I have had this feeling about the DePIN track in the cryptocurrency ecosystem for the past few years:

The vast majority of projects describe application scenarios and business models that are pseudo-needs.

This is one aspect.

On the other hand, what I find very forced is that many so-called DePIN projects originally intended to utilize idle resources, but as they developed, not only could idle resources not fulfill the tasks assigned by the network, but the projects completely turned into a game of monopoly by major players.

The most typical case in this regard is Filecoin-------from the earliest depiction of being able to participate in storage mining with idle hard drives at home to later completely becoming a competition among storage server players.

Not only Filecoin, but most subsequent DePIN projects are also like this.

Once a DePIN project turns into this, it not only fails to make full use of idle resources but also tends to consume more resources additionally.

Why does this happen?

I think an important reason is:

In many DePIN projects, the idea of connecting idle resources to the DePIN network is that in reality, idle resources cannot complete the tasks assigned to them by the network, or that the actual requirements to execute tasks in such networks impose a very high hardware threshold. In reality, nodes possessing such high-threshold hardware do not typically connect to these types of DePIN networks unless they are specifically aimed at participating in DePIN projects.

Thus, in these types of DePIN projects, resource nodes with low hardware thresholds have almost no chance of obtaining network rewards.

In this regard, Bitcoin mining is significantly different from these DePIN projects.

The computational tasks of Bitcoin can be split and parallelized, thus giving rise to the strategy of mining pools. This strategy allows both high-performance devices and low-performance devices to connect to mining pools to participate in mining. Simply put, high-performance devices receive more rewards, while low-performance ones receive less.

Therefore, in the Bitcoin network, after meeting a certain hardware threshold, nodes, whether high or low performance, have the opportunity to obtain rewards.

In my view, this is an important distinction between the two systems:

Whether nodes with different amounts of resources have fair opportunities to receive network rewards. In the former, nodes with small resources have almost no chance; in the latter, nodes with small resources still have opportunities, although the rewards are just smaller.

Recently, a DePIN project in the AI + Crypto track seems to have followed a path similar to that of Bitcoin, breaking through the predicaments faced by previous DePIN projects.

This project is dphn.ai.

It (currently) is a decentralized network focused on AI inference, aimed at incentivizing systems (servers, clouds, etc.) that run various large models (mainly various open-source large models at present) to join the network to handle AI inference work.

Taking the systems running DeepSeek as an example.

These systems all join the network and become its nodes. Their computing power differs, their work efficiency differs, and the speed at which they produce tokens also varies.

But that doesn’t matter.

Once they join the network, when the network receives a request that needs to process inference tasks with DeepSeek, it randomly selects one node from this group to handle the task. After the task is processed, the system runs a verification mechanism to validate the results. If all goes well, the node receives a reward; otherwise, the node faces a penalty.

When the network allocates tasks, it randomly selects a node. This node may process tasks slowly or quickly, but all have equal opportunities.

However, this approach also requires users to have a certain tolerance for performance feedback. Because if the chosen node for processing tasks is low in computing power, the user will experience longer feedback times; if the selected node is high in computing power, the feedback time will be shorter.

This project has been online for a while now, and its current development momentum seems to be good.

To some extent, this seems to be validating the feasibility of DePIN in this sub-track.

However, as for the project itself, it has several obvious issues:

The biggest problem, in my opinion, is that the threshold is still not high enough, especially there is no strong technical threshold.

The reason it appears outstanding right now is mainly that this market is still relatively small and not very attention-grabbing. But if this model ignites demand on the inference side in the future, a large number of competitors will certainly join. At that time, where will its moat be?

Furthermore, this system currently mainly relies on open-source large models and cannot truly support robust and efficient closed-source large models like those from Anthropic, OpenAI, etc.

Even among the open-source large models it currently supports, more and more are narrowing the scope and extent of their openness. If this trend continues, in the future, when the capabilities of open-source large models are too weak and closed-source large models do not support it, can this model still continue to operate?

So the impression this project leaves on me is not so much about the quality and investment value of the project itself, but that the project seems to prove that there is a possible path for DePIN in certain AI + Crypto tracks.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。