The B-side of American AI: "Working for Chinese Bitcoin miners"

Author: Lin Wanwan

At the end of 2025, a Chinese cryptocurrency equipment company, Bitmain, was placed on the U.S. national security review list.

On November 21, the U.S. Department of Homeland Security launched Operation “Red Sunset” under the guise of national security, putting Bitmain under scrutiny. The accusations are pointed: investigating whether its devices have remote backdoors and whether they could deliver a fatal blow to the U.S. power grid in extreme situations.

Why is a Chinese mining company being pointed to as a possible threat to the US power grid?

This is America's extreme anxiety over core resources. Because at this moment, Silicon Valley is staging the most expensive “silence” in the history of technology.

In the AI data center, tens of thousands of Nvidia H100 GPUs are quietly lying on the ground gathering dust. These chips, priced at $30,000 each and referred to as “industrial gold” by Jensen Huang, should be running at full speed, injecting soul into GPT-5 or Sora, but at this moment—they have no power.

The most advanced assets of humanity are now being choked by the most primitive physical bottlenecks.

The United States is experiencing a level of power shortage that is difficult to comprehend. The gap is 44 gigawatts, equivalent to the entire electricity generation capacity of a moderately developed country like Switzerland. In this country, which claims to be the most technologically advanced, the average waiting time to power a newly built AI data center has now stretched to over 48 months.

The American power grid is like an aging old man.

At the moment when AI giants, holding hundreds of billions of dollars, found themselves in despair without a power outlet, they discovered that their lifeline appeared in the place they least expected - Bitcoin mining farms.

Then Wall Street suddenly realized: the group of people held in their hands the most scarce asset of the AI era - a massive amount of electricity that has already been contracted with energy companies.

But they are realizing: this set of survival rules about “computing power is electricity” was vividly interpreted by a group of Chinese engineers across the ocean ten years ago.

Because the first “power training ground” built for the American AI era has already been completed in China ten years ago, and it was relocated to the United States three years ago due to a ban.

The game between the two shores of the ocean hides inevitability in the randomness. Just as the torrent of the times cannot be diverted, each generation has its own destiny, and every footnote tells us: greatness cannot be planned.

The US Electricity Inherits the “Chinese Heritage”

History always tends to write down the answer first, and then wait for the questioner to appear.

In June 2024, American Bitcoin mining company Core Scientific announced a shocking piece of news to Wall Street: they signed a $3.5 billion agreement with CoreWeave, which claims to be the favored child of Nvidia, to lease the electricity infrastructure originally intended for Bitcoin mining to the latter for training AI models.

These news stories have caused a sensation in Silicon Valley, referred to as the “Power Marriage.” However, across the ocean in China, for those miners and officials who experienced the “5·19” storm back in the day, reading these messages brings a different flavor.

Because the infrastructure used by mining companies like Core Scientific, IREN, and Cipher to house NVIDIA H100 has a significant portion that actually carries Chinese genes.

To some extent, the first round of 'power defense works' in the American AI era is the industrial legacy that fully received the massive diversion of computing power from China.

And the person who inadvertently drew the blueprint is named Zhan Ketuan.

Zhan Ketuann, a typical tech guy who graduated from the Institute of Microelectronics, Chinese Academy of Sciences, originally had a life trajectory that should involve writing code, drawing circuit diagrams, and quietly being a technical expert in a technology park.

Until 2013, Jihan Wu and Micree Zhan founded the company Bitmain.

It is said that Zhan Ketuán only spent two hours reading the Bitcoin white paper. He may not fully understand the future of currency, but he grasped the essence of the mathematics behind it — it is an arithmetic game about hash collisions.

In 2016, Bitmain made a shocking decision in the industry: it placed a massive order for wafers with TSMC. The Antminer S9, equipped with TSMC's most advanced 16nm FinFET process, emerged, which not only represented a miracle of production capacity in the history of chips but also created an unprecedented “thermodynamic furnace.”

In Zhang Kuan's eyes, S9 is a chip; but in the eyes of the State Grid, it is a purely industrial load.

It does not rotate day and night like a factory, nor does it fluctuate with temperature changes. It operates 24 hours a day with a smooth power curve like a straight line, not picky about voltage and indifferent to its origins. From that moment on, a new system was born in the world: electricity, which transformed from a public service into a “B-end raw material” that can be instantly priced, traded, and monetized; electricity, an energy that is difficult to store at a low cost once generated, has embedded its value in a string of numbers in another form; Bitcoin mining began to emerge as an industry: from hydropower in the mountains of Sichuan to wind power on the grasslands of Inner Mongolia, Bitcoin mining machines operate on every piece of land in China where electricity is redundant.

At that time, Jian Ketuann may not have realized that the industrial standard he defined for Bitcoin mining machines inadvertently rehearsed a perfect energy supply solution for the extremely thirsty American AI a decade later.

In the craziest year of 2018, Bitmain alone swallowed 74.5% of the global market share. But that’s not the scariest part; the scariest part is that the remaining share was also entirely monopolized by Chinese individuals. Whether it’s Bitmain's former chief chip designer Yang Zuoxing's Shenma Mining Machine or the ancestor of ASICs, Canaan Creative, they all have Chinese faces.

This is not a global competition at all, but rather a “Chinese engineer civil war” spanning over 2000 kilometers: from the Aobei Technology Park in Haidian, Beijing, to the Zhiyuan in Nanshan, Shenzhen, where 99% of the world's computing power heart beats with a Chinese pulse. An absolute closed loop that is completely locked by the Chinese supply chain, which Silicon Valley has to look up to.

Until May 2021, with a regulatory ban, the roaring sound that had lasted for several years by the Dadu River came to an abrupt halt.

For the country, this marks the end of an energy-consuming industry; but for the industry, it is the beginning of an epic “technological migration.” Thousands of containers are loaded onto cargo ships, crossing the ocean, carrying not only the latest generation of Ant miners designed by Janke Tuan, but also a unique “power survival philosophy” honed in China.

Destination one: Texas, USA.

This place has an independent ERCOT power grid and boasts the freest and wildest electricity trading market in the United States. For this group of “computing power refugees” from the East, it is simply a magnified version of “Sichuan + Inner Mongolia.”

However, when this group of Chinese people actually arrived, the American energy sector was surprised to find: these are not refugees at all, they are clearly a well-equipped “energy special forces” team.

When mining companies were in Sichuan, the mine owners relied on drinking heavily with the power station managers and building relationships to obtain cheap electricity. They signed a kind of “understanding” based on personal connections. However, when it came to Texas in the United States, this logic rapidly evolved into high-frequency trading algorithms.

The electricity prices in Texas fluctuate in real-time, changing every 15 minutes, and in extreme cases can soar from 2 cents to 9 dollars. Traditional Silicon Valley data centers (like Google and Meta) avoid such fluctuations as if they were a plague, accustomed to lying on fixed rates like greenhouse flowers.

But what was the reaction of the members of the “Zhang Ke Tuan”? It was excitement.

They transformed their experience of manually controlling the power on and off back in the day into an automated demand response program. When the electricity price is negative (in Texas, USA, during times of excess wind power, the price can be negative), they operate at full capacity, wildly consuming electric power, to the extent that the power grid even has to pay them to use electricity; when a heatwave strikes and electricity prices soar, they can cut off hundreds of megawatts of load within seconds, “selling” electricity back to the grid, earning a price difference that far exceeds that of mining.

This method of “energy arbitrage” has left veteran power traders in the United States dumbfounded. The reason why current mining giants like Riot Platforms and Marathon can thrive and transition to AI data centers is precisely due to this set of power algorithms brought over from China.

Another major legacy of the Zhan Ketuantuan era is the extreme pursuit of the speed of physical infrastructure.

The traditional construction cycle of American data centers is 2-3 years, which is a meticulous process belonging to elite engineers. However, the “mining circle” does not subscribe to this; their logic is: every second of downtime is a crime against profit.

Thus, on the plains of Texas, a “Chinese speed” that left local builders astonished emerged: no exquisite glass curtain walls, no complex central air conditioning, only huge industrial fans roaring. This “modular, container-style, minimalist heat dissipation” infrastructure solution has compressed the construction period to 3-6 months.

This rugged yet extremely efficient engineering capability was initially mocked by Silicon Valley as “electronic waste”, but has now become highly sought after—because the explosion of AI computing power is too rapid, and organizations like OpenAI cannot wait for three years; they need this “plug-and-play” infrastructure capability right now.

It is obvious that in Silicon Valley, you can buy graphics cards if you have money, but time cannot be bought.

These “times” are the crazy legacy from ten years ago. Back then, in order to mine Bitcoin, Chinese miners and their successors frantically acquired land and built substations in the United States, hoarding the “grid capacity” that is now worth a fortune.

Power quotas are the new hard currency of American capital. The so-called 'inheritance' is not inheriting that pile of silicon scrap, but inheriting the right of access to the power grid.

The reason mining companies can secure hundreds of millions in contracts is that they hold the key to starting the AI era tightly in their hands amidst the current power shortages across the United States.

The Migration Night of the 'Invisible Champion'

These brutal pleasures will ultimately end in brutality.

2018 was a hidden watershed in the history of business. That year, ChatGPT's founder, Sam Altman, was still worrying about the survival of non-profit organizations; Musk had just come back from the brink of bankruptcy, and the computing power in their eyes was still just the docile servers in the data center.

But across the ocean, Jihan Wu and his Bitmain have turned computing power into an industrial giant. They may not understand the future of AI, but that does not prevent them from having mastered the key to the future: how to tame those greedy silicon chips measured in gigawatts.

This is a story about grassroots heroes, national will, and historical jokes. China took seven years, quietly nurturing a giant beast that devoured electricity amidst the rapids and coal fields of the west; and on a summer night in 2021, in pursuit of higher financial security and dual carbon goals, it uprooted it by hand.

To understand how the United States can bend to mining companies in accepting the power explosion of AI today, one must comprehend the “energy drill” that took place a decade ago by the Dadu River in Sichuan, China.

Pull the lens back to August 2019.

That was the most glorious time for Bitmain and also a brief window for China's mining industry to “turn from gray to white.” At that time, the Sichuan provincial government introduced a policy called “Hydropower Consumption Demonstration Zone” to solve the long-standing problem of “wasted water during the flood season” (i.e., when electricity generated from water could not be sent out and had to be wasted).

This is a red-headed document that actually exists in places like Ganzi and Aba in Sichuan.

According to reports from Caixin that year, under this policy, Zhan Ketuán's mining machines were no longer the “black households” hidden in the deep mountains, but had become honored guests helping local power grids to “peak shaving and valley filling.”

At that time, Bitmain actually served as a “supercapacitor” for the energy network in western China. What Jihan Wu takes pride in is not only the 7nm chips but also the ability to instantly convert excess electricity into digital assets.

At that time in China, it held 75% of the global Bitcoin hash rate. From Wall Street to the City of London, everyone who wanted to participate in this game had to look at the face of the Bitmain team and relied on the electricity load from Sichuan and Xinjiang in China.

However, behind this “grayscale prosperity,” there are always two Damocles' swords hanging.

The first point is “financial security.” Regulatory authorities have long realized that this is not just a technological innovation, but a huge channel for funds that operates outside of foreign exchange controls.

The second aspect is “dual control of energy consumption.” With the proposal of the “3060 dual carbon” target in 2020, the flow of every kilowatt-hour has become a political account. The mining industry, characterized by “high energy consumption, low employment, and no physical output,” is destined to be sacrificed on the balance of macro-strategic considerations.

The turning point of history is precisely marked on May 21, 2021.

On that night, the Financial Stability Development Committee of the State Council held its fifty-first meeting, in which a sentence with very few words but significant weight appeared in the meeting transcript: “Crackdown on Bitcoin mining and trading activities.”

This is no longer the past “risk alert” or “development restriction”, but the highest level “zeroing order”.

The following month was the most thrilling 30 days in the history of China's computing power industry. Inner Mongolia took the lead in responding by directly cutting off the power supply to coal-fired mining sites; Xinjiang quickly followed suit, conducting a comprehensive investigation.

The climax occurred in the late night of June 19, 2021.

On this day, the Sichuan Development and Reform Commission and the Energy Bureau issued a notice requiring the cleanup and shutdown of virtual currency “mining” projects. This is famously known in the industry as the “Sichuan Shutdown Night.”

The true video from that night is still circulating online: at a super mining site in Aba Prefecture, as the clock struck midnight, the duty staff, with tears in their eyes, pulled down the circuit breakers of the high-voltage distribution cabinet one after another. The roar of the cooling fans, which had lasted for years and sounded like an airplane taking off, disappeared in an instant.

The indicator lights of millions of mining machines went out at the same time. The world suddenly became terrifyingly quiet, with only the sound of the Dadu River still rushing.

At that moment, the global Bitcoin network's hash rate plummeted by nearly 50%. China decisively severed this industry, which consumes over a hundred billion kilowatt-hours of electricity annually, from the national power grid.

We successfully defended the financial line and freed up valuable energy space. But in the gaps of this grand narrative, an unexpected foreshadowing was buried: we left the electricity, but expelled those who “know how to use electricity” the best.

However, the machines that were cut off from power did not disappear; they began to wander.

In the second half of 2021, the Yantian Port in Shenzhen faced unprecedented congestion. According to descriptions from freight forwarding companies at the time, tens of thousands of containers piled up like mountains, all filled with S19 mining machines dismantled from Sichuan and Xinjiang.

This is a “Dunkirk evacuation” of computing power.

The story returns to the beginning.

In 2024, when ChatGPT takes the world by storm, AI giants suddenly realize: lack of electricity, lack of substations, lack of high-power data centers that can be quickly deployed.

China eliminated “backward production capacity” in the past, but completely packaged and delivered the ability to “build and operate ultra-large-scale, high-energy-consuming computing centers” to the world.

This is a strategic choice concerning the national financial sovereignty, resolutely giving up this high-risk digital stronghold. From a macro-prudential perspective, this was an absolutely correct and necessary strategic action at that time. However, the irony and paradox of history lie in the fact that those massive bubbles and excess computing power that were actively squeezed out and expelled ultimately solidified across the ocean into the most unbreakable foundation of the opponent's power grid and energy system.

But if one thinks that the end of this great migration of computing power is just “the East loses, and the West benefits,” then they are only seeing the chips on the table and not the table itself.

The arms race in AI is essentially an endless consumption of energy by computing clusters, which will ultimately lead to a battle over electricity costs. In this war of consumption, no country has more strategic depth than China.

The United States needs miners as a “flexible load” to patch and prolong life, treating miners as a medicinal agent to cure the “age-related diseases” of the power grid.

But China is different, having the national grid as its central brain. Using ultra-high voltage transmission (UHV), the cheapest clean energy from the west is continuously and low-loss transmitted to the data center clusters in the east, like blood flowing through arteries.

Nevertheless, amidst the torrent of history, Bitmain, the Chinese powerhouse of the computing era, inadvertently became a strategic force reshaping the global energy landscape. They have unintentionally dedicated their skills honed along the banks of the Dadu River to the other side of the ocean, repairing the first circle of the energy fence for the impending American AI era.

The Fate of the 'Pacification' Mining Enterprises

So, have these “former Bitcoin miners” who have been recruited really jumped straight to the top and taken a seat at the AI era's table?

The answer may lie in the calculations of the giants. Have you ever wondered why Microsoft and Google, with their hundreds of billions in cash flow, would really hand over the lifeblood of their power to mining companies? Is it just because they find the timeline for building it themselves too long?

Of course not. The root cause is that they are more wary of the lessons of history than anyone else.

Looking back at business history, on the redwood desks of Silicon Valley tycoons, there actually lies an invisible tombstone, engraved with a once-resounding name: Global Crossing.

This is the infrastructure giant that suffered the most during the 2000 internet bubble. At that time, the elites in the United States firmly believed that the entire world would enter the internet age within a few years, and by then people would need faster and faster internet speeds. In this almost religious fervor, founder Gary Winnick borrowed hundreds of billions of dollars in just a few years and crazily laid down hundreds of thousands of kilometers of fiber optic cables in the deep sea, connecting the Americas, Europe, and Asia.

After the internet bubble burst, the .COM websites only needed to shut down their servers and lay off employees to complete their bankruptcy liquidation. However, infrastructure providers faced a huge asset burden: the fiber optics buried at the bottom of the Pacific Ocean, capable of transmitting trillions of bytes per second, overnight became the most terrifying “zombie assets” in the eyes of shareholders—unsellable, immovable, they could only silently lie in the dark ocean depths, slowly rotting on the balance sheet.

In 2002, Global Crossing collapsed under a debt of 12.4 billion dollars. The most ironic outcome was that Li Ka-shing's CK Hutchison Holdings later wanted to pick up these assets for less than 1% of their value, as if picking up scrap metal.

Global Crossing proved a harsh truth with its own demise: in the early stages of technological change, those who bear the burden of irreversible heavy assets become the primary scapegoats during economic downturns. They believed they held the data arteries of the future world, only to turn themselves into sacrifices for infrastructure.

Today's Microsoft CEO Satya Nadella and Google CEO Sundar Pichai must remember this tombstone more than anyone else.

So, when you look at the financial reports from the past two years, you will find that their core risk control consists of only four words: Asset Isolation.

The capital expenditures (CapEx) of major AI companies are skyrocketing, but every penny is accounted for down to the bone: on one hand, there are GPUs and custom servers, which are relatively “generic” assets that can be quickly turned around, and if that doesn't work out, they can still be sold at a discount; on the other hand, there are data center buildings, cables, and cooling systems, which are typical “specialized heavy assets,” and efforts are made to separate these hardest-to-exit assets.

The real calculation is right here: they want to share that “pit” with others.

AI companies try to use long-term computing power contracts, electricity contracts, and park leases to create a chain that “looks like OpEx operating expenses, but essentially shifts CapEx risks to others.”

For those miners who have been pacified and infrastructure players eager to transform, the giants' pitch is enticing: “You are responsible for pouring money into building factories, you handle the liquid cooling upgrades, and I will take care of signing contracts for electricity. As long as AI becomes a boon of the era, you collect rent as per the contract, while I enjoy business growth and stock price returns.”

It sounds like sharing risks, but upon closer inspection, it feels more like that popular saying: “If friends die, the poor monk survives.”

But what if AI ultimately proves to be another illusion like Global Crossing?

The giants can at most pay a penalty, record an asset impairment, and then exit gracefully, ready to tell the next story. However, it is the group of infrastructure buyers, who thought they had finally “made it to the table,” that will have to face the bank's demand letters and explain how to deal with those factories specially designed for high power density that can only accommodate H100, and nothing else.

Furthermore, some may ask: What if the AI bubble bursts? Wouldn't mining companies just unplug the GPUs and plug the mining machines back in to continue mining coins?

More realistically, most “turning to AI” mining farms do not simply replace hardware at the push of a button: AI data centers use GPU + liquid cooling, while Bitcoin requires extreme cost-cutting with ASIC containers, making the two systems almost non-interchangeable. The capital market has already given you a round of premiums based on “AI infrastructure stocks,” and announcing a return to mining is essentially throwing the valuation anchor from AI back to “high-energy miners,” with the factory still in place, while the story and market value are first liquidated.

So history does not repeat itself, but it always rhymes with the same foot. The optical fibers of the past were buried on the seabed, and today's server rooms stand on the wasteland. The people who foot the bill have changed, but the roles have never changed.

Greatness cannot be planned

Today, in the chess game of AI competition between China and the United States, computing power and electricity are two key winning hands.

Although the United States lags behind China's ultra-high voltage construction speed in terms of power grid construction efficiency, it unexpectedly accumulated a huge “shadow inventory.” When the construction of data centers in Silicon Valley is choked by environmental regulations and supply chain issues, these mining farms can immediately step in to provide power for the training of GPT-5 and GPT-6.

The charm of the business world lies in its unpredictability. All strategic planning is essentially looking at the road through a rearview mirror.

This is a strategic aid that no one anticipated. It was not planned by the policy makers in the White House, nor was it simulated by the Pentagon, but was inadvertently constructed by wandering Chinese engineers and a group of profit-seeking speculators in the chaotic market game.

The world is always filled with 'precise errors' and 'vague truths'. This may be a parable left by business history: greatness can never be planned.