The Vision
As AI scales to industry-wide adoption, it is becoming increasingly clear that the future economy will be denominated in compute cycles more than in human labor.
We believe a new form of money must emerge from that reality - an open, auditable AI-native currency, issued directly by AI inference and training. Such a monetary system will improve LLM unit economics (which are currently nonfungible) and would shift a portion of AI-generated value to users, who are driving production and model improvement, but have zero upside in the AI revolution.
“If we can find some useful computation which is easy to verify, then cryptocurrency mining could actually become a huge boon to society… but it is probably infeasible.”
Since the inception of Bitcoin, an outstanding open problem in distributed systems was whether it is possible to implement Proof-of-Work consensus on top of real-world computation, as opposed to useless random hashing. While long considered impossible, last year we answered this question affirmatively. Our mathematical breakthrough shows that GPUs can implement Proof-of-Work as a by-product of native AI workloads.
We’ve spent the past year turning this “2-for-1” breakthrough into a working infrastructure, building from the linear algebra all the way down to the CUDA kernels, alongside world-class mathematicians and low-level engineers.
Today, we’re pleased to announce that the Pearl Network is ready - a fully operational blockchain, with vLLM and SGLang plugins, and a cryptocurrency secured directly by AI computation.
What this means is that the hundreds-of-billions (and soon trillions) of dollars of compute being deployed for AI workloads will double - for effectively free - to secure Pearl’s proof-of-work chain; all the properties of Bitcoin, but secured as the by-product of AI training and inference, i.e., by the native operation of GPUs: matrix-multiplication.
Pearl changes the unit economics of training and inference - it transforms compute from a sunk expense into an asset. An AI-native store-of-value, anchored directly to the production of intelligence.
We hope that you find our vision and technology interesting and compelling enough to help us build the future economy and trust layer of AI.
Fig. Data, Energy, Money
The transition from AI as a tool to AI as currency represents a deep shift in the meaning, distribution, and fundamental purpose of money. In a world where the majority of GDP is produced by non-humans and intelligence becomes a mechanical process, the Pearl protocol demonstrates how intelligence itself can be made into a currency - auditable, fungible and portable.
Intelligence is the process of fusing energy with data - the two scarce resources required to produce Pearls. Since data (“useful work”) is ultimately produced by AI consumers, who are collectively contributing to model improvement and progress of AI, a monetary system rewarding unique data production alongside computational power can lead to a more equitable distribution of AI-generated wealth - allowing users to participate in the value created by AI systems, rather than remaining passive consumers of datacenters, frontier labs and AI enterprises.
Pearl will also enable a more efficient way of trading compute: in the same vein that Bitcoin is used to trade energy outside the dollar system, Pearl can be used to trade AI compute directly, without commissions, regulations and overheads of banking and the dollar system.
Pearl is the Bitcoin of the AI Era.
Beyond money, Pearl introduces a new trust layer for AI systems, intertwined with compute itself, enabling a new class of applications on the network. Autonomous agents that plan and act on our behalf require an asynchronous medium of exchange they can natively verify. Proof-of-Work (inference) provides exactly that: a currency embedded in the native atomic operation agents use to think, operate, and communicate: matrix multiplications.
The Pearl protocol described in this manuscript is merely an infrastructure for a new digital economy. By making AI computation auditable, attestable and monetizable, the Pearl network could serve as a non-sovereign, immutable database containing inference traces of all AI systems (with post-quantum ZK encryption, ensuring the largest enterprises never leak private information). This feature will enable critical applications such as model tracing, GPU telemetry and financing, and will mitigate the problem of semantic verification of LLMs (through PoW “credit score”), to mention a few.
We are entering an era where compute cycles are the primary denominators of social welfare. Pearl ensures that the monetary and trust layer of that era are grounded in the same principles that govern its technology: a trustless, asynchronous and incentive-compatible system, backed by math, and open to anyone…
“The curious task of economics is to demonstrate to humankind how little they really know about what they imagine they can design.”
