BEIJING / INNER MONGOLIA — October 30, 2026. China’s artificial-intelligence infrastructure race is moving into the wind-rich plains of Inner Mongolia, where renewable-energy company Envision Group is working to develop one of the country’s largest planned AI computing campuses.
According to Bloomberg’s October 30 report, Envision aims to complete its massive data-center project by 2028, turning a region traditionally associated with energy and agriculture into a critical part of China’s technological ambitions.
The project, called Envision Galaxy Campus, is located in Ulanqab, approximately 350 kilometers northwest of Beijing.
It is designed to expand toward 2 gigawatts of power capacity, with renewable energy playing a central role in its operation.
Envision announced in August that an initial 120-megawatt phase had begun operations, with capacity allocated to two Chinese technology companies.
But the larger campus remains a work in progress.
The central question is whether Envision can convert the region’s abundant electricity, available land and relatively cool climate into a sustainable advantage in the global AI race.
If the project succeeds, China could gain a major new source of AI computing power. If construction, chip supply or customer demand falls short, the enormous expansion could become another test of whether the world’s AI infrastructure boom is growing too quickly.
Envision’s Inner Mongolia data center targets 2028 completion
Envision’s Galaxy Campus is designed to grow into a gigawatt-scale AI computing center.
The project’s planned 2-GW capacity would make it exceptionally large by data-center standards.
One gigawatt equals 1,000 megawatts of electrical power.
However, the figure describes the planned scale of the facility, not the amount of electricity it is consuming today.
The company launched an initial phase in August 2026.
Bloomberg reported that the first 120 MW had been allocated to two domestic technology companies.
Envision has also described a 120,000-square-meter facility designed to accommodate extensive computing infrastructure.
The 2028 completion target represents a development timetable, subject to construction progress, equipment availability and the commissioning of additional capacity.
The exact amount of computing hardware currently installed is not publicly established in the material reviewed.
That distinction matters because opening a campus, delivering usable electrical capacity and operating fully equipped AI servers are different milestones.
Why Inner Mongolia is becoming China’s AI infrastructure capital
Ulanqab has emerged as an important location for Chinese data-center investment.
Its attractions include relatively inexpensive electricity, large development sites, wind-energy resources and cool weather that can reduce some cooling requirements.
The city is also close enough to Beijing to support high-speed data connections with China’s major technology industry.
China’s national strategy to move computing workloads from eastern cities to western regions has encouraged projects in areas with greater access to energy and land.
WWC ONE MEDIA previously explored the transformation in China Is Building an AI Data-Centre Empire in the Middle of Nowhere — But There’s One Problem Beijing Can’t Easily Fix.
That report examined Ulanqab’s rapid infrastructure growth and the practical difficulties of converting large data-center developments into productive AI computing capacity.
The location offers potential advantages, but cheap land and electricity are not sufficient on their own.
Data centers also need processors, networking infrastructure, cooling systems, trained personnel and customers willing to pay for computing services.
Envision wants renewable electricity to become an AI advantage
Envision’s approach differs from that of many conventional data-center developers.
The company began as a wind-turbine manufacturer and expanded into energy storage, energy-management systems and other clean technologies.
Its Galaxy Campus is designed around an integrated system involving renewable generation, storage and power transmission.
The company argues that combining energy and computing infrastructure can reduce operating costs while limiting dependence on conventional grid arrangements.
That strategy addresses one of the largest challenges in the international AI boom: obtaining enough electricity to operate increasingly powerful computers.
Advanced AI workloads can require large amounts of electrical power.
Electricity must also remain dependable because interruptions can disrupt computing jobs and damage service reliability.
Renewable energy can reduce emissions associated with electricity consumption, but wind and solar generation vary with weather conditions.
Battery storage, backup supply and intelligent power management are therefore important components of any claim that a large AI campus can operate reliably on renewable power.
The effectiveness of Envision’s approach will depend on actual operating performance as the facility expands.
The project forms part of a 5-GW global strategy
Galaxy Campus is the first flagship development under Envision’s Mission Gobi initiative.
The company announced the program in June 2026 with an ambition to develop approximately 5 GW of green AI computing capacity in desert and arid regions around the world by 2030.
The strategy is based on the idea that regions with abundant renewable-energy resources can become competitive locations for AI infrastructure.
Envision has also promoted the potential to accommodate very large numbers of AI accelerators and deliver extensive computing performance.
But these numbers describe the company’s planned development capacity.
They should not be presented as verified installed chip counts or fully operating computing output.
Mission Gobi’s 5-GW ambition is also a broader international target, not an additional 5 GW already operating at Ulanqab.
For investors, the important question is how much of the proposed expansion will be completed, connected to reliable electricity and supported by paying customers.
China’s AI race is spreading beyond Beijing and Shanghai
Envision is not alone in investing in Inner Mongolia.
Chinese AI developer DeepSeek has reportedly been planning a large data-center development in Ulanqab.
Bloomberg’s earlier reporting indicated that DeepSeek was considering around 1 GW of computing capacity, with part of the project potentially operating by late 2027 or early 2028.
Other Chinese technology companies have also established or planned facilities in the region.
The Financial Times reported in October that Ulanqab had become one of China’s most important computing clusters, with a substantial pipeline of announced projects.
However, planned capacity should not be treated as equivalent to fully operational facilities.
Some projects may be completed in stages, while others could change in scale depending on financing, demand and access to hardware.
The growing concentration of development activity also creates practical questions about transmission infrastructure, water availability and local economic benefits.
Advanced chips remain a critical obstacle
Electricity is only one ingredient in a successful AI data center.
The computers also need powerful processors capable of training and running advanced artificial-intelligence models.
US restrictions on exports of certain advanced semiconductors to China have complicated access to leading hardware.
Chinese companies are therefore expanding their use of domestically designed processors, including Huawei’s Ascend products.
But domestic chip production, memory availability, networking equipment and software compatibility can all influence the actual performance of AI systems.
The Financial Times identified advanced processor availability as a major potential constraint on China’s ambitious data-center construction program.
The risk is straightforward.
A company could build a large facility and secure sufficient electricity but still struggle to reach its planned computing capacity if equipment cannot be obtained on schedule.
For Envision, the success of its 2-GW ambition will depend partly on whether customers can install and operate the necessary hardware.
The global AI data-center boom is approaching a critical investment test
Envision’s project is part of a worldwide surge in AI infrastructure spending.
Bank of America Global Research has forecast that annual global data-center spending could reach approximately $2.1 trillion by 2028.
WWC ONE MEDIA examined that outlook in Bank of America Predicts $2.1-Trillion AI Data Center Explosion by 2028 — But Could a Global Power Shortage Bring the Entire Boom Crashing Down?.
The projection illustrates the enormous financial commitment required to expand AI infrastructure.
But it remains a forecast rather than money already invested.
Data-center developers face high upfront costs for land, power systems, buildings, servers, cooling and network equipment.
They also need confidence that customers will purchase enough computing capacity to justify those investments.
The profitability of individual projects may vary substantially.
Cheap electricity can improve the economics, but it cannot guarantee full occupancy or strong returns.
Google’s power agreements reveal how important electricity has become
Major technology companies outside China are also securing long-term electricity supplies.
Google, for example, signed large power arrangements with Constellation Energy involving existing nuclear-generation facilities and additional electricity capacity.
WWC ONE MEDIA covered the development in Google Signs Massive 3.6-GW Power Deal With Constellation — But the Real AI Bottleneck Is No Longer Chips.
The strategy differs from Envision’s focus on renewable generation in Inner Mongolia.
But both approaches reflect the same industry reality.
Power availability, grid access and operating costs can determine where AI facilities are built and how quickly they become usable.
Different regions are responding through nuclear power, renewable energy, gas generation, energy storage and grid expansion.
No single energy source can be assumed to provide the optimal solution for every data center.
Local conditions, reliability, emissions and total costs all matter.
Construction speed is becoming a strategic advantage
Competition is also intensifying over how quickly developers can construct and equip AI facilities.
Equipment shortages can delay projects even after financing and power agreements are secured.
Electrical transformers, cooling systems, memory chips and specialized components may have long procurement periods.
WWC ONE MEDIA examined these constraints in Nokia Says AI Data Centers Could Be Built Twice as Fast — But Power and Memory Are Becoming the Real Bottlenecks.
The report discussed the difficulty of matching customer demand with the pace of infrastructure delivery.
Envision’s 2028 target therefore depends on more than completing buildings.
The company and its customers must coordinate power systems, network connections, computing equipment and operational readiness.
A facility can be structurally complete without being able to deliver its full planned AI capacity.
That is why investors should assess completed phases and installed equipment separately from announced project targets.
What could the project mean for the Philippines and Southeast Asia?
The expansion of China’s AI data-center industry could influence technology supply chains and regional infrastructure investment.
The Philippines is seeking to attract more digital infrastructure and technology-related investment, but developers face their own challenges involving electricity costs, transmission reliability and permitting.
The Envision model highlights the potential advantages of coordinating energy and data-center planning.
For Southeast Asian markets, greater access to reliable power could become a competitive factor in attracting AI investment.
However, conditions in Inner Mongolia cannot be directly replicated everywhere.
Renewable-energy resources, climate, land availability, connectivity and customer demand differ across locations.
The project also does not establish a specific new Envision data-center investment in the Philippines.
Any local implications remain potential lessons for infrastructure planning rather than a confirmed domestic development.
Can Envision turn its early launch into a fully operational 2-GW campus?
The next two years will be crucial for Galaxy Campus.
Envision must continue developing the site’s electricity and computing infrastructure.
Its customers must secure processors and other equipment.
The company must also demonstrate that renewable generation and storage can reliably meet intensive computing requirements.
Financial performance will depend on the pace at which customers begin using capacity and paying for services.
The project’s announced ambitions are substantial.
But the difference between commissioned capacity, installed computing hardware and commercially productive workloads will determine its real economic significance.
The bigger question: Will clean electricity decide the next AI superpower?
Envision’s planned Inner Mongolia expansion offers a glimpse of how artificial intelligence and energy infrastructure are becoming increasingly interconnected.
China is using regions with abundant electricity and available land to build computing capacity at enormous scale.
Renewable-energy companies are moving beyond selling electricity equipment into developing entire AI infrastructure systems.
And major technology companies are searching for ways to secure enough power and processors to support increasingly demanding AI models.
But construction announcements do not automatically translate into profitable, operational computing capacity.
The project must overcome hardware constraints, power-management challenges, financing requirements and competitive pressure.
Envision is targeting 2028 for its enormous Inner Mongolia data-center development. But the decisive question is whether China’s combination of renewable energy, rapid construction and domestic AI technology can deliver a lasting computing advantage—or whether advanced-chip shortages and uncertain returns will slow the revolution before it reaches full scale.
