Is anyone still using PoW in 2025? When I first saw Irys announce they would use PoW, my first reaction was: Are they serious? In 2025, Ethereum has long completed The Merge, Solana and Sui have been PoS from the start, and even the Bitcoin community is discussing whether to consider some form of proof of stake. In this era where PoS has become the industry consensus, Irys actually wants to use PoW? My first reaction was: Is this team out of their minds? But as I delved deeper, I realized: I was wrong, and wrong in a big way. Irys's choice is not retro, not stubborn, but a profound understanding of the essence of the problem. Why do we feel that PoW is "outdated"? Let me first explain why everyone thinks PoW is outdated. Over the years, PoW has been criticized mainly for several points: High energy consumption. Bitcoin's electricity consumption exceeds that of many countries, which is indeed a big problem in today's world where environmental awareness is growing. Poor scalability. PoW has slow block generation speed and low throughput, making it difficult to support large-scale applications. Centralization risk. Hash power is concentrated in large mining pools, which goes against the original intention of decentralization. So when Ethereum announced it would switch to PoS, the entire industry cheered: Finally, we are saying goodbye to this old, inefficient, high-energy-consuming mechanism! I thought the same way at the time. But data chains are not financial chains. It wasn't until I started researching Irys that I realized a neglected issue: Data chains and financial chains face completely different challenges. What is the core problem that financial chains need to solve? It is the confirmation and ordering of transactions. Who transfers first, who transfers later, does this money really exist—these are the problems that financial chains need to solve. For these issues, PoS is indeed a better choice. It is faster, more energy-efficient, and has better scalability. But what about data chains? The core problem that data chains need to solve is: how to prove that the data has really been stored? This is a completely different question. The fatal flaw of PoS in data storage. Let me tell you a story. Suppose you have a data storage chain that uses PoS. Someone uploads 1TB of data to your chain, how do you prove that this 1TB of data has really been stored? The most direct way is: to submit storage proofs regularly. Nodes must submit a proof to the chain at intervals, saying "I have indeed stored this data." This proof could be a hash value, or a Merkle tree root, in short, something verifiable. Sounds reasonable, right? But here comes the problem. If your chain stores 1PB of data, with 1000 nodes, and each node submits a storage proof once an hour, then there will be 1000 storage proof transactions every hour. In a day of 24 hours, that’s 24000 transactions. This is just for storage proofs, not including other transactions. Your blockchain will quickly be filled with storage proof transactions. Block space becomes very expensive, or you have to increase the block size, but that brings other problems. Worse, this is not the end. As the amount of stored data increases and the frequency of storage proofs rises, it will eventually reach a critical point: the chain simply cannot handle so many storage proof transactions. At this point, what will many projects do? They will "prune" historical data and host it in centralized databases. Yes, it’s cheaper, but this completely goes against the original intention of a data chain—decentralized storage. This is the fatal flaw of PoS in data storage scenarios. Irys's PoW: An elegant solution. Irys's approach is completely different. They treat PoW as a sampling mechanism. Let me explain what this means. In Irys's system, miners continuously generate storage proofs. But not every proof is submitted to the chain; only when a proof meets specific conditions (like a hash value being less than a target value) will it be packed into a block. This is like a sampling. The miner generates 1000 storage proofs, but only 1 is packed into a block. But the existence of that 1 proof proves that the miner indeed performed those 1000 validations. Because if the miner did not actually store the data, they would not be able to generate valid storage proofs, let alone happen to generate a proof that meets the conditions. This design is so elegant. It uses one proof on the chain to represent countless off-chain validations. The burden on the chain is greatly reduced, but the security is not compromised. This is why Irys can support large-scale data storage without being overwhelmed by storage proof transactions. But just having PoW is not enough. To be honest, if Irys only used PoW, I wouldn't be this excited. Because pure PoW also has problems. The biggest issue is: speculative play. In a pure PoW system, miners can constantly switch partitions, mining whichever partition is more profitable. This leads to network instability and can compromise the durability of storage. Irys's solution is to overlay a staking mechanism on top of PoW. For every 16TB partition, miners must stake a deposit to be eligible to participate in block production. This design has two clever aspects: First, it prevents account spamming for mining. If you want to mine by creating a large number of small accounts, sorry, each account must stake a deposit. This greatly increases the cost of attacks. Second, it solves the network propagation delay issue. Because of the staking, miners are more likely to store data long-term and stably, rather than frequently switching. This makes the network more stable and reduces propagation delays caused by nodes frequently entering and exiting. PoW + staking, this combination is executed beautifully. This reminds me of some past events. To be honest, Irys's design reminds me of many past events. I remember back in 2017, everyone was saying "blockchain can solve all problems." Supply chain, copyright protection, identity verification, everything had to be on-chain. At that time, everyone was using Ethereum or Ethereum forks. What happened? Most projects died. Why? Because they used the wrong tools. Ethereum was designed for financial applications; if you insist on using it for data storage, you will certainly encounter various problems. It’s like trying to hammer a nail with a screwdriver; of course, it won’t work. Tools must match the scenario. This principle is easy to say, but very few truly understand and practice it. Irys has shown me this understanding. They did not blindly follow the trend of PoS but seriously considered: for the data storage scenario, what is the most suitable consensus mechanism? The answer is PoW. Not because PoW is more advanced, but because PoW is more suitable. Data chains and financial chains: two different paths. This matter has made me rethink a question: should blockchains have different forms? In the past few years, we have been pursuing a "universal blockchain"—one chain to solve all problems. Ethereum wants to be the world computer, Solana wants to be a high-performance public chain, and every chain wants to be a universal infrastructure. But perhaps, this direction itself is wrong. Perhaps different application scenarios require different blockchain architectures. Financial applications need fast confirmation and high throughput; PoS is more suitable. Data storage needs durability proof and large-scale expansion; PoW + staking is more suitable. Computationally intensive applications may require other consensus mechanisms. Blockchains should not be a universal solution but a toolbox. Different problems require different tools. What does this mean for the industry? What insights does Irys's choice provide for the entire industry? I think there are three points: First, do not blindly follow trends. PoS is a trend, but it is not the only answer. In specific scenarios, "outdated" technology may be the most suitable. Second, deeply understand the essence of the problem. Irys was able to make this choice because they deeply understood the fundamental challenges of data storage. If you only superficially look at "PoW consumes a lot of energy, PoS is more environmentally friendly," you will never think of this solution. Third, innovation often comes from combinations. Irys did not invent a new consensus mechanism; they cleverly combined PoW and staking. Innovation does not always start from scratch; sometimes it’s about combining existing things in new ways. My personal judgment. To be honest, I don’t know if Irys will ultimately succeed. Data storage is a fiercely competitive field, with Filecoin, Arweave, and Storj having been deeply involved for many years. As a newcomer, Irys has a tough road ahead to prove itself. But what I do know is: Irys's technical choice is correct. They were not trapped by the narrative that "PoS is the future" but chose the most suitable technical solution based on their own needs. This ability to think independently is too rare in this industry filled with FOMO and FUD. Perhaps true innovation is not about chasing the latest technology but finding the most suitable technology. Final thoughts. As I finished writing this article, I suddenly thought of a question: Ten years from now, when we look back at the blockchain industry of 2025, how will we evaluate Irys's choice? Perhaps at that time, everyone will say: "Of course, using PoW for data chains is common sense; what’s there to discuss?" But in 2025, this choice requires tremendous courage. Because you have to withstand the doubts of being "backward" and "retro" and stick to what you believe is the right direction. This reminds me of a saying: truth is often held by a minority. Not because the minority is smarter, but because they are willing to think independently, willing to question consensus, and willing to bear the risk of being misunderstood. Irys has done this. And this may be more valuable than the technology itself. Written on September 30, 2025. In this era where everyone is embracing PoS, I salute those who dare to stick to their judgment.