"Power Famine Looms in 2030"... The Company Smiling Amid AI's Relentless Appetite

"268GW Power Supply Gap in AIDC"

1–3 Years to Build AIDC, 5–15 Years to Expand Grids

Korean Power Firms See Order Backlogs... Entering Supercycle 2.0

As artificial intelligence (AI) servers, which consume massive amounts of electricity, proliferate worldwide, bottlenecks in the global energy infrastructure’s power supply are worsening. As a result, global big tech companies and power utilities are placing an increasing number of orders with Korean power equipment manufacturers, known for their advanced technology and rapid delivery capability. The industry anticipates that the looming global "power famine" will provide record-setting positive momentum for Korea’s power companies.

"268GW Power Supply-Demand Gap by 2030… U.S. Alone Facing a Shortfall of 170GW"

According to market research firm Trendforce, as of September 26, 2026, the global data center power demand is estimated to reach 161 gigawatts (GW) this year—up 31% from the previous year. The rapid growth is expected to continue next year with sustained capital expenditure (CapEx) by cloud service providers (CSPs) and rising overall server shipments, driving annual power demand to increase by over 30%.

The overwhelming driver of this power surge is, without question, AI servers. The share of general servers, which previously accounted for 40% of data center power usage, is expected to decrease to 25.6% next year. In contrast, the proportion of power usage by AI servers is estimated to rise from 25% last year to 33.4% this year, surpassing 40% in the following year.

The major issue is that grid (power network) supply expansion is failing to keep pace with the growing power demands of data centers. Trendforce predicts that this power supply-demand imbalance will begin this year and that the gap will widen dramatically after 2028.

Global Data Center Power Demand Scale Trends. TrendForce

By 2030, global data center power demand is projected to reach 490.7GW, but the grid’s maximum supply capacity is estimated at only 222.6GW. Consequently, a supply-demand gap of 268GW will emerge, equivalent to the total output of about 270 standard Korean nuclear power reactors.

In the United States, home to the world’s largest concentration of data centers, the power shortage is forecast to exceed 170GW by 2030. While data centers can be constructed in 1 to 3 years, building and permitting ultra-high-voltage grids takes 5 to 15 years. Moreover, over 70% of U.S. ultra-high-voltage transformers are outdated, having exceeded their lifespan of 30 to 40 years.

Thus, rather than passively waiting for grid access, big tech firms are actively adopting "on-site self-generation" by building their own power facilities or shifting to ultra-high-voltage direct current (HVDC) transmission and 800V direct current (DC) architectures to minimize power loss. This explains why global power companies such as Hitachi and Siemens Energy expect this power infrastructure boom to usher in "Supercycle 2.0," continuing through at least 2035—well beyond 2030.

3–4 Years’ Worth of Order Backlogs… Korea’s Power Equipment Makers Enter Ultra-High-Margin Boom

Paradoxically, this global "power famine" is providing Korean power equipment and cable manufacturers with unprecedented financial leverage. Owing to supply constraints in global power grids, Korean firms equipped with unrivaled technology, stringent quality, and quick delivery responsiveness are seeing a surge in orders.

As of the first half of this year, HD Hyundai Electric’s order backlog reached 12.3105 trillion won—almost three times its annual sales. Its first-half operating margin was 25.0%, demonstrating top-tier profitability in manufacturing. The company is also expanding its supply portfolio, operating the Cheongju "Smart Distribution Campus" alongside expansion of its Ulsan and Alabama factories, covering everything from ultra-high-voltage transformers to medium- and low-voltage distribution equipment.

LS ELECTRIC achieved a record high quarterly revenue of 1.577 trillion won in the second quarter this year, fueled by increasing demand for internal power distribution solutions for AI data centers. The company has raised its annual new order target to 6.5 trillion won and is enhancing local responsiveness by expanding its Utah factory in the U.S. It has also secured commercialization technologies for next-generation 800V DC grid semiconductor transformers (SST) and solid-state circuit breakers (SSCB) as indicated by Nvidia, building solid technical barriers as well.

Hyosung Heavy Industries surpassed an order backlog of 17.5 trillion won in its heavy industries division. In addition to its Memphis plant—a U.S. production base for ultra-high-voltage transformers—it has formed a joint venture for ultra-high-voltage gas circuit breakers (GCB) with Quanta Services, the No. 1 U.S. power EPC company, establishing a turnkey supply system for "transformers and circuit breakers" locally.

Status of Major Domestic Power Equipment Companies. Pinlit

Sanil Electric, which supplies fuel cells for self-generation in AI data centers, boasts an operating margin of 37.8% and is accelerating expansion of its Ansan Plant 2 and 3. Iljin Electric, too, is profiting from increased order backlogs for heavy electrical equipment overseas—totaling USD 1.04 billion—further leveraging the expansion of its Hongseong factory. Taihan Cable is transforming into a turnkey provider of ultra-high-voltage HVDC subsea cables, with order backlogs surpassing 4 trillion won, the construction of a second subsea plant, and the addition of cable-laying vessels.

Lee Sanghyun, analyst at BNK Securities, commented, "The dual effect of replacing aging U.S. power grids and the explosive growth of AI data centers is powering growth in the global power infrastructure sector. With global power giants now reassessing the industry’s peak from 2030 to at least 2035, a robust supplier-driven market is expected to persist post-2030, even under a conservative expansion stance."

This content was produced with the assistance of AI translation services.