Hello! I'm your financial analysis assistant. This article about "AI computing infrastructure and the construction industry" is quite extensive, but the core logic is actually quite clear.
To help you understand it easily, I'll first "translate" this long article into plain language and then break it down into five key dimensions to help you grasp the business logic and potential pitfalls behind it.
📝 Summary of Key Points
In one sentence:
The era of AI has arrived, and data centers (computing power centers) have become the new "essential resources." This has brought a new wave of orders to the construction industry, which is currently going through a tough time. However, this benefits are not evenly distributed: leading companies with power, mechanical and electrical capabilities, and advanced construction skills are reaping huge profits, while small and medium-sized enterprises that only build ordinary buildings may get a small share or even suffer losses due to reckless cross-industry ventures. At the same time, the entire industrial real estate industry is undergoing a dramatic shift from the old model of "building factories and collecting rent" to a new model that requires a deep understanding of the industry and technology.
---
🔍 In-depth Analysis: Five Key Dimensions
1. The Trend is Here: AI Has Made Building Data Centers a Priority
People used to think the construction industry was declining because residential and commercial real estate was struggling. But now, AI requires a lot of computing power, which in turn requires data centers. Building data centers involves constructing buildings, installing equipment, and connecting them to the power grid.
- Data speaks for itself: The new contract value for data centers in China's construction sector increased by 328% in the first half of the year, indicating that the giants have smelled the opportunity.
- Essential Change: Building data centers now focuses more on mechanical and electrical systems and power supply, rather than just traditional construction. It's like building a warehouse has turned into building a super-precise "electronics factory."
- Who's Paying? Internet giants like ByteDance, Alibaba, Tencent, and Baidu, as well as foreign companies like Google and Microsoft, are the main buyers. Their capital expenditures for 2026 are expected to reach $886.7 billion, with a significant portion going towards infrastructure.
2. A Harsh Hierarchical Distribution of Benefits
This is the core point of the article: The benefits of AI infrastructure are distributed based on capabilities, not equally among all participants.
- First Tier (Reaping the Benefits): Leading state-owned enterprises (such as China State Construction Engineering Corporation and China Power Construction Corporation)
- Advantages: They not only know how to build buildings but also understand power systems, grid expansion, and complex mechanical and electrical installations.
- Result: They get the large EPC (Design-Procurement-Construction) contracts and the highest profits, with the most influence.
- Second Tier (Getting a Share of the Benefits): Specialized subcontractors
- Advantages: They focus on high-tech areas like liquid cooling systems, clean engineering, and special fire protection.
- Result: Although their contracts are smaller, the technical barriers are higher, and competition is relatively less, so their profit margins are acceptable.
- Third Tier (Suffering or Facing Competition): Small and medium-sized construction companies that only do traditional construction
- Disadvantages: Their main contribution is in building foundations and walls, which account for less than 20% of the cost of data centers. The bulk of the cost goes to equipment and power.
- Result: They are left with low-value tasks like building fences, roads, and facades, resulting in thin profits and fierce price competition.
- Warning Case: Chengdi Xiangjiang, despite winning many large contracts, lost money in the first half of 2026 due to its single business structure or management issues. This shows that winning big contracts doesn't necessarily mean making money.
3. Hidden Pitfalls: Making Money from Internet Giants Isn't Easy
Many construction companies think they'll make easy profits by working with these giants, but there are two major challenges:
- Pitfall 1: Low Profit Margin from Construction:
- Building a 1GW data center costs about 160 billion yuan, with IT hardware and power accounting for over 90% of the cost, and construction only 10%-15%.
- This means construction companies play a supporting role, while equipment and power suppliers reap the bulk of the profits.
- Pitfall 2: High Requirements and Uncertainty from Clients:
- Internet giants demand speed, efficiency, and customization. Their needs can change quickly, and budgets can be slashed.
- Real Cases: Chengdi Xiangjiang encountered two contract changes. Other non-construction companies like Lianhua Holdings and Jingyuan Environmental Protection also tried to enter the computing infrastructure business but ended up terminating their contracts and losing money.
- Expert Warning: Wall Street bear Jim Chanos has noted that the return on investment in AI infrastructure is declining from 40% to 10%. If giants find the investment unprofitable, they may slow down construction, leading to a sharp drop in construction orders for construction companies.
4. Industry Disruption: From "Landlord Thinking" to "Industry-Oriented Thinking"
The rules of the industrial real estate sector (including data centers, semiconductor factories, and new energy plants) have changed dramatically.
- Old Rules (Landlord Thinking): Acquire land → Build generic factories → Rent them out to companies → Collect rent.
- Current Situation: This model is no longer effective. The prices of ordinary industrial REITs (Real Estate Investment Trusts) have dropped because no one wants to rent low-quality factories.
- New Rules (Industry-Oriented Thinking): Deep understanding of the industry is essential for customized development and partnerships with leading companies for shared growth.
- Changes: Industrial real estate now requires expertise in specific industries, such as nanoscale vibration control for chip factories or explosion-proof and anti-corrosion technology for battery factories.
- Result: Only leading operators who understand these industry needs and can provide customized services can secure good assets. Core industrial parks like Huaxia Jinyu Smart Manufacturing Park are still in high demand, with occupancy rates of 92%.
- Conclusion: Industrial real estate is no longer just about renting out properties; it's about becoming industry partners. Without industry knowledge, you can't enter this market.
5. Future Outlook: No Universal Benefits, Only Precise Matching
The article concludes with a clear message: Don't expect AI to make all construction companies wealthy, as it's a high-tech field that requires specialized skills.
- Who Has the Opportunities?
- Power Companies: Integrating power supply with computing to reduce costs using green energy is a new opportunity.
- Traditional Construction/Building Materials Companies: Reusing idle facilities and quickly upgrading data centers can be profitable.
- Advice for Professionals:
- If you're in construction or related fields, consider learning about HVAC, electrical engineering, and automation; companies like DeepSeek are looking for such talents.
- If you're a small business owner, don't blindly cross-industry into computing infrastructure; it's a game for giants. Focus on your niche (e.g., special fire protection or clean engineering) or provide supporting services.
- Core Logic: This cycle is about skill differentiation. Those who can keep up with industry advancements will thrive; those who can't will be left behind.
💡 Insights for Everyone
1. Career Choices: If you or your family works in construction, civil engineering, electrical engineering, or HVAC, now is a good time for transformation. Focus on areas like data center operations, smart grids, and industrial IoT, which offer higher-paying jobs in the future.
2. Investment Perspective: Don't blindly trust all construction stocks based on the AI trend. Look at the company's business structure: Does it focus on construction or on mechanical and electrical systems? Its clients are internet giants or ordinary factories? The former carries higher risks but greater potential for growth, while the latter is more stable but has slower growth.
3. Cognitive Shift: The economy is shifting from scale-driven to technology-driven. Success now depends on technology and industry expertise. Whether you're running a company or looking for a job, ask yourself: Does the value you provide have an irreplaceable technical edge?