Asia Energy Security and AI Power Demand: A US$5 Trillion Investment Supercycle Reshaping Return Maps

Core Conclusion

Asia is entering an investment supercycle driven by energy security and AI power demand as dual engines: by 2030, regional energy investment will roughly double from an average of about US$660 billion per year over the past decade to approximately US$1.1 trillion, totaling over US$5 trillion, and is expected to generate about US$9 trillion in enterprise value. Three core drivers: first, Asia's energy consumption will grow 50% over ten years, but investment has stagnated, with traditional supply chain investment falling to historical lows; second, AI data center demand will consume about one-sixth of the region's new electricity by 2030; third, geopolitical fragmentation forces countries to shift from "just-in-time" efficiency to "just-in-case" resilience, with policies prioritizing domestic reliable supply. This cycle is not linear expansion but a structural reset—energy security shifts from optional to strategic necessity, making capex more rigid, thus supporting valuation re-rating, especially in traditional energy, power equipment, fuel refining, etc.

What the Market May Be Underestimating

Investors generally view energy investment as cyclical or transitional spending, but the core characteristic of this supercycle is "risk-driven" rather than "demand-driven." AI power demand is only a catalyst; the deeper driver is geopolitical risk premiums and supply chain vulnerabilities forcing countries to prioritize domestic energy self-sufficiency as a national security core. This means traditional fossil fuel investment (coal power, gas power, refining) will not fade quickly due to energy transition, but instead gain long-term support from the need to ensure baseload and respond to supply disruptions. Coal will still maintain about 45% share of Asia's primary energy consumption through 2030. The predictability and sustainability of investment in coal power, gas power equipment, and grid infrastructure are underestimated. Additionally, Asian corporate balance sheets are healthy, able to support about 75% of investment (potential net debt increase up to US$1.8 trillion), with capital allocation shifting toward domestic growth intensifying.

Evidence Chain

Theme 1: Investment Scale and Pace Drastically Exceeding Expectations

• Annualized Asian energy investment will increase from an average of about US$668 billion in 2015-2025 to an average of about US$1.1 trillion in 2026-2030, growth rate jumping from 2% over the past decade to 11%.
• China's 15th Five-Year Plan unprecedentedly plans US$3-3.8 trillion investment; India, Indonesia, Malaysia, and others simultaneously increase investment in upstream, grid, refining, and strategic reserves.
• Investment focus shifts to power: about two-thirds of new investment flows to the power sector, with grid (average ~US$280 billion per year) and battery storage (average ~US$115 billion per year) seeing the largest increases.

Theme 2: Coal and Gas Baseload Status Solid

• Asia consumes nearly three-fifths of global coal reserves; China's coal power levelized cost is about US$80/MWh (30% load factor), far lower than gas power's US$110/MWh.
• By 2030, Asia needs to add about 600 million tonnes/year of coal consumption and 100 million tonnes/year of gas consumption, with new coal power investment of about US$318 billion.
• Economies such as Japan, South Korea, Taiwan, and Australia have coal power and nuclear units averaging over 20 years in age, urgently needing replacement or life extension; equipment OEMs and EPC contractors benefit significantly.

Theme 3: Structural Reshaping by AI Power Demand

• Global data center power demand will increase from about 1,200 TWh in 2025 to about 1,800 TWh in 2030, with Asia accounting for about 45% of the increment.
• Power costs have minimal impact on AI hyperscaler profitability: a price increase from US$120/MWh to US$140/MWh only affects ROE by about 50 bps; model efficiency, GPU asset life, and per-token revenue have 10-30x larger impacts.
• AI inference workloads make power fluctuations sharper, making energy storage shift from "optional" to "system-level necessity"; global storage deployment is expected to reach 3 TWh by 2030, 6 times current levels.

Theme 4: Potential Re-rating of Traditional Energy Sectors

• In 2024, the power sector saw valuation multiples rise 20-75% driven by AI demand; the energy sector (fuels, fertilizers, chemicals) may replicate a similar path. Currently, the one-year forward P/B for traditional energy sectors has recovered from about 1.0x to about 2.4x, but supply chain tightness (refining, fertilizers, chemicals) will drive 20-30% valuation re-rating.
• Asia's refining capacity growth lags demand growth by about 2x: by 2030, over 100 million tonnes/year of new capacity is needed (~US$90 billion investment), with only about 5 new builds currently. Fertilizer self-sufficiency investment is about US$37 billion.

Key Divergences and Risks

• Execution risk: India and Southeast Asian countries have a history of severe project delays; land permits, equipment supply (transformer lead times already extended to 2.5-4.5 years), and labor shortages may slow investment pace.
• Policy shift risk: Dependence on government guarantees for key components (strategic reserves, nuclear); if subsidies decline or energy transition policies reverse, investment pace may fluctuate significantly.
• Oversupply risk: In globally tradable commodity value chains (plastics, fuels), new capacity may create an oversupply cycle by 2029-2030; but power and domestic coal face lower oversupply risk due to AI demand and higher capital costs.
• Renewable curtailment risk: Asia's wind and solar curtailment rates are rising (China ~9% in 2024); if grid and storage supporting are insufficient, renewable investment returns will be compressed.
• Geopolitical shocks: Trade frictions, sanctions, or Middle East conflicts may disrupt energy logistics, but also act as catalysts for energy security investment.

Valuation and Trading Implications

The report screens 70 globally most preferred stocks, expected to provide over US$1.5 trillion in enterprise value growth, with 5-30% being upside to consensus for 2028. Among sub-sectors, grid operators have a market-cap-weighted average upside of about 29%, data center operators about 21.5%, power equipment manufacturers about 18%, while gas pipeline sub-sector shows negative returns (-19%) due to stagnant demand. Core beneficiaries include: CNOOC (0883.HK), PetroChina (0857.HK), Reliance Industries (RELIANCE.IN), GE Vernova (GEV), CATL (300750.SZ), etc. Least preferred areas: China solar supply chain (capacity nearly twice global demand), some battery components (except LiPF6), some power operators (facing per-kWh profit pressure and curtailment).

Appendix Data Summary

Table: Key Numbers for Asia Energy Investment Supercycle