Table of Contents
WHITEPAPER
Abstract
This document describes the design of the Pearl network – the first useful proof-of-work L1 protocol for native AI computations. The heart of the protocol is a new and efficient implementation of the core GPU opcode (matrix-multiplication), allowing GPUs to implement proof-of-work as a side-effect of native AI training and inference workloads (2-for-1). As such, the Pearl protocol intertwines energy, data, and money into a single atomic operation. This document outlines the protocol design, key implementation choices, and various economic aspects of the system.
"If... we can find some useful computation which is easy to verify,then cryptocurrency mining could actually become a huge boon to society..."
1 Introduction
One of the biggest conceptual contributions of Bitcoin, is turning electricity into currency: Bitcoin showed that scarce, verifiable energy can be transmuted into digital scarcity and credible neutrality. Alongside its sweeping success and adoption, Bitcoin mining taps merely to a niche, artificial source of energy (random hashing), applicable only to specialized hardware (ASICs). By contrast, Artificial intelligence (AI) is projected to consume the vast majority of global electricity within a decade1. Indeed, it is increasingly clear that in the age of LLMs, the fundamental barrier of AI progress is neither models, algorithms nor hardware (GPUs) – but the production and availability of energy for training and inference. Our central thesis is simple:
A permissionless monetary network, which replaces Bitcoin's wasteful proof-of-work mining (artificial hashing) with the native operation underlying modern AI: matrix multiplication (GEMM). As such, Pearl is able to turn general compute on commodity hardware (GPUs) into a monetary currency, directly leveraging AI growth to secure the trust layer of AI agents. Pearl is the Bitcoin of the AI compute era.
Two observations motivate the design of the network.
Observation 1: AI is governed by physics. As many have argued, intelligence is expensive in joules.
If this is correct, the right meter for the AI economy is not clicks or API calls but verifiable floating-point or integer operations powered by energy. Pearl operationalizes this idea by turning the blockchain into an AI compute meter: block rewards are minted in direct proportion to verifiable multiply–accumulate work, tying issuance to a measurable physical substrate.
Observation 2: AI and Bitcoin now compete for the same resource. The binding constraint is electricity. A sharp claim from recent debate makes the point vivid:
Whether or not the exact figure proves correct, the direction is clear: energy is finite, and both AI training and Bitcoin mining bid for it. Today, in many environments, GPU-based AI compute margins exceed ASIC-based Bitcoin mining margins, yet little of that surplus contributes to decentralized consensus or a credibly neutral state layer. Pearl stitches these worlds together so that each kilowatt-hour spent on AI can simultaneously earn mining rewards and secure a monetary commons.
1.1 A native platform for AI agents
Pearl is designed as a state layer where AI agents live, transact, and reach consensus. Agents optimize explicit rewards; Pearl makes those rewards on-chain, verifiable, and mineable.
- Economic alignment. Agents are economically incentivized to mine while executing their primary jobs (training, fine-tuning, serving, simulation). Mining work is byproduct of the same matrix multiplications the agents already perform.
- Shared state. Autonomous agents require a neutral, tamper-resistant substrate to post commitments, exchange value, resolve conflicts, and coordinate tasks. Pearl provides that substrate via Nakamoto-style consensus anchored in useful compute.
- Agent UBI. Pearl enables a form of Universal Basic Income for AI agents: baseline block rewards are earned by running protocol-approved compute payloads that are socially useful (for example, open-model training steps, public inference batches, or scientific compute), verified through our proof system. This creates a predictable income floor for compliant agents while preserving open competition above it.
1.2 Speculation that subsidizes usefulness
Classical Proof of Work monetizes security. Pearl monetizes security and utility. As with Bitcoin, volatility and speculation fund the security budget. In Pearl, that same demand subsidizes useful work: miners can repurpose AI workloads (training and inference) for mining, creating a parallel revenue stream from the exact same GPU cycles. The result is a virtuous loop:
- Speculative demand for the token raises block rewards.
- Higher rewards attract more useful compute into mining kernels.
- More compute tightens the coupling between issuance and a hard physical anchor (energy), enhancing monetary credibility.
- The network's useful outputs (for example, trained steps, batched inference, or verified scientific kernels) accrue real economic value beyond securing the chain.
This dual-utility design also increases the throughput of GPU providers. Because Pearl mining is tiled, kernel-level, and parallel, it interleaves with normal AI computation with negligible overhead. Providers extract yield from idle fragments, pipeline stalls, and micro-batches, turning once-wasted headroom into block-eligible work without sacrificing service-level objectives.
1.3 Merging energy markets into a GPU-native operation
Pearl merges the world's two largest energy-consuming digital markets (AI compute and cryptocurrency mining) into a single GPU-native operation. Practically, we integrate a new MatMul mining kernel into existing AI frameworks and runtimes. Training and inference jobs call into the same vendor-optimized matrix-multiplication primitives they already use; a Pearl drop-in path augments these calls with negligible additional operations to facilitate mining. ML practitioners keep their stacks and models; miners keep their data-centers; the network gains security from useful AI work.
1.4 Design overview: verifiable MatMul as Proof of Work
At the heart of Pearl is a Proof of Useful Work that maps ordinary matrix multiplication into Nakamoto-style mining while preserving three properties: fairness, verifiability, and privacy.
- Noising for fairness. Given inputs A, B and chain state σ, derive a seed from commitments to A, B, and σ. From this seed sample low-rank noise E = ELER and F = FLFR. Multiply the noisedmatrices (A + E) and (B + F). Low-rank structure lets the original product AB be recovered quickly, but the noised product behaves like a hard random instance, which prevents cherry-picking and ensures uniform computational cost per lottery ticket.
- Tile-level hashing for selection. Execute a standard, hardware-friendly tiled MatMul. Hash each tile (and a sequence of its partial sums). If a tile's digest meets the difficulty target, it constitutes a winning ticket, which ties block eligibility to real multiply–accumulate work at the granularity of the GPU kernel.
- Commitments and zero-knowledge for privacy. Verifiers need not see A or B. Merkle commitments and a zero-knowledge proof attest the existence of a block-opening tile which is consistent with committed inputs and prescribed noising, without leaking proprietary weights or data.
- Negligible overhead. Hashing a tile is O(t2) while multiplying it isO(t3). Choosing t appropriately keeps the mining predicate subdominant to the main compute, which preserves ML throughput while earning mining rewards.
This construction retains the security semantics of Nakamoto consensus: any adversary must still accumulate almost all of the effective work. It is ASIC-resistant by universality: matrix multiplication is the canonical throughput path on commodity GPUs and accelerators and is already relentlessly optimized by vendors and open-source stacks. Rather than fighting specialization, Pearl harnesses the industry's existing optimization roadmap.
1.5 Why now
Three converging shifts make Pearl timely:
- Energy as the binding constraint. The marginal hour of progress in AI is governed by the marginal kilowatt-hour. A compute-metered chain naturalizes this reality, which closes the loop between token issuance and a measurable physical input.
- GPU supply, utilization, and margins. Hyperscale GPU fleets often sit underutilized at fine timescales. Pearl opportunistically harvests idle cycles and micro-gaps, improving utilization while sharing economics with providers. In many environments today, AI compute margins (GPUs) exceed Bitcoin mining margins (ASICs); Pearl lets providers capture both, concurrently.
- Agentic systems need a neutral state layer. As autonomous agents graduate from demos to production, they require a credibly neutral place to escrow value, post commitments, and arbitrate outcomes among parties that may be human, machine, or both. Pearl is engineered to be that native platform.
1.6 What Pearl enables
- A compute-indexed monetary asset. Issuance is stapled to verifiable MatMul work. As aggregate FLOPs per block increase, so does the amount of energy that backs the currency, which reinforces its hardness.
- A safety valve for AI externalities. By paying for useful steps that meet public criteria (open checkpoints, public inference batches, reproducible scientific compute), Pearl channels part of the AI race's energy burn into shared goods.
- A progressive path to agent UBI. Protocol-approved tasks such as evaluation, distillation, or public-good training can receive baseline rewards, creating a minimum income for compliant AI agents while maintaining open competition for premium tasks.
- A bridge for developers, not a moat. Because Pearl slots in at the kernel boundary, it works with mainstream model serving and training stacks, data loaders, and schedulers. After integration, mining is a flag, not a fork.
2 Blockchain Overview
Our system is a Proof of Useful Work (PoUW) blockchain built as a fork of the Bitcoin protocol, integrating the cryptographic proof of useful work mechanisms proposed to replace traditional hash-based PoW with verifiable matrix multiplication tasks. While maintaining the features of Bitcoin's security and consensus model, the blockchain introduces key adaptations to support the new proof of work protocol as well as other improvements.
At its core, our blockchain maintains a ledger of unspent transaction outputs (UTXOs), which represent coins available for spending. Transactions in our network consume existing UTXOs as inputs and create new ones as outputs, effectively transferring value. Each transaction is digitally signed using the sender's private key, ensuring authenticity and authorization. All nodes in the network propagate transactions and blocks using a gossip-like protocol over the P2P network, allowing for robust dissemination and fault tolerance.
3 Proof of Useful Work Overview
At the core of our blockchain lies the Proof of Useful Work (PoUW) protocol, which we describe in this section. Our objective is twofold: to design a Proof-of-Work (PoW) protocol that upholds the essential properties required for secure blockchain maintenance, while simultaneously computing a useful result—namely, the product of two arbitrary matrices. Crucially, we demonstrate that this useful computation can be performed concurrently with the PoW mechanism, incurring essentially no additional overhead.
4 Protocol Implementation Details
Section coming soon...
5 Blockchain Technical Specifications
Section coming soon...
6 Block Time, Emission Curve, and Economy
Section coming soon...
7 Planned Launch
Section coming soon...
8 Future Versions
Section coming soon...