is being able to trade like in the same vein that bitcoin is used to trade energy outside the dollar system pearl can be used to trade you know ai compute from an inference outside the the the dollar system in a much more seamless and you know uh much more efficient way financially

13:15yeah and i want i want to address some stuff people are asking around around like getting exposure or useful work it's there we're in the current mode i think um especially especially when it comes to the concept really the idea is to let non-technical people get access and to the network and assist in bootstrapping it um we're already running useful workloads as part of that in the sense that we're running ai workloads and doing inference and we're going to work and we're scaling that up rapidly as we're able to scale that up the economics for everyone participating can

13:48get a lot better and a lot of us also has to do with compute availability as as we've all seen especially in the past few weeks perhaps because of pearl compute prices have skyrocketed and it's been hard to get on demand compute so these these are things that we're that we're working on uh very hard right now to try and get uh useful work like exposure to consumers that don't necessarily have it obviously the way to do that is to have a little bit of a longer term view and there's things that we're working on now which will hopefully be able to talk about

14:18more uh in the coming weeks so i see pistachio is asking um can you mimic useful ai workloads and mining what determines your uniqueness so i want to be clear what like a me a pair of matrices a and b are you know x useful if someone is someone external is willing to pay x dollars for their multiplication that's a definition of a useful work so useful work is not defined by uniqueness not some information theoretic property it's uh matrices are useful if someone is willing to pay for their multiplication obviously um you can't resubmit work for you know you can't

15:03resubmit a pair of matrices because no one is going to presumably want to pay for a multiplication that he or she already knows the answer for so i think it's important to define mathematically what useful means which is just um the usefulness is determined by by by by cost i hope it makes sense um the latest question asks about latency where that's that's introduced by running an ai workload in a distributed network and so on so forth this is a common misunderstanding or misconception of what pearl is pearl is not a marketplace or not or not uh distributed system for running ai workloads

15:49it's uh it's a blockchain where miners do their own useful work so they're not going and sourcing miners are not sourcing useful work of somebody else and so there's no communication privacy issues miners are the ones who are already doing the useful work or non-useful work if they wish to and so and this is i think the biggest distinction between our blockchain and many and many of the marketplaces of the gpu marketplaces that you are or that use tokens to incentivize gpu's to participate in the network

16:34yeah i think like going back there's a bunch of questions i think from other like from other from various users how we compare to other other things out there that are currently i think but most of the ones that have been mentioned are just are variations on on on gpu marketplaces or clouds that maybe have a token involved or maybe do some payments on chain but ultimately um they're uh just like tokens maybe revenue tokens uh they're not they're not created they're secured by by matrix multiplication and by AI workloads

17:10they're generally a token living on another blockchain or maybe they're their own proof of stake style blockchain there's a question about an async being a hundred times more efficient than the gpu and whether this breaks the security or something like that and i i'm not sure i mean this number is correct because the gpu is already essentially an async for matrix multiplications it does also other things it can also do general compute and so on but just in terms of of of matrix multiplication i think i'm not sure that an async can be a

17:48hundred times faster than a gpu but irrespective of the of that even if an async is some somewhat more efficient and you lose the ability to do useful work and so you might be x times faster in finding a block in the blockchain but you can't earn money from doing your useful work you can't be an AI company if you do that so in some sense we the the protocol as we designed it is in some sense asic resistant from from a game theoretic perspective it's not it's not asic resistant because it's impossible to build an async it's just asic resistant because it's not

18:29worthwhile for anybody to design an async that loses the ability to do useful work that makes sense yeah and regarding dizzles question about new super fast chips for inference so they're already on our radar again as long as they do math moles you know there are certain chips that are just you know tallas or you know or wafer like there are cerebras which are you know are doing things sometimes which are relevant you know only to a specific like you know you know attention

19:10mechanism or computation or decoding phase or whatever but as long as they support general math moles they're definitely they can you know they can we're planning to add support and in fact we're already in touch with at least two hardware like major hardware companies that express interest in integrating perl the perl protocol into their design and into their spec yeah but again like to eons point like if you can do general math moles which is what at least most of the you know moe pre-fill feed forward you know

19:57layers and lm's require if you managed to do this substantially faster than nvidia then you know then gpu's then you know you know we're in in some sense we are relying on nvidia being the core the main you know core essentially the asic for math moles which is what you on said so and then in terms of the other questions there we are talking to the different ai asics so i think for us in when when you know anyone that's creating an asic that's would even do mining and ai and parallel most efficiently that's that's like fair game

20:40that's the purpose if when you guys are referring to asics we're talking about an asic that would just be doing a random matrix multiplication and able to mine more profitable than someone doing ai workloads if someone's able to create such a thing then you know that's um uh that's that's that's that would be problematic but um that's from from what we've looked into that doesn't seem um like like a realistic concern well would be the hardest part of getting large compute owners and labs to buy in so and so ai you know infer like the the ai companies are generally divided into two are the closed

21:25source and open source obviously it's much easier to start with the open source models because there we can just prolify the models you know all the gemmas and quans and kimmy's and you know eapsie those are the models we're already working on and that we've mentioned part of them are already served on together ai and they're soon going to be served on uh a couple other like top tier hyperscalers obviously the long haul like most of the inference in the world i would say probably done correctly for probably probably like 80 at least 80 percent of the work

22:04um is closed source inference right it's the cursors of the world and the meta and open ai and i'm you know claud is a closed source model right we can't just prolify it we need we would need the actual model companies and the ai labs to um let us in and do this like surgical operation of basically prolifying their their their model of course pearl is open source so they could in principle do this themselves with the next scheme i i describe the next protocol upgrade later this uh summer or fall that then you know that's going to be almost seamless so

22:44which is we view this as a good thing so adoption would be by the major model companies will be way way simpler than today today just getting low overhead is still requires a lot of optimizations and quantization like a lot of things that are model specific with the new scheme it's going to be easier uh but i think the the hardest challenge is this is getting um the frost and abortion span from the biggest companies in the world you know anthropic to for anthropic to to be able to just open the kimono let us in just you know prolify their models um we believe this this

23:26will require um it's it's it's not so much about um maybe about um uh um the side of the network or the valuation i think it's more the thing that our thesis is that what will force the big labs to dopril is consumer pressures just market dynamics so um if you think about it open ai they do own the models and maybe even the hardware even though they don't but you know let's say they own the model and the hardware and the um the weights and and the compute and they pay for it but they're not the producer of the useful work right who actually owns

24:08the useful work when when walmart is doing an api call to claude um walmart's customers let's say us right we we're the ones that are basically doing the um data labeling the prompting we drive the demand with all the reinforcement learning so we're providing that the users are actually providing the useful work uh the useful prompts and matrices and sooner or later our thesis is that consumers are going to realize that they earn a stake or a piece of the pearls produced by those ai inference workloads and so consumer pressure will ultimately

24:46force those labs to opt in and serve those tokens um back to the users where do we see the protocol in the year um so maybe the right question where do we see the chain in the year but let me try to answer both at least from my perspective and i'm reading it down can't say what they think so in terms of the protocol uh we we mentioned a couple of times that we we have a new version that's going to support seamlessly every model not require anybody to quantize it it will support a floating point eight or four and natively without any intervention essentially the overhead is also

25:33going to be much smaller than it is today um so that that's what we think will be really a a knockdown to to to the to the dream of useful work and that's in terms of the protocol in terms of the network well and the general feedback that we've received so far from any companies is nothing but positive they're all eager to to to try it out to integrate and play around with it hopefully a year from now it's it's something standard that you don't need our plugin to vlm it just comes with vlm and when you download vlm from the official website you just get it

26:20with with pearl mining plugins that's the dream yeah so um i see an earlier question but by tom how do we plant incentivize actual ai workloads random matrices are almost six to 10x more hash rate so again i i want to be like the whole premise here is that if the overhead were literally zero right or let's say one percent i hope we all agree that it doesn't matter even if um diagonal or all zero mat math moles on the chain are even 50 times faster right because the point is that at zero overhead the cost of opting in

27:07is is bit is essentially zero um forgetting forgetting the you know the human integration overhead but um at this point the a the useful ai player it's um all we care about is incentivizing and throbbing and opening out all the you know inference workloads to occur on pearl so at zero overhead it doesn't matter that someone is multiplying random matrices or all zeros matrices faster even 50 times faster right they still have the incentive to opt in i hope it makes sense that we're making this argument precise in the in the paper

27:47mathematically you can show that um amount of speed up we can allow like from a game theoretic perspective like elon said is roughly one over the overhead so you know if the overhead is is five percent and you can allow a speed up of of 20 roughly for non-useful miners and you can still your all we care about is just the cost of production of pearls at zero overhead the cost of production of pearls for useful miners is literally zero so in this case i hope we all agree that um it doesn't matter how much faster useless mining occurs the cost of production of useful

28:30miners will always be cheaper and now what happens if three three percent overhead five percent overhead which is what we're realistically targeting in the next upgrade um that's somewhere in this ballpark so that's that that's exactly why the the um and i think i mean that's that's why we're doing our best to optimize the overhead why it's really crucial and that's the biggest challenge and proof of useful work as to james uh question about speaking on specific uh next like ai companies that are planned to onboard let us not mention specific names i will say um we're

29:11already in touch with like three in like advanced progress with um um two other two out of like hyperscalers uh in the world one of them is hopefully going to be signed um this week we're in touch with two other major hardware companies who are very bullish on this one of them hopefully you'll see a joint blog post in in about a month or so um and um that that's going to be a major thing that we never expected to happen that early uh in the chain uh neolab like ai companies or model labs which are the the like closed source companies which are the hardest to

29:57penetrate uh so we have already two of two major neolabs in the battler i would say three even in the leo in the valley that are um waiting to integrate the protocol most of that like we're already in touch with about five of them three of them are waiting for blackwell kernels which are going to be released in a few months um but um i think that's a very strong sign because these again these are not open source models so they're literally willing to integrate those protocol into their you know those source ai companies or model companies

30:40they're extremely sensitive to their you know this is their holy grail right their ways the models the architecture and the fact that they're willing to um take a bet on this network with it just you know being as far from crypto as you can imagine i think it's a it's a big sign that this is something they see here's something different if the cost of pearl production is zero who will support the cell pressure why would anybody buy pearl if the cost is zero so that's uh that's something that we need to clarify i think if the overhead in compute

31:25is zero it doesn't mean that the cost of production is zero because miners that don't have useful work that somebody else is paying for their cost of production is not zero and there will will always be such layers in the game if uh a player a miner has an infinite amount of useful work and the cost of production is zero then yeah that miner is producing pearls for free that's essentially equivalent to some to a bitcoin miner that has a free electricity for instance but it's it doesn't make sense that

32:02something like that will happen in some sense you can think about it as the real the real asset that miners should have as opposed to bitcoin where it's only cheap electricity here it's cheap electricity and also data or essentially consumers that are willing to pay for a service so even if there's a company that has a cheap electricity and they still need to spend money on procuring users who will buy their AI service and this is a resource that's not free and so the cost of production is never free for anybody that makes sense

32:52how will you be able to spend pearls in the future so we we think that pearl is we're trying to close the circle so AI creates pearls we our vision is that pearls will also be able to drive AI as the native currency that is generated by AI it also makes sense that it can be used to buy AI so one very natural way of spending pearls is to buy compute or pay for GPUs business being recorded I have no idea yeah so I I see that questions again about the current question about you know EC funding strategy side investment investors and so forth so again we

33:45you know this is not the purpose of this chat I will just say that the one thing you should know like there was a company is is is well funded we will address this we will release this information when when the time comes the important thing is that there were as you know like no pre allocations no pre-mine no no one including team members including investors any pre-sold token no allocations no not on balance sheet not anywhere else that that was like our our guiding principles for that we never you know that we swore by

34:29and I think we we kept um anyone that invested in the company was you know basically is believe that in the in the technology in the value proposition of the company all the token side was completely you know sidetracked so so all we wanted to ensure is that you know we can we can the team has the resources to just to keep supporting this this the massive R&D that's required to to to support this chain which isn't to a large extent very dynamic right has to evolve with the state of the art of AI and that's that's something that's going to require a ton of R&D

35:14and we can assure that you know we we did what's needed to ensure that this this project will is not a fling that you know something that's really going to integrate to the AI stack and become hopefully something you know a native part of the AI stack yeah I want to address the question around like agentic payments and settlement time and block time so like I think our general design design principle around around that was we wanted to you know showcase this new proof proof of work primitive and and and be able to focus on that

35:57with really really short block times it would have made um it would have made running this efficient with AI a lot more difficult um that was a conscious decision we made around how uh like quicker like L2s or or or roll-ups etc should look we're still actively researching this internally this is some of the research that we actually are like discussions that we also love to have in public and understand from other builders what they would want to build I think also it's cool that this thing is completely permissionless anyone can build whatever they want there's

36:31probably room for a light you know lightning like payment payment channels there can be room for roll-ups we've also considered enshrining some kind of um smart contract there we have like the another person that created uh the consensus algorithm that that tempo uses and that Solana uses he works for us so we're spending time um I'm trying to understand how to do that well but it's just not our was just kind of wasn't in our initial focus Elon do you want to take jonathan's question about lower block times

37:11um sir would you be interested in lowering the block time in the future so payments are faster so just follow up on what did I said lightning is not a bad solution for payments and we're quite flexible to adopting to adapting the l1 who have better lightning than what bitcoin has and more secure and more efficient and so on it's not clear that the l1 has to be super efficient and super flexible and so on so forth there are other solutions there are also drawbacks in reducing the the block time we made like a conscious decision to have it at around three

37:58minutes and we're getting there it's more hashrate than we thought there's going to be further questions so I see Pegasus is still asking about the compute marketplace and the chain how are they connected so again the compute marketplace is basically is going to be a just a cloud inference or training an inference service as as you know um if we if in one we choose to expand to to training to the training use case as well the post training more precisely um and again the marketplace will be just like a neocloud where you would you

38:55you know just to have consumer base come to pearl cloud um we will have our own purlified model garden where we will serve you know like base 10 and cruzo and together ai and you know all those um uh rest of those neoclouds we are we are cooking some you know proprietary like um variants of open source models which i won't share at this point but we are the hope is to make ai not just cheaper um but also better in a in a sense so you know to demonstrate capabilities and ai on on on the chain which which i think will make um

39:41this the pearl cloud more even more um attractive for users um yeah and there are a lot of ways we could you know implement this this this cloud but the essence is that our compute platform will basically revamp from mining packages early or genesis mining to an actual inference cloud uh in the near in the near future hopefully and again the demand the best way to create demand for as i'm sure you all know better than i do the best way to create um demand for the tokens to have is to being able to um like the natural the native use case

40:31for pearls is just paying for compute right that's that's precisely what pearl represents is very analogous to the bitcoin and energy or bitcoin and electricity um and and and we believe we need to control our own destiny and and serve those models of course we could have competition from from from a ton of other you know hyperscalers which which is fine i think uh ultimately i think we all share the goal that of making you know as early adopters we all share the goal of of just uh letting letting it grow right so will you let others provide third party

41:26compute for the platform too so i'm not sure idan do you yeah that that's the idea of the marketplace it connects people that have hardware but don't have useful work or customers that are really willing to pay for it and on the other side you have and when he's over entities that want to do it and the compute marketplace essentially to do the matching yeah i think there is a faucet on the test net and you can also ask for tokens from that faucet and yeah we none of the hash none of the pools that exist are endorsed or backed by the team so i actually have no idea

42:26how they measure hashrate or what they actually do unfortunately i can't answer any of the questions about how they work or how they account hashrate okay so that's a good question about hashrate um how does the team refer to hashrate we we're not doing bitcoin for right so we're not actually doing hashes as lottery tickets in a sense and what we call hashrate is essentially the the number of useful ops that we are utilizing for mining so it's a different measure like i saw lots of posts on twitter saying that our hashrate is like one percent of bitcoin hashing

43:38but in parenthesis they say but the different protocol so yeah it's not the same it's not the same measure uh our hashes or our useful ops is not the same as a single hash on bitcoin so ours is exactly what they said it's the number of a useful operations that gpu is doing which is math of math operation may multiply add this is the number of math operations that we're able to utilize for uh for creating a lottery ticket hope that makes sense um tiles so is tiles per second is not is not precise because every miner can choose their own tile size like you're probably using

44:39the default one but essentially the protocol is agnostic to the tile size and we we normalize the hardness of that of creating a proof of work based on the size of the tile so tile per second is not precise we're just counting the number of ops which is hopefully uniform and correct yeah there is a test net there's a false set uh and to answer pistachio on what exactly was a mathematical breakthrough so it's it's essentially to do uh consensus or distributed consensus from a real-world computation um if you think about proof of work in in bitcoin um

45:25or any other proof of work it's just uh the problem is is the same for all the different miners and that's why you can implement the proof of work because no one has an advantage uh matrix multiplication especially when the miner can just select whichever matrix multiplication they want is inherently as different hardness uh and and the protocol is able to flatten the hardness for all the participants um while without affecting the actual work being done so it's essentially to do proper uh proof of work um but where the

45:57workload being done can be selected by the participant and it's and it's not gameable yeah the user selected is is the the key questions you know it's uh basically flattening the the hardness it's um and that's and that's the total the total address will compute as a result of that for pearl is theoretically all the matrix multiplications happening in the world can secure the network whereas other um other attempts at doing this are generally focused on a very narrow subset of work and then therefore the the amount of compute and then hence security that

46:39they can achieve is is um is going to be a lot smaller don you want to answer uh question listings or um yeah the work we're not like focused on that at all honestly on any of the I think like you the OTC stuff or any of that stuff that's happening it's just been like organic from the community our approach and including I mean all the different uh community uh investment reports that have gone out which which are which are an awesome to read it's all organic uh and we want it to continue to continue to be that way that's that's how we think the best way

47:35to build this community is we want people that are intellectually curious about the technology and whether no matter how technically deep you are today with with with clod and gpt you can get stuff to work and you can do optimizations and we and that's for us like really what we're focused on and we assume that once there's enough interest on it um then people will list it just like when there was enough interest people built OTC uh apps and we I can assume that stuff will continue to happen but that's not like a really a focus for for the team yeah so for for Dizzle's question

48:09on promising applications for math moles beyond AI absolutely so you know before even AI came along so still you know all the quantum simulations you you hear about graphics rendering gaming all the high frequency trading uh firms everything is just math moles it's like 90 95 percent math moles um so um way beyond that right like graphs like you know vector databases graph search uh compression uh scientific computing everything is math moles so uh uh basically um you know essentially any industry scale solution uh you can think of uh about is

49:01is probably between 50 and 90 percent math moles so yeah so that that that was the biggest design decision here is not not doing something verticalized or not focusing on training or inference or routing or whatever just focusing on math moles which is a proxy for general compute as he done mentioned earlier um yeah well CASPA is is a different thing CASPA was random matrix like was basically if anything was like a random matrix multiplication which is very very different right um in fact if you think about it you can literally implement bitcoin like shock

50:01250 like universal hashing is something you can easily implement with random matrix multiplication so the fact that you can implement bitcoin proof of work with random matrix multiplication that's almost trivial like it would surprise no one implement to get with arbitrary matrix multiplication where the miner chooses her own or his own uh workloads at her own discretion that's a whole different story so i think the comparison to CASPA is not even you know it's not it's just a very different uh creature um yeah again we're not focused at all in exchanges and so we're not

51:07talking to anyone we're obviously receiving a ton of of inbound requests we're focusing purely on the tech and integrating into ai partners and uh and the enterprise um uh move and building the beta c consumer facing uh inference compute marketplace and and that's it we're currently um we feel it's much healthier for the network for everyone um you know to defer listings and and all sorts of funding and you know um any crypto economics things to to this is just noise as far as we're concerned at this point everyone this was fun we should do it again

52:21absolutely um thank you everyone for participating and once again we really appreciate all the the you know genuine curiosity and support and you know feedback and we we keep monitoring this this forum closely and we really appreciate it thanks everyone bye bye see everyone thanks guys bye bye

00:00One could argue that the real currency is not money, it's energy and data. The two, what we collectively call compute, so the two scarce resources whose fusion creates intelligence is data and energy, data and compute. We believe that this dramatic shift in the production of

00:17knowledge compels us to rethink the fundamental properties purpose in the creation process of money. What money actually is? The basic scientific question that started this, the PRO project, and we, Elon and I, asked ourselves about a year and a half ago,

00:38was whether we can build a monetary currency directly from these two key scarce resources, data and compute. This is the motivation for this question, is not merely mathematical or philosophical. It is to address at least three core incompatibilities of the fiat system with emerging AI economy.

01:01The first one is that the unit economics of AI today is simply non-fungible, not all LLM tokens are born equal. Producing a deep-seek R1 LLM token on Nvidia Hopper has a very different physical cost, production cost, then producing a reasoning token, a chain of thought of Quen14B on AMD MI3s, right?

01:27So these are just two very different physical costs. The current crude market pricing of, you know, X dollars per million LLM tokens is at best a crude proxy for this actual energetic cost, but it hardly serves as a proxy for the data dimension, which is much more elusive and harder to articulate.

01:52The second drawback or limitation is that in the current economy, AI consumers are completely left out of the upside of the wealth generated by AI. So AI users like myself who are, you know, just using cloud for coding or chatbots, we don't get it in a prop or Nvidia share, well, most of us, at least,

02:16by doing so, yet consumers are the ones driving demand and model improvement through prompting RL data labeling, green for our legit and so forth. But we don't really have a stake in the wealth generated by the AI revolution, right? So that's another part.

02:37The third one and final one is that in this era where autonomous AI systems are slowly penetrating or, you know, economy and driving a lot of the value, we must have an economic system that allows machines to participate alongside humans in the same ecosystem without, you know, onboarding KYC and identification, right?

03:01That's, we just have to allow machines to somehow manage the two only resources we deposit in their hands or in their, you know, in their system, which is compute and data, right? So AI agents just receive two researchers, the data, the prompts and the compute budget.

03:21And we believe AI agents should be able to transact seamlessly and directly using these resources to manage them without any permission or regulatory or social restrictions with their, obviously not subject to, right? So this is what we often call among ourselves

03:41the money for machines, right? So it's money for machines and humans.

00:00The real protocol, the eventual protocol, which we call Pearl Gem, I think this is the probably Pearl's biggest innovation, is to follow the same idea, but in order to make the peeling of the noise the decoding cheap, maybe you can just add structure to the noise. So instead of just using brute force random Gaussian matrices, what we can do is we can just add the noise matrices which have,

00:22they look like kind of marginally random, but they actually have structures. So if we add what's called low-rank random Gaussian matrices, low-rank matrices are still end-by-end matrices if you open them up. But they have linear correlations. So actually you can decompose them into the outer product of two strip matrices. And the thing you should know is that this completely solved the peeling problem

00:46because now Alice needs to peel off those extra three products. If you do the math, you'll see that all those peeling products are going to be low-rank, rank R matrix multiplication, that's going to cost Rn squared, which is negligible compared to R cubed. Okay, so in practice, Elon and Vippo will talk about this, we'll see what's the actual overhead in practice,

01:08but that's great because now peeling is just negligible, right? But I claim this screws up another thing. This protocol is no longer secure because if Alice now chooses the all zero's matrices, Alice is left with a much easier problem than an honest player that's multiplying dense matrices. Alice now just needs to multiply low-rank matrices, rank R matrices.

01:31Alice can do this operation in Rn squared time, whereas an honest miner running actual LLM matmoles is going to pay N cubed. That's completely breaks the security. The system is no longer fair, right? There are easy instances. Does that make sense?

01:47It looks like we're trying to hold the string from both ends. We tried to make the system secure, so we added 300% overhead, and we tried to reduce the overhead, we screwed up the security, right? It seems like fundamentally this trade-off seems inherent. And I think the key innovation here, the key conceptual idea, was the following that.

02:08You know, for an honest miner, for an honest Alice that wants to run LLMs, all Alice cares about is the output, is the product of A and B, right? She doesn't care about the intermediate computations, but no one said we need to do the proof of work on the output. We can, we're free to do this on intermediate computations as well. And I think this is the key idea.

02:28So we would like to prevent Alice from exploding the low-rank structure of the noise in the intermediate space, right? And this is how we do it. So instead of just verifying or hashing the output, of A prime times B prime, which is gameable, we just saw it. This is easy.

02:48Instead, we're going to verify we're going to do the proof of work, not on the output, but on the intermediate computation that the GPU already does. So when the GPU multiplies those two R by R tiles, we will just for every such R by R tile that the GPU is going to multiply anyway, we're going to force Alice to hash the product, the result of those intermediate tiles,

03:15and check if it starts with, let's say, 19 zeros. This is exactly Bitcoin's adjustable difficulty puzzle problems. And the nice thing is that about low-rank random matrices is that marginally every R by R tile set that you see here is going to be uniformly random. This is just a R-wise independent distribution.

03:36So that's a standard property. Of course, there's tons of correlations between those tiles.

00:00Each and every proposal for useful proof of work was shown to fall short of satisfying one of those core definitions or properties of useful work. And indeed none of those attempts have succeeded. One of the most vocal votes was by Vitalik, the co-founder of Ethereum, who said that if we can find some useful computation, which is easy to verify, then crypto mining could actually become a huge boon to society, not only removing the objection that Bitcoin wastes energy, but also being societally beneficial. And then a few lines later, Vitalik adds is probably infeasible. A few years later, Ethereum switches from proof of work to proof of stake, and the rest is history.

00:00What's cool about our system, and I think you all understand it by now, you don't need any special hardware, you don't need ASICs to mine it, you can use your GPUs, the ones that you have at home, or the ones that you are renting from a data center. You can use your favorite inference engine. We have plugins for SGelang and for VLLM.

00:19So you can just do pip install and start mining basically. There's minimal setup customization right now. The existing system supports int computations. So we do need to quantize models to int right now if you want to do useful work. But we have a next generation scheme that we're working fiercely on now, and it should be ready in a couple of months supporting floating point computations.

00:46And then you won't need to do any quantization whatsoever.

00:00So I mentioned that we need to quantize models to int. In fact, we need to quantize them to int 7, which is highly weird, which is somewhat weird. And the reason why we need to quantize things to int 7 is because we're adding noise. And we need the signal plus the noise to fit into int 8.

00:16Int quantization to int 7 was not a thing. So we had to create our own quantization technique. And as usual, we came up with something that we believe is SOTA in terms of quantization quantization. We came up with a way to use existing tools, GPTQ and smooth quant, which are existing tools for quantization.

00:36But we combine them in a very non-trivial way to get an end-to-end quantization of existing models that outperforms any previous quantization, even to FP8.