Crypto mining is not dead, it's just hidden in the office buildings of Shanghai.

Author: Liu Honglin

Many people’s impression of cryptocurrency “mining” is still stuck in the era of Bitcoin’s “moving with the water and grass” — using wind power in the northwest during winter and hydropower in the southwest during summer. Thousands of machines are often crammed into metal houses in the desert, built by rivers in Sichuan, roaring day and night, consuming electricity like a mountain flood.

But the reality is that what we see more in the industry now is a kind of “lightweight mining”: not relying on hydropower, not going deep into the mountains, but quietly running a few devices in office buildings in the city, without the roar of fans or the smell of burnt circuit boards, just silently “computing” and silently producing Tokens.

Due to work reasons, Lawyer Honglin often interacts with Web3 project parties, developers, and investors in Shanghai and Shenzhen. Many familiar friends take me to visit their offices, pointing at a bunch of hardware machines to introduce me, saying, “This is our cryptocurrency mining farm.”

Outside the room is China’s most decentralized financial center, bustling with activity. Inside the room, machines operate silently, without any sound or noticeable change in temperature, supporting decentralized finance and dreams.

This method of “lightweight mining” is actually a state that has naturally evolved within the industry under high regulatory pressure in recent years. On one hand, due to policy risks, large-scale deployments are no longer sustainable; on the other hand, as many new projects abandon the Bitcoin-style PoW route in favor of lower-power PoS, distributed storage, and edge computing mechanisms, the physical form of mining itself has also become “invisible.”

From the perspective of compliance, this is actually a typical “invisible” state - device compliance, network compliance, and running nodes themselves are not illegal, but its income methods and incentive logic do belong to the category of cryptocurrency. If you want to say that this is not mining, it seems that it can’t be completely brushed off; You want to say that it is illegal, but it lacks the characteristics of substantive illegality. This gives the industry a subtle space to survive: it continues to run in the gray area, not big, not small, not noisy, but still alive.

To truly understand this reality, we must start with China’s regulatory path regarding “mining.”

In May 2021, the Financial Stability Development Committee of the State Council explicitly stated during a meeting: “We need to crack down on Bitcoin mining and trading activities.” Subsequently, a systematic “clean mining” campaign was launched across the country. Traditional “mining areas” such as Xinjiang, Inner Mongolia, and Sichuan were the first to respond, successively issuing power restriction notices and shutting down mining sites. In September of that year, the National Development and Reform Commission officially included “virtual currency mining activities” in the “Elimination Category” of the “Guidance Catalogue for Industrial Structure Adjustment,” thus establishing the policy direction.

The official reason given is that such activities “consume a lot of energy, produce high carbon emissions, and contribute little,” which does not align with national industrial policies and the “dual carbon” goals. At that time, this characterization had a certain basis in reality. The PoW mechanism dominated by Bitcoin was indeed a representative of high energy consumption and high density, with electricity usage once exceeding that of some medium-sized countries, much of which came from “gray” power sources.

However, with the evolution of industry technology, many crypto projects no longer rely on the PoW algorithm, but instead maintain the network through methods such as PoS, DPoS, and distributed storage. The computational resources required under this model are significantly reduced, and the deployment scenarios gradually shift from “shanty houses in the suburbs” to “office buildings in the city.” You can call it mining, but it really doesn’t consume much electricity.

What’s more complicated is that the rapid rise in AI development and computing power demand has turned some underlying facilities originally belonging to the crypto industry into “policy-supported targets.” Edge computing, distributed storage, and general-purpose GPU nodes—these technologies, which were once the foundational infrastructure for blockchain applications, are now being “acquired” by the AI industry. Additionally, at the level of computing power and architecture, the boundaries between the two are inherently unclear—running an AI training model and running an on-chain verification node may use the same set of servers; it’s just that the software and objectives called upon are different.

This raises a very real question: the identification logic that regulatory agencies are accustomed to using, such as “Is the power consumption excessive?”, “Is the equipment special?”, “Is it deployed in a concentrated area?”, has almost become ineffective today. You can’t tell which project is engaging in legitimate AI computing business, which project is using a shell to mine tokens, and which project is doing both. Reality has long since blurred the boundaries of regulation.

So often, what we see is not “mining is being revived,” but rather “it never died, it just changed its appearance.” You will see many Web3 projects that superficially promote AI collaboration and edge node scheduling, but when implemented, they are still running the verification logic of a certain chain; there are also projects that, under the guise of data security and encrypted computing, are actually just building their own token issuance mechanisms.

For local governments, this situation is equally tricky. On one hand, there is a clear prohibition from the central government on “mining,” while on the other hand, there is significant support for areas like “computing power infrastructure” and “AI large model training.” If a project’s business model straddles both lines, whether to support it, how to regulate it, and whether it constitutes a violation, there are actually no clear answers.

This ambiguous state further leads to many projects in reality being “run if they can, hide if they can,” which in turn gives rise to a more covert, mixed, and flexible “underground mining ecosystem.” You can’t check it, you can’t calculate it clearly; the electricity is residential, the space is an office, the accounts are compliant, and the entity has a license, but it is still calculating a Token. At this point, if you apply the traditional regulatory logic, it is already out of pace.

As a legal compliance practitioner in the Web3.0 industry, Lawyer Honglin’s personal judgment is that among China’s “three bans” policy on cryptocurrencies (ICO, cryptocurrency exchanges, cryptocurrency mining), if there is indeed room for relaxation in the future, the first to loosen may be “mining.”

It’s not because the country’s attitude has shifted, but rather because the “new miners” have already deviated from the original definition. It’s hard to describe them as “high energy consumption, low contribution” anymore. On the contrary, they might already be the “computing power entrepreneurs” you encourage, holding subsidies from technology parks, participating in AI competitions, seriously registering companies, paying taxes, and issuing salaries, with profits generated not only in Renminbi but also in globally exchangeable tokens.

Moreover, the integration of AI and Web3 is becoming increasingly tight. Many blockchain teams are actually participating in AI model pre-training, data labeling, or algorithm optimization; while many AI companies have also realized that on-chain incentive mechanisms are more efficient in “crowdsourced computing” and “edge participation.” At this point, forcibly separating the relationship between Web3 and computing power will only become more and more unrealistic.

Of course, I am not saying that regulation should be completely relaxed, but rather that we must acknowledge that the shape of this industry has indeed changed, and we can no longer use the standards from three years ago to govern the reality five years later. Especially in the “blurred areas” involving computing power infrastructure and AI service capabilities, what needs to be done may not be a complete denial, but rather to clarify which behaviors should fall under the category of the data industry, which behaviors belong to the financial regulatory objects, and which behaviors can operate in compliance but must be registered and reported through a “positive list + industry classification” approach.

Otherwise, if we forever equate the term “mining” with illegal and backward practices, we will indeed miss out on a part of the future.

Mining, as of today, is no longer just a compliance issue or an energy issue; it is also a question of “how we understand the evolution of infrastructure.” From Bitcoin’s “hashrate for blocks” to the AI era’s “computing power as a resource,” what we are essentially witnessing is that an increasing number of underlying computing power nodes are becoming the universal interface of the digital society. If the past decade was characterized by “whoever can mine coins can make money,” then the next decade is likely to be defined by “whoever masters flexible computing power will have industrial initiative.”

In this increasingly intense era of global computing power competition, if we cannot build a mining and computing power integration mechanism at home that respects the underlying technological path and can be included in the regulatory view, we are likely to be absent in the next wave of global computing power infrastructure competition.

Instead of blocking it, it’s better to see its true nature clearly; instead of hiding it, it’s better to incorporate it into an open regulatory system. This way, at least it can reduce the concerns of projects that could operate in the sunlight, and also lessen the motivation for gray operations.

This is indeed a new issue that needs to be discussed.