ULANQAB, INNER MONGOLIA — A remarkable transformation is sweeping across China’s northern grasslands, where fields that once produced millions of tonnes of potatoes are being replaced by enormous data centres powering Beijing’s artificial intelligence ambitions.
For decades, Ulanqab in Inner Mongolia was known as China’s potato capital.
Now, it is emerging as one of Asia’s fastest-growing artificial intelligence infrastructure hubs, attracting technology investment, reshaping rural livelihoods and raising questions about whether its fragile water resources can support the next phase of China’s AI revolution.
According to Chinese state news agency Xinhua, Ulanqab has signed 109 computing-centre projects, while artificial intelligence workloads accounted for more than 90% of the city’s computing capacity as of August 2026.
Technology companies are drawn to the region by relatively inexpensive electricity, abundant land, a cool climate and access to renewable energy.
But the development boom is creating a striking contradiction.
The region’s natural environment makes it attractive for cooling electricity-hungry computers, yet its long history of drought creates potential risks as more facilities are built.
A November 8 investigation by Channel News Asia reveals how the transformation is changing the lives of farmers, food vendors and rural communities.
And behind the extraordinary construction boom lies a question that could determine whether China’s AI expansion remains sustainable.
Can Beijing build enough computing power to compete globally without placing unacceptable pressure on the land, electricity and water resources that make the industry possible?
From Potato Fields to Artificial Intelligence: The Reinvention of Ulanqab
For 64-year-old potato farmer Zhang Zhiyong, the transformation of his hometown is deeply personal.
Zhang has spent more than four decades working farmland in Shengjiaying, a village near Ulanqab’s Jining district.
Growing potatoes was not simply his occupation.
It was the livelihood his family had known for generations.
But approximately two years ago, developers began approaching villagers to acquire agricultural land for data-centre projects.
Zhang eventually surrendered about 0.7 hectares of his more than 1.3 hectares of farmland.
He told CNA that he received more than 200,000 yuan, equivalent to approximately US$30,000, in compensation.
Today, Zhang cultivates a much smaller plot, while renting out much of his remaining farmland to other farmers.
The compensation has provided financial flexibility and allowed him to reduce his workload.
Yet he has no intention of abandoning agriculture completely.
His experience illustrates how China’s AI infrastructure expansion is reshaping local economies.
Land that once produced agricultural crops is now being used to support digital services, cloud computing and artificial intelligence.
For some residents, that means new income opportunities.
For others, it means losing access to farmland or grazing areas that have supported livelihoods for decades.
China’s Potato Capital Is Becoming an AI Powerhouse
Ulanqab’s agricultural reputation was built on enormous potato production.
According to a 2023 report by People’s Daily, the city produced nearly four million tonnes of potatoes annually, representing approximately 5% of China’s national output.
The region’s dry climate and extensive agricultural land helped establish it as a major farming centre.
But today, large-scale data-centre developments are changing its economic identity.
Xinhua reported in October 2026 that Ulanqab had signed agreements involving 109 computing-centre projects.
That figure represents signed projects, not 109 fully operational facilities.
As of August, AI-oriented computing represented more than 90% of the city’s computing capacity.
A separate August report by WIRED, citing Goldman Sachs research, said nearly 100 data centres had opened or begun construction in Ulanqab since 2016.
The publication also reported estimated commitments totalling 12.5 gigawatts of capacity.
However, that number describes planned and committed infrastructure capacity, not fully installed and continuously operating AI hardware.
The distinction is important.
A data-centre project can be announced, funded, under construction or operational.
Those are different stages of development.
China’s challenge is turning large-scale physical infrastructure into computing services that businesses and AI developers can actually use.
Why China’s Biggest Technology Companies Are Moving Into Inner Mongolia
Ulanqab’s rise is closely connected to Beijing’s national computing strategy.
Known as East Data, West Computing, the initiative seeks to shift suitable computing workloads from heavily developed eastern regions into western areas with greater access to land and energy.
Cities such as Beijing, Shanghai and Shenzhen remain major centres of technology development.
But building additional data centres near these crowded urban markets can be expensive.
Electricity, land and cooling infrastructure are major operating costs.
Inner Mongolia offers several advantages.
Ulanqab is approximately 350 kilometres northwest of Beijing, allowing developers to connect with a major technology market while locating infrastructure in a less densely developed region.
Its climate can significantly reduce mechanical cooling requirements.
The region also has substantial wind and solar resources.
According to CNA, the Chahar High-Tech Development Zone contains facilities operated by major technology companies including Huawei, Alibaba and Apple.
China’s leadership is closely monitoring the expansion.
President Xi Jinping visited the zone on October 16, examining green computing infrastructure and calling for greater coordination among computing capacity, electricity supply and water resources.
The visit underscored the strategic importance Beijing places on AI infrastructure.
Cheap Electricity Could Be China’s Biggest Competitive Advantage
Artificial intelligence requires enormous amounts of computing power.
That computing power depends on processors, networking equipment, cooling systems and continuous electricity supplies.
As AI models become more demanding, the cost of operating data centres becomes increasingly important.
Inner Mongolia’s electricity resources offer a significant potential advantage.
According to an October Xinhua report, electricity prices for some computing facilities in the region can be as low as approximately 0.35 yuan per kilowatt-hour.
That can be substantially cheaper than electricity in China’s major eastern cities.
The region’s energy resources include wind, solar and conventional power generation.
This combination allows developers to pursue different energy-supply arrangements while trying to maintain reliable operations.
Xinhua reported that Ulanqab’s computing industry consumed approximately 4.3 billion kilowatt-hours of electricity in 2025.
Consumption was estimated to reach approximately 9 billion kilowatt-hours in 2026.
Those figures illustrate how quickly the sector’s energy requirements are expanding.
However, access to renewable electricity does not mean every data centre operates entirely on renewable power.
Actual emissions depend on the electricity mix, operating conditions and the source of energy used by each facility.
As previously reported by WWC ONE MEDIA in its analysis of China’s AI data-centre expansion in Inner Mongolia, cheap electricity and rapid construction are important advantages.
But developers still need sufficient processors, customers and reliable infrastructure to make their investments productive.
A 2-Gigawatt AI Campus Shows the Scale of Beijing’s Ambitions
Ulanqab is also attracting developers planning exceptionally large computing campuses.
Envision Group, a renewable-energy technology company, is developing its Galaxy Campus in Inner Mongolia.
According to Bloomberg’s October 30 reporting, the project is designed to expand toward two gigawatts of power capacity.
Envision aims to complete the development by 2028.
An initial 120-megawatt phase began operating in August 2026, with capacity allocated to two Chinese technology companies.
The larger planned facility remains under development.
The proposed scale demonstrates how AI infrastructure is becoming increasingly interconnected with energy projects.
Developers must secure not only land and buildings but also substantial electricity supplies.
Renewable generation, energy storage and grid infrastructure may all be necessary to support growing demand.
As WWC ONE MEDIA detailed in its report on Envision’s proposed 2-gigawatt AI campus, the project represents a major test of whether clean-energy resources can support high-density AI computing.
But planned capacity is not the same as operational capacity.
The ultimate economic value of the development will depend on whether customers install computing equipment, use the services and generate sufficient revenue.
The AI Boom Is Already Changing Local Businesses
The transformation of Ulanqab is visible beyond data-centre construction sites.
Local food vendors are finding new customers among the thousands of workers involved in building and equipping computing facilities.
At the Bayin data-centre cluster in Chahar, 46-year-old food vendor Gao Lina sells pork rib rice to workers.
She told CNA that she serves approximately 100 portions daily, often selling out by lunchtime.
Her business is one of around 30 roadside food operations observed by CNA in the surrounding area.
Some vendors serve customers from vehicles.
Others operate temporary stalls or tables near construction sites.
The demand reflects the labour-intensive nature of building large data-centre campuses.
Construction workers need meals, transportation and everyday services.
But the economic benefits can change as projects are completed.
At the Yiwutang data-centre cluster, another vendor reported that customers increasingly consisted of engineers rather than construction workers.
This transition illustrates a critical economic issue.
A construction boom can create temporary demand for local businesses.
Once facilities become operational, the workforce may be smaller and require different skills.
For Ulanqab, long-term economic development will therefore depend on more than the construction of server buildings.
The city must also develop technical employment, training opportunities and related industries.
Can Inner Mongolia’s Water Supply Keep Up With the AI Boom?
Despite the economic opportunities, water availability remains one of the most important concerns.
Ulanqab has a history of severe drought.
Residents interviewed by CNA recalled periods when household wells ran dry and farmers struggled to grow crops.
In June 2011, drought affected more than 23,000 square kilometres of Ulanqab.
Approximately 15,000 residents in Chahar Right Middle Banner experienced difficulties accessing drinking water.
Since then, authorities have invested in water infrastructure.
A 163-million-yuan project completed by early 2023 included six wells, a water-treatment plant and 45 kilometres of pipelines.
Additional rural water-supply improvements were announced in 2024.
Residents interviewed by CNA said household water access had improved.
Importantly, they did not report widespread current water shortages directly caused by the data-centre expansion.
Nevertheless, the region remains vulnerable to drought.
Municipal agricultural monitoring in May 2026 identified approximately 491,000 hectares of farmland experiencing varying degrees of drought.
The concern is therefore not that data centres have already exhausted Ulanqab’s water supply.
It is whether rapid future expansion could increase pressure during periods of limited availability.
Data Centres Currently Use Just 0.33% of Ulanqab’s Water
One of the most significant findings in CNA’s investigation challenges the assumption that China’s new AI data centres are already consuming most of the region’s water.
According to the Ulanqab municipal water bureau, big-data enterprises used approximately 1.54 million cubic metres of water in 2025.
That represented just 0.33% of the city’s total reported water consumption.
The figure suggests that current direct industry water demand remains relatively small compared with total municipal consumption.
But it does not eliminate future risks.
A sector’s percentage of total water use can remain low while still creating pressure in specific locations or during dry periods.
Water availability also changes throughout the year.
An industry that consumes relatively little water annually may face operational challenges when drought, high temperatures and peak demand occur simultaneously.
Experts therefore argue that annual consumption figures should be considered alongside local supply conditions.
The quality and source of water matter too.
Using reclaimed wastewater can have different implications from extracting groundwater or consuming treated drinking water.
For Ulanqab, managing these distinctions will become increasingly important as more computing projects begin operations.
How One Data-Centre Operator Reduced Cooling Water Use by 92%
A major part of Ulanqab’s water strategy involves cooling technology.
Servers produce heat while operating.
If that heat is not removed, equipment can overheat, potentially disrupting services.
Traditional cooling systems may use evaporation to release heat, consuming significant amounts of water.
But Ulanqab’s relatively cool climate creates opportunities to rely more heavily on outside air.
At UCloud’s data-centre campus, operators use air-cooled chillers and chilled-water circuits to regulate server temperatures.
Campus manager Feng Yongcheng told CNA that outdoor conditions allow the facility to reduce its reliance on mechanical refrigeration for approximately eight to 10 months each year.
UCloud said the approach reduces cooling-related water consumption by approximately 92% compared with conventional methods.
That is a company-reported comparison, not proof that every data centre in Inner Mongolia can achieve the same savings.
Actual water use depends on facility design, computing density, cooling technology and environmental conditions.
The approach nevertheless demonstrates how climate can influence infrastructure efficiency.
Locating data centres in cooler regions may reduce the need for water-intensive evaporative cooling.
Why More Powerful AI Chips Could Change the Cooling Equation
There is another complication.
Artificial intelligence servers are becoming increasingly powerful and energy-intensive.
Higher-density computing equipment produces more heat within a smaller physical space.
That can make traditional air cooling less effective.
UCloud has already installed liquid-cooled server racks and plans to expand the technology.
Liquid cooling can transport heat more efficiently from dense computing systems.
But liquid cooling is not automatically equivalent to consuming large amounts of water.
Closed-loop systems can circulate fluid repeatedly with limited direct water losses.
Other designs may rely more heavily on evaporative cooling.
The environmental impact depends on the entire system, including how heat is ultimately released.
That means technology choices matter as much as the label attached to a cooling method.
Researchers are continuing to examine the relationship between climate, water consumption and server efficiency.
Beijing Institute of Technology associate professor Wang Yongzhen told CNA that cooling systems in northern and western China generally have advantages compared with those in hotter and more humid southern regions.
However, he also warned that improvements in efficiency might not fully offset rapid growth in overall computing capacity.
Even if each individual facility becomes more efficient, total industry water consumption could still increase as more facilities open.
Reclaimed Water Could Become a Critical Resource
Ulanqab is exploring ways to make greater use of treated wastewater.
According to municipal data cited by CNA, the city produces approximately 72 million cubic metres of reclaimed water annually.
Around 30 million cubic metres remains available for allocation.
Using reclaimed water could help reduce competition between industrial facilities and households for freshwater supplies.
It could also provide a more sustainable source for certain cooling applications.
Xinhua reported in October that local regulations prohibit data centres from using groundwater as cooling water.
The agency also described facilities that rely on reclaimed water and low-water-consumption cooling technologies.
But future sustainability depends on implementation.
Developers must ensure that suitable reclaimed water is available where facilities operate.
Transport infrastructure, water quality and treatment capacity can all influence whether the resource is practical.
Officials must also account for existing agricultural, industrial and residential demand.
A water resource that appears sufficient at the city level may still be constrained in a specific district.
That is why researchers are calling for closer examination of actual operating data.
Experts Warn That Efficiency Alone May Not Be Enough
Tongji University associate professor Zhong Shifa told CNA that the most important test would occur when several pressures coincide.
Those include unusually high temperatures, low water availability and heavy computing demand.
During such periods, operators may face greater cooling requirements precisely when local water supplies are under pressure.
Zhong emphasized the need to examine individual facilities rather than relying solely on broad regional estimates.
Relevant information includes actual water consumption during peak periods, cooling methods and the combined requirements of existing and planned projects.
Ulanqab water bureau director Yang Yong has also warned against dismissing resource concerns simply because data centres currently account for a small share of total consumption.
His caution reflects the difference between existing environmental conditions and the demands of a rapidly expanding industry.
A sustainable development strategy must consider how much water remains available during difficult periods, not only how much is consumed in an average year.
China’s AI Expansion Faces Another Bottleneck: Advanced Chips
Water and electricity are not the only challenges facing China’s AI infrastructure strategy.
The country’s technology companies also need advanced processors to turn data centres into productive computing facilities.
US restrictions on exports of sophisticated semiconductors have complicated China’s access to some of the most capable AI chips.
Domestic manufacturers are working to expand their capabilities.
But constructing a data-centre building does not guarantee that it can immediately be filled with advanced computing equipment.
This creates a potential mismatch between announced infrastructure and usable computing capacity.
An October report in the Australian Financial Review highlighted the extraordinary speed of China’s data-centre construction.
An earlier WIRED investigation similarly documented the enormous scale of proposed capacity in Ulanqab.
But analysts continue questioning whether developers can secure enough processors and customer demand to support every project.
The challenge is economic as well as technological.
An expensive computing campus must generate sufficient revenue to justify its development and operating costs.
Unused or underutilized facilities could become a financial burden.
This is why China’s AI strategy must be evaluated through actual computing output and commercial demand—not simply the number of data centres being built.
Is China Spending Too Much on Artificial Intelligence?
The rapid construction of data centres has also raised questions about the concentration of investment in artificial intelligence.
Companies are committing substantial resources to chips, electricity infrastructure, cooling systems and networking equipment.
Governments are supporting developments through industrial policies and infrastructure planning.
But enthusiasm for AI does not guarantee that every project will generate adequate returns.
As WWC ONE MEDIA reported in its coverage of China’s warning about concentrated AI investment, Chinese financial regulators have raised concerns about capital flowing rapidly into technology companies and potential financial vulnerabilities.
Those warnings do not mean the entire AI sector is experiencing a financial bubble.
They highlight the importance of examining debt, investment concentration and long-term profitability.
For Ulanqab, the central question is whether planned facilities will become productive assets or whether construction could outpace demand.
The Global AI Race Is Becoming a Battle Over Natural Resources
China’s experience reflects a global infrastructure challenge.
Artificial intelligence is no longer simply a competition to develop advanced software.
It increasingly depends on access to land, electricity, semiconductors, cooling technology and reliable communications networks.
Countries with inexpensive and dependable electricity may enjoy advantages.
Regions with cooler climates can potentially reduce cooling costs.
But land use, water availability and environmental impacts can create political resistance.
That resistance is becoming visible beyond China.
As WWC ONE MEDIA previously reported on the growing backlash against AI data centres across Europe, Asia and Africa, communities have raised concerns about electricity demand, water consumption, construction activity and the cost of infrastructure upgrades.
In some locations, opposition has delayed or complicated proposed developments.
Inner Mongolia’s experience is different because state-led infrastructure planning plays a significant role in the construction process.
But even rapid development cannot remove physical constraints.
Electricity must be generated.
Water must be managed.
Land must be allocated.
And communities must adapt to economic changes.
The global AI race is therefore also becoming a contest over how efficiently countries manage essential resources.
Farmers and Herders Face a Different Kind of Future
The transformation of Ulanqab is not experienced equally by everyone.
For Zhang Zhiyong, giving up farmland provided compensation and reduced the physical demands of farming.
For food vendor Gao Lina, construction activity created opportunities to earn more money.
But for others, the changes are more complicated.
CNA interviewed a 50-year-old sheep farmer who identified himself only as Hu.
He maintains a flock of approximately 40 sheep near a data-centre construction site.
As development expands, land previously used for grazing has become less accessible.
Hu said the surrounding environment had changed considerably.
He also acknowledged benefits, including relatives finding employment connected to construction.
His experience illustrates the trade-offs involved in economic transformation.
New industries can create jobs and income while disrupting traditional activities.
A farmer may receive compensation but lose productive land.
A food vendor may experience strong business during construction but weaker demand once a project is completed.
A livestock owner may find grazing areas reduced even without facing a direct shortage of drinking water.
The long-term success of Ulanqab’s transformation will therefore depend on whether the new economy creates lasting opportunities for local residents.
Why Inner Mongolia’s AI Boom Matters to Southeast Asia
Although Ulanqab is thousands of kilometres from Southeast Asia, its development offers lessons for countries seeking to attract artificial intelligence investment.
Singapore, Malaysia and the Philippines are among the regional economies examining opportunities connected to data centres, cloud computing and digital infrastructure.
But the conditions differ substantially.
Inner Mongolia benefits from relatively cool weather, available land and major energy resources.
Tropical markets face higher ambient temperatures and different electricity-system challenges.
That can increase the importance of efficient cooling and reliable power supply.
Water availability is also a growing consideration.
In Malaysia, the expansion of data centres in Johor has prompted concerns among some residents about construction activity, electricity infrastructure and changes in nearby communities.
As WWC ONE MEDIA reported in its coverage of Johor’s data-centre boom, some homeowners in Iskandar Puteri have questioned how new developments could affect their surroundings.
The circumstances are not identical to Inner Mongolia.
But both cases show why environmental planning and community considerations must accompany large technology investments.
For the Philippines, the broader lesson involves the importance of dependable electricity, grid capacity, investment planning and responsible resource management.
China’s model demonstrates the possible advantages of coordinating power development with digital infrastructure.
It does not establish that the same approach can be replicated directly in every country.
Could Inner Mongolia Become the World’s Next AI Infrastructure Superpower?
Ulanqab’s rapid transformation is remarkable.
A city once associated primarily with potatoes and grasslands is becoming an important part of China’s national computing network.
Major technology companies are expanding their presence.
Farmers are receiving compensation for land.
Construction workers and food vendors are finding new economic opportunities.
And renewable-energy resources are being integrated into increasingly ambitious infrastructure plans.
But physical scale alone does not guarantee global technological leadership.
China must still secure the processors, networking infrastructure and commercial customers needed to turn capacity into productive computing.
It must also demonstrate that rapid development can remain compatible with local water supplies and environmental conditions.
That combination presents a demanding challenge.
The country’s advantages in construction speed, manufacturing and energy infrastructure are substantial.
Its constraints involving advanced chips, future demand and resource sustainability remain significant.
The outcome will depend on how effectively these competing factors are managed.
The Bottom Line: China’s AI Empire Is Rising, but Its Most Important Test May Be Beneath the Surface
China’s transformation of Ulanqab into a major artificial intelligence computing hub is reshaping both the country’s technology infrastructure and the lives of local residents.
According to Xinhua, 109 computing-centre projects have been signed in the city.
AI-related processing accounted for more than 90% of its computing capacity as of August 2026.
The expansion has brought investment, business opportunities and new employment prospects.
But it has also affected agricultural land and traditional grazing areas.
Environmental sustainability remains a critical concern.
Big-data enterprises accounted for only 0.33% of Ulanqab’s reported water consumption in 2025.
UCloud says its cooling approach can reduce water use by 92% compared with conventional methods.
Those figures suggest that efficient technologies and careful water management can help address some of the industry’s environmental challenges.
However, researchers caution that total consumption could rise as more facilities begin operating.
Periods of high temperatures, drought and heavy computing demand could create additional pressure.
At the same time, China must ensure that its enormous data-centre investments translate into productive computing services rather than underused infrastructure.
For Beijing, Ulanqab represents an opportunity to transform traditional economic strengths into a competitive advantage in artificial intelligence.
For local communities, it represents a future in which farmland and grasslands increasingly coexist with vast server campuses.
And as China races to construct one of the world’s most ambitious AI infrastructure networks, one question remains: Can Inner Mongolia provide enough electricity, water and advanced computing hardware to sustain Beijing’s digital ambitions—or could the very resources behind its remarkable rise become the biggest obstacles to China’s AI revolution?
