Pearl’s Two Leading Mining Pools Control Most of Its Hashrate

Pearl faces a decentralization test as two pools control most of its hashrate and developers have made three hard forks in a few months to close mining vulnerabilities.

Pearl launched in April 2026 as a hard fork of Bitcoin (BTC). Its proof-of-work system uses matrix multiplication, a core operation in AI inference, aiming to turn the work that secures blocks into computing that can also serve AI customers. That design makes the distribution of mining capacity and technical authority central to the network’s security.

Kryptex and PearlPool together control most of Pearl’s hashrate. Miners can switch pools, but most mining activity remains concentrated in two operators. In proof of work, hashrate is the computing power used to compete for blocks. Concentration can give a small number of pools a larger role in organizing that competition.

Pearl also made three hard forks over a few months this summer to address mining vulnerabilities and sharply restricted supported hardware. The fork cadence and hardware rules have become part of the debate about who can shape the network and participate in mining.

Bitcoin’s block subsidy declines at each four-year halving, while transaction fees have not grown enough to fund network security on their own. Miners sell some of their Bitcoin to cover electricity and operating costs. Meanwhile, demand for AI computing is drawing energy and data-center resources toward commercial AI work, adding to competition for mining capacity.

Pearl’s economic model seeks to link those commercial workloads with mining. An operator that already serves AI customers may use idle or overlapping computing capacity to mine, collecting Pearl token (PRL) rewards alongside customer payments in fiat or stablecoins. If those payments cover a meaningful share of expenses, the operator may have less need to sell mining rewards to pay bills.

But mining rewards do not show whether a paying AI job took place. Pearl allows miners to earn PRL for matrix calculations using random inputs; the network can verify the calculation without requiring it to come from a customer’s model or inference request. The distinction between a valid matrix calculation and a customer workload is central to Pearl’s reward design.

Pearl is in its first year of issuance, when token inflation is high. Subsidies are available for random-input calculations as well as commercial inference, so miners can earn rewards without serving paying customers. That leaves the project’s economic case dependent on whether customer revenue becomes significant alongside token rewards.

A longer-term plan would use PRL to pay providers in a decentralized inference market. Pearl is working with Attestable on a Proof of Inference layer that remains in early development. It is intended to keep prompts and private data from node operators while allowing users to verify that an output came from a specified model. The proposed verification and privacy features are meant to distinguish the service, since low prices alone may not be enough to compete with large cloud providers.

For Pearl’s security-budget case to work, customer payments must become meaningful to miners while mining remains broadly distributed. The network’s inference layer is still early, and its protocol continues to reward calculations that do not establish paid customer use.