L2 queue collapses, PoW begins to pile up: What are foreigners really saving up for in the highly discussed pearl-2: native of TOKEN2049? Recently, I have seen several PoW projects emerge consecutively, and coincidentally, L2 projects such as Blast and Abstract are being shut down one after another. During TOKEN2049, one project that foreigners discussed fiercely was Pearl (PRL). After studying it for a while, I found that its team, technology, and underlying funds do have something, and it is trying to solve a long-standing problem: can AI computing power easily mine a currency similar to Bitcoin? What exactly does PRL do first? It is an independent PoW public chain that will be launched on April 27, 2026, using Proof of Useful Work, PoUW）。 Bitcoin mining machines perform SHA-256 hash operations to compete for block rewards, while Pearl replaces mining work with matrix multiplication, which is widely used by AI models. If an AI company is already using GPUs for customer inference, then the same batch of calculations can also participate in Pearl mining and receive PRL rewards. In an ideal situation, a GPU can earn both AI customers' money and mining money, allowing miners to earn more and AI service providers to use some of the profits to lower prices. Team background is one of the biggest bonus points for PRL. CEO Omri Weinstein is a Ph.D. in computer science from Princeton and an associate professor at Hebrew University. He has previously taught at Columbia University and has experience with NVIDIA and VAST Data Another co-founder, Ilan Komargodski, is an associate professor of cryptography at Hebrew University and a PhD from the Weizmann Institute of Science. He has conducted research at Cornell Tech and NTT Research. The third co-founder, Idan Sugarman, has a background in business and investment, and Cornell cryptography researcher Rafael Pass is also involved in economic security research. The core team has been researching computational complexity, cryptography, and verifiable computing for a long time, and published the original PoUW paper in 2025. The technology is not just a temporary attempt to ride on AI hotspots. There has also been substantial progress in commercial implementation. On May 15th of this year, Together AI officially announced a partnership with Pearl Research Labs to launch Gema-4-31B-it-Pear inference service, which reduces the inference price of the model by more than 25% with mining subsidies. Pearl has also built its own inference platform compatible with OpenAI API, supporting open models such as GLM and DeepSeek, which can be recharged in US dollars and charged based on token consumption. We have customers, products, and charging mechanisms, but currently we still lack key data such as real payment income, daily average inference volume, and customer retention, which cannot prove that this business has been successfully launched on a large scale. Who else is supporting it? No Limit Holdings (NLH) has publicly stated that it is actively accumulating PRL. Its investor Brian Breslow published a lengthy study in September, calling Pearl a potential next-generation Bitcoin and even setting a forward target of over $30. The core figure of NLH, Gin Chao, has served as Binance's Chief Strategy Officer and a member of BinanceUS's board of directors, indicating that there is indeed a certain background in the funding circle that is bullish on PRL, but this cannot be understood as Binance investing in or endorsing Pearl. PRL does not have traditional VC token rounds and pre mining allocations. In the early days, chips were mainly generated through GPU mining, and then circulated through third-party over-the-counter markets such as Pearl OTC and Black Pearl OTC. Pearl OTC website reported a cumulative transaction volume of approximately $31.5 million, and spot trading has also appeared on exchanges such as BigONE and SafeTrade. The total limit of PRL is 2.1 billion coins, without the conventional team pre mining and VC unlocking. The issuance method is a smooth decreasing block reward, without the Bitcoin style four-year sudden halving. The problem is that the early mining difficulty was very low. In the first 5.9 days, 121.7 million coins were mined, accounting for about 36% of the current circulating quantity. It is not transparent how much the team and early miners each took. Based on a reference price of approximately $1.28 on October 8th, the circulating market value is about $430 million, FDV is close to $2.7 billion, and theoretically, there are still about 1 million new PRLs added every day. The potential selling pressure formed by the continuous output of miners cannot be ignored. There is still a big controversy in terms of technology: Pearl can prove that miners have indeed completed matrix calculations, but cannot prove that these calculations must have come from real paying AI customers. In June of this year, a study observed the software and actual mining behavior of miners and found that a large amount of calculations could be completed using random matrices, without the need for anyone to purchase inference services. The logic of the project is that in the future, miners who truly serve AI customers will have two sources of income, while pure miners will only receive coin rewards. As competition increases, the cost advantage of the latter will disappear. This economic model makes sense, but it has not yet been proven by large-scale real business data The MoE mining vulnerability that occurred in July also reminds us that consensus implementation is still in its early stages, and the FP8 upgrade that supports a wider range of AI workloads still needs to verify the implementation of the main network. At present, Pearl's inference platform mainly collects US dollars, and miners earn customer payments and PRL rewards. The token does not have a confirmed large-scale income dividend or mandatory repurchase and destruction mechanism. Expanding business can improve miners' economic efficiency, enhance network security, and boost market confidence, but in order to sustain the conversion of AI demand into direct PRL buying, it still depends on whether there will be real demand for token payments, coin holding settlements, mortgages, and reserves in the future. Otherwise, even if the inference income increases, the coin price may still be diluted by continuous mining and issuance. The death wave of L2 indicates that the market does not need so many redundant chains, and projects like PRL are trying to find a new way out for PoW. PoW+AI has the opportunity to become the next important narrative, but ultimately determining how far PRL can go is whether AI companies are willing to pay for this computing power in the long term. Enabling AI computing power to mine coins is a technology, and making this coin a real asset that the market needs to hold is a big business.