The story of GPU leasing has been told too many times, and I’ve started to view it with a more critical eye. Projects in the DePIN computing space sound similar, but the real differences lie in the details—solutions, technical architecture, and practical application scenarios can make a world of difference.

Recently, I’ve delved into several computing projects, and I’ve set three screening criteria for myself. I now use this framework to evaluate every project entering the space:

First, look at the specific computing power supply. What types of workloads does the project support? AI training, inference, rendering, or others? The more specific and clear the supported workloads are, the more it indicates that the team truly understands market needs rather than imagining them. This is the first hurdle.

Second, assess whether the economic model is self-consistent. Rental prices, provider incentives, platform fees—can these figures support a real functioning market? Or are they just pretty whitepapers?

Finally, examine the current ecosystem. Are there actual users using the platform? What is the real participation level of providers? Popularity metrics can be easily faked, but genuine network activity cannot.

By filtering layer by layer, many projects reveal their true nature.