OTR EMERGING DATA
Emerging Data
Effect on you
If you sell cooling equipment, Ulanqab’s AI expansion is worth tracking. But existing low-water designs mean a new proposal must demonstrate additional savings, not just point to more data centres.
What changed
A remarkable transformation is taking place in Inner Mongolia. On 8 November 2026, CNA reported from Ulanqab, a Chinese city known for potato farming, where agricultural land is being converted into sites for AI data centres. The report describes farmers selling or giving up access to land for developments serving China's growing demand for computing power. It also raises concerns about water sustainability in an already dry region.
The connection we found
Most people will see another example of China's rapid expansion into artificial intelligence. Our analysis points toward a different connection. AI data centres need more than powerful computer chips. They need land, stable electricity, cooling equipment, communications networks and reliable maintenance. Building in a dry region creates an important question: How can a growing concentration of large computer facilities manage cooling needs without placing unacceptable pressure on local resources? Not all data centres consume large volumes of water. The answer depends on their cooling design, climate and operating methods. CNA reports that Ulanqab’s computing industry used 1.54 million cubic metres of water in 2025, about 0.33% of the city’s water consumption. Low-water systems are already in use. UCloud describes air-cooled chillers, chilled-water circuits and liquid-cooled racks, and claims 92% lower water use than conventional methods. This is a company claim, not an independent measurement of every facility. The opportunity depends on whether an additional improvement saves enough resources to justify its cost. We think the overlooked opportunity may lie in technologies that reduce the resources required to run AI facilities. That makes water-efficient cooling systems, heat-reuse engineering and real-time electricity and cooling monitoring worth investigating. These businesses sell the systems needed to operate computing facilities rather than the AI models themselves.
How it could work
Who could benefit? Possible publicly traded companies to study include: Vertiv (NYSE: VRT), which provides infrastructure and cooling systems for data centres. Schneider Electric (EPA: SU), which supplies electricity-management and cooling-related systems. Ecolab (NYSE: ECL), which provides industrial water-management technologies. These businesses operate in relevant markets, but the retrieved article does not establish that they supply the Ulanqab projects or that their shares are undervalued. Chinese domestic competitors may also be important. Local supplier awards and purchasing records have not been verified here. The hidden commercial test An operator selling a cooling upgrade must prove that the customer saves more money over time than the installation costs. For example, consider a hypothetical S$500,000 cooling project that reduces electricity and water-related operating expenses by S$150,000 annually. Its simple recovery period would be approximately 3.3 years before financing, maintenance changes and other costs. That calculation becomes attractive only if independently measured savings support it. No such project-level saving has been verified for Ulanqab.
Research context
Publication and review times
Published
Research review No fixed date
Key points
A remarkable transformation is taking place in Inner Mongolia.
On 8 November 2026, CNA reported from Ulanqab, a Chinese city known for potato farming, where agricultural land is being converted into sites for AI data centres.
The report describes farmers selling or giving up access to land for developments serving China's growing demand for computing power.
It also raises concerns about water sustainability in an already dry region.
Bottom line
Ulanqab’s data-centre expansion may create work for cooling and monitoring suppliers, but efficient systems are already operating.
This is research and interpretation, not an investment price forecast. Source verification limits are listed below.
Evidence and sources
Review the information supporting this view and the limits on what it can show.
Material claims and sources
Claim sources 1
- Claim id
AIW-F1
- Material claim
CNA reports data-centre expansion on former farmland in Ulanqab and concerns about water sustainability.
- Kind
fact
- Supporting source links
- Original publication time status
unavailable
- Source limit
Displayed 8 November 2026 06:00AM; retrieved clock has no timezone label. Field reporting includes existing low-water cooling and 2025 industry water use. UCloud’s 92% reduction is a company claim. No named-supplier orders or independent project savings established. Checked 8 November 2026; exact retrieval clock not recorded.
Claim sources 2
- Claim id
AIW-F2
- Material claim
CNA reports 1.54 million cubic metres of computing-industry water use in 2025, 0.33% of city consumption, and existing low-water UCloud systems.
- Kind
fact
- Supporting source links
- Original publication time status
unavailable
- Source limit
Claim sources 3
- Claim id
AIW-I1
- Material claim
Cooling upgrades could earn revenue only if measured incremental savings justify total costs compared with existing systems.
- Kind
inference
- Supporting source links
- Original publication time status
unavailable
- Source limit
OTR’s interpretation, not an observed commercial outcome. The report describes some existing cooling systems and annual industry water use, but does not establish independently measured project savings, peak-period requirements, purchasing orders, supplier margins or whether the market already expects strong growth.
Evidence behind this view
Recorded connection
Most people will see another example of China's rapid expansion into artificial intelligence.
Our analysis points toward a different connection.
AI data centres need more than powerful computer chips.
They need land, stable electricity, cooling equipment, communications networks and reliable maintenance.
Building in a dry region creates an important question: How can a growing concentration of large computer facilities manage cooling needs without placing unacceptable pressure on local resources?
Not all data centres consume large volumes of water. The answer depends on their cooling design, climate and operating methods.
CNA reports that Ulanqab’s computing industry used 1.54 million cubic metres of water in 2025, about 0.33% of the city’s water consumption. Low-water systems are already in use.
UCloud describes air-cooled chillers, chilled-water circuits and liquid-cooled racks, and claims 92% lower water use than conventional methods. This is a company claim, not an independent measurement of every facility.
The opportunity depends on whether an additional improvement saves enough resources to justify its cost.
We think the overlooked opportunity may lie in technologies that reduce the resources required to run AI facilities.
That makes water-efficient cooling systems, heat-reuse engineering and real-time electricity and cooling monitoring worth investigating.
We have not established a separate count of independent evidence clusters for this card. Several reports may rely on the same event or data provider.
Source ledger
Source references in the research point to these original reports.
| Reference | Source | Publication and retrieval |
|---|---|---|
| S1 | China's potato capital cultivates AI data centres CNA Displayed 8 November 2026 06:00AM; retrieved clock has no timezone label. Field reporting includes existing low-water cooling and 2025 industry water use. UCloud’s 92% reduction is a company claim. No named-supplier orders or independent project savings established. Checked 8 November 2026; exact retrieval clock not recorded. |
Consequences and economics
Follow how this part of the research connects to the wider idea, including its limits.
Who this affects
- Intended audience
Data-centre operators, cooling engineers and investors researching resource-efficient infrastructure.
- Geography
Ulanqab, Inner Mongolia, China; global listed companies are research candidates only.
- Who gains summary
Qualified cooling, water-management and monitoring providers could gain profitable orders; no Ulanqab award verified for the named stocks.
- When this could matter
At project approvals, tenders, equipment awards and verified operating measurements. No fixed return date established.
Calculation and commercial economics
- Assessment status
qualitative
- How it could happen
Operators could purchase cooling upgrades when independently measured resource savings justify installation, financing and maintenance costs.
- Limits
The S$500,000 project and S$150,000 annual saving are assumptions. Include financing, maintenance, downtime, performance risk and existing-system comparisons. No observed unit economics or buyer commitment established.
Indirect effects
How this effect would happen
More AI facilities could increase demand for cooling systems, electrical equipment and specialised engineers. Firms supplying these systems may see additional orders if construction plans are completed.
Third-order effects
How this effect would happen
If local authorities place greater emphasis on water conservation, facility operators may adopt systems that use less water or recycle heat. Companies providing these technologies could gain a competitive advantage.
Fourth-order effects
How this effect would happen
Successful cooling improvements could make other dry or remote regions more attractive for data-centre development, opening opportunities in construction, electricity and communications infrastructure.
Fifth-order effects
How this effect would happen
As more regions compete to host AI facilities, the cost and reliability of electricity and cooling could become important factors in where technology companies invest. Locations that provide reliable infrastructure at lower operating costs may attract more projects than locations offering land alone.
Action and timing
What to do today
- Specific action
Track new data-centre construction, cooling equipment tenders, water-use requirements and confirmed supply contracts. Smaller engineering firms could seek partnerships with established suppliers rather than build their own cooling equipment. Before investing, compare the suppliers' existing earnings, share prices, debt, customer exposure and confirmed orders.
- Timing
- Costs and practical limits
The S$500,000 project and S$150,000 annual saving are assumptions. Include financing, maintenance, downtime, performance risk and existing-system comparisons.
- When this would not be suitable
Unsuitable for committing capital before paid demand, delivery capability and total costs are established. Data-centre developers use low-water cooling methods that require little new equipment, construction slows, cheaper local competitors take the contracts, or electricity and water savings do not justify installation costs.
Challenge and uncertainty
Check the opposing evidence and other explanations that could weaken this view.
What could change this view?
- Project savings, peak water needs, supplier margins and stock valuations remain unverified. No named company is confirmed as a project winner.
Evidence against this view
Recorded opposing evidence and risks
Counter-evidence and limits
- Counter-evidence summary
The report describes some existing cooling systems and annual industry water use, but does not establish independently measured project savings, peak-period requirements, purchasing orders, supplier margins or whether the market already expects strong growth.
- Alternative explanations
Existing air and liquid cooling may already be efficient; local competitors may meet the need at lower cost.
- What would prove this wrong
Data-centre developers use low-water cooling methods that require little new equipment, construction slows, cheaper local competitors take the contracts, or electricity and water savings do not justify installation costs.
- What the sources actually establish
CNA, China's potato capital cultivates AI data centres, published 8 November 2026, 6:00am, timezone unlabelled in retrieved text. Ulanqab, Inner Mongolia, China. Firsthand reporting on development, not proof of specific equipment orders. Vertiv, corporate source for its product range. Specific 2026 Ulanqab contracts not verified. Schneider Electric, corporate source for its product range. Specific 2026 Ulanqab contracts not verified. Sources checked November 8.
Company-level contracts, valuations, ownership changes and complete historical database comparisons were not verified. Source publication clocks are unavailable unless recorded in the source list. Consequences are conditional hypotheses.