Today, there are several good reports.
This time, I would like to introduce a report that forecasts the outlook for construction and power sectors based on numerical data.
Although there has been a lot of noise about new AIDC construction for some time, positive signals are emerging for new DC construction in the US, as well as Europe, China, and other Asian countries including Korea. Therefore, the outlook for construction and power - particularly gas-fired power plants to quickly meet power demand - is becoming brighter.
For reference, a summary based on numerical data using AI will also be uploaded. Please refer to the attached report for details.
One-line Conclusion | AI data centers are expanding investment beyond construction contracts to power plants, power equipment, and renewable energy, with the report citing a potential increase of 13% in domestic power demand by 2030 based on global demand of 80.9GW.
1. Report Key Summary
「The ripple effect brought about by data centers」is a Kiwoom Securities industry report that connects the benefits of AI data center expansion to construction companies → gas/nuclear power plant equipment → renewable energy power generation projects.
Global AI data center demand is estimated at 80.9GW by 2030, but current announced global construction plans are only around 34.5~34.9GW. This discrepancy between planned capacity and demand estimates serves as the starting point for the report.
The domestic AI data center target is 8.4GW in 2029 and a cumulative 15GW in 2035. Plans announced by major domestic companies also reach approximately 10.9GW in 2030 and 14.3GW in 2035.
Reflecting new semiconductor fabs and data centers, maximum power demand could increase by approximately 13% to 118.0GW → 133.0GW in 2030 and 138.3GW → 156.7GW in 2036.
In the construction sector, DL E&C is the Top Pick, while GS Construction is a stock of interest. In utilities, Doosan Enerbility and VHI, and in renewable energy, SK Datonics are considered major beneficiaries.
2. Key Figure Summary
AI Data Center Demand and Supply Plans
Item | Report Figure | Meaning |
|---|
Training Power Increase Rate | Annual 2.17 times | This is a baseline scenario assuming a total computational volume increase of 3.5 times, a training period increase of 15%, and a 40% improvement in hardware efficiency. |
Global Demand in 2030 | 80.9GW | This is an estimated value calculated based on the demand for learning Frontier AI models. |
Global Construction Plan | 34.5~34.9GW | This logic suggests that there is room for additional construction, as the demand estimate is approximately 46GW short. |
Domestic Construction Goal | 8.4GW by 2029 | |
15GW by 2035 | This is the Korean government's mid-to-long-term plan for AI data centers. | |
Domestic Company Announced Plan | 10.9GW by 2030 | |
14.3GW by 2035 | This is the sum of the plans announced by SKT, GS, Naver, KT, and Samsung SDS. | |
Changes in Domestic Power Demand
Category | Existing Outlook | New Outlook | Increase |
|---|
2030 Maximum Power Demand | 118.0GW | 133.0GW | +15.0GW · Approximately 13% |
2036 Maximum Power Demand | 138.3GW | 156.7GW | +18.4GW · Approximately 13% |
2030 Power Consumption | 625.3TWh | 708.8TWh | +83.5TWh |
2036 Power Consumption | 707.8TWh | 806.1TWh | +98.3TWh |
The total power demand of 14 semiconductor fabs is estimated to be approximately 22GW.
The new generation capacity needed by 2035 is 40~50GW, and the existing gas power plant expansion plan of 19.5GW needs to be increased by at least 50~100%.
The domestic gas turbine and steam turbine market accessible to Doosan Enerbility is estimated to expand from 9~11 trillion won to 15~20 trillion won, and the HRSG market of BHI is estimated to expand from approximately 2 trillion won to 3~4 trillion won.
3. Beneficiary Companies and Target Stock Prices
Company | Opinion | August 25 Stock Price | Target Stock Price | Key Rationale |
|---|
DL&C | Buy | 70,300 won | 120,000 won | Top Pick. Owns both US power plant and SMR, domestic data center references. |
GS Construction | Buy | 35,000 won | 43,000 won | Watchlist. Expected to directly benefit from the 1.2GW AI data center in Donghae and the 130MW pipeline. |
Hyundai Construction | Buy | 121,300 won | 178,000 won | Has references for 18 data centers and EPC capabilities for nuclear power plants and SMRs. |
DaeWoo Construction | Buy | 17,620 won | 23,000 won | Promoting the 200MW Jangseong and 300MW Gangjin data center projects in Jeollanam-do. |
Doosan Enerbility | Buy | 80,700 won | 130,000 won | Key beneficiary of the expansion of the gas turbine, steam turbine and nuclear power plant equipment market. |
Korea Electric Power | Buy | 32,700 won | 41,000 won | Upgraded opinion based on the possibility of alleviating transmission network burden and lowered valuation. |
SK Data & Networks | No Opinion | 55,200 won | Not Presented | Paying attention to the increasing demand for renewable energy PPA due to RE100 implementation by large corporations. |
BHI | No Opinion | 63,500 won | Not Presented | Expected to benefit from the expansion of the HRSG market for gas combined cycle power plants. |
HD Hyundai Energy Solutions | No Opinion | 135,100 won | Not Presented | Based on a domestic solar module market share of about 30% and a carbon certification market share of over 50%. |
4. Things to Consider When Interpreting Numbers80.9GW is not a confirmed order volume. It is a baseline scenario calculated assuming computational workload, training period, and hardware efficiency. Therefore, if technological efficiency improves faster than expected, demand may decrease.
Domestic plans require power system impact assessments, securing power plants and transmission networks, lease agreements, and financial procurement to generate revenue.
Revenue for GW-scale data centers is recognized in stages. The announced capacity and a construction company's sales for a specific year are not the same figure.
Increasing gas power generation is a realistic alternative to address short-term power shortages, but the scale of the plan may change depending on carbon reduction policies and fuel prices.
The target stock price and current price are both based on the report as of August 25, 2026 and are not real-time prices.
5. Simple Explanation of Terms
Term | Explanation |
|---|
GW·TWh | GW represents the size of power facilities needed at a given moment, while TWh represents the amount of electricity actually used over a certain period. |
Hyperscaler | Companies like Amazon, Microsoft, and Google that operate massive cloud and data centers. |
Neocloud | AI infrastructure companies that specialize in providing GPU computing resources. |
HRSG | A heat recovery steam generator (HRSG) that recovers waste heat from a gas turbine to produce steam for additional power generation. |
PPA | A contract between a company and a power generation company where the price and quantity of electricity are fixed for a long period. |
6. Additional Opinions from AIThe most important number in this report is not the data center construction plan itself, but the upward revision of power demand forecasts. This is because contracts for data center units may be shorter than those for power plants, transmission networks, cooling facilities, and renewable energy agreements.
However, when selecting beneficiary companies, it is better to first check the actual order backlog and construction status rather than the announced GW. GS Construction, which directly undertakes construction, may see faster performance reflection. DL E&C, Doosan Enerbility, and VHI are structured to benefit more when investment in power generation facilities expands. Therefore, it is crucial to distinguish between anticipation before construction confirmation and performance recognition after order placement.
7. Sources and Limitations
“Ripple Effects Brought by Data Centers,” Kiwoom Securities, Shin Dae-hyun·Jo Jae-won, August 26, 2026.
This post focuses on the core outlook of the report and company-specific figures at the user's request.
Global and domestic construction plans include media reports, company announcements, and policy goals, which may differ from actual construction and completion scales.
Some government plans and tentative power supply and demand forecasts were not independently verified against original sources at the time of posting. Therefore, they are distinguished as citation figures in the report.
로그인 후 더 많은 기능을 이용하세요!
로그인
7.83
▼ -0.01
2026.08.26 KEB 하나은행 고시회차 470회
