Energy security, global refining and AI-powered electricity demand Report Interpretation

Overview

Morgan Stanley presents a thematic energy-and-power framework centered on energy security, stronger refining economics and AI-driven electricity demand. Its central conclusion is that these forces are converging into a long-duration investment cycle, particularly across Asia.

Core views

Morgan Stanley argues that more frequent energy shocks have elevated energy and economic security into strategic priorities, particularly in an AI-intensive economy. Its analysis covers 85 countries representing 93% of global primary-energy demand, 87% of oil demand and 95% of natural-gas demand. The institution expects a US$5tn-plus Asian energy-security investment supercycle: dependable-energy investment should nearly double relative to the last decade, while renewable investment plateaus after five years of significant growth. Power is estimated to account for about 65% of the region's energy-security investment needs. Morgan Stanley estimates that investment of this scale could secure technology, food and energy supply chains and unlock roughly US$9tn of value creation, implying nearly 1.6x value-creation torque. The energy-security theme changes the preferred power mix rather than simply increasing total spending. Morgan Stanley expects Asia's need for power-sector natural gas to encourage greater use of coal, energy storage and other dependable fuels. Its power-resiliency comparison highlights high import reliance in several Asian markets, including Japan, South Korea and Taiwan, supporting the case for diversified and more self-reliant supply. The report also expects the US shale revolution to be exported to Asia at scale for the first time in 2027, with implications for both energy security and AI power supply. The second theme is a global refining “golden age.” Morgan Stanley says global refining-capacity additions have undershot IEA estimates by about 0.36mbpd on average over the past five years, while demand has exceeded IEA estimates by about 0.8mbpd on average. It forecasts about 2.5mbpd of incremental demand over the next three years against only about 1.2mbpd of capacity growth, creating tighter supply-demand conditions. New refinery construction takes more than 10 years on average outside China, and brownfield expansion still takes roughly seven years, limiting the speed of supply response. As a result, the report expects high operating rates and Asian gross refining margins to remain above mid-cycle levels as gasoline and diesel cracks settle higher after conflict-related disruption, while hardware upgrades reduce fuel-oil production. It notes that refiners have already seen about 25% earnings upgrades and expects a further upgrade cycle as margins establish a higher base. The third theme is the changing face of power demand as AI scales. Morgan Stanley highlights rising global token demand and estimates that an average large-language-model query requires ten times the energy of a Google search; each new generation of AI model is described as requiring ten times more compute, and therefore energy, to train. It expects Asia's data-center power demand to grow at about 24% annually from 2023 to 2030, faster than in the United States and Europe, driven by AI inference demand and greater regional investment commitments. China is expected to account for a significant part of Asia's data-center growth as inference demand and semiconductor self-sufficiency support commoditization and faster commercialization. The report therefore connects expanding data-center capacity with accelerated deployment of grids, nuclear power, energy storage and dependable generation.

Analysis framework

The report links energy-security needs to required capital spending, then assesses how supply reliability and import dependence affect the generation mix. It separately compares refining demand growth with capacity additions, construction lead times, operating rates and gross refining margins. For AI power demand, it connects compute intensity and data-center expansion with future electricity demand and the infrastructure needed to supply it.

Methodology notes

• Industry AnalysisSupply-demand framework

  Refining supply-demand balance

  Morgan Stanley compares expected fuel-demand growth with refining-capacity additions and adds construction lead times to explain why utilization and margins may stay elevated.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  Energy security and AI power-infrastructure transmission

  The report traces energy-security and AI-compute demand through dependable generation, fuels, storage, grids, refining and related supply chains.

• Industry AnalysisCost curve analysis

  Round-the-clock power-generation economics

  The report compares the economics of reliable power-generation options to support its view on the changing Asian power mix.

Key data

• Asian energy-security investment supercycleUS$5tn+Morgan Stanley expects dependable-energy investment to nearly double versus the last decade.
• Power share of Asia energy-security investment needs~65%Morgan Stanley estimate.
• Potential value creationUS$9tnAssociated with US$5tn-plus of regional investment; described as nearly 1.6x value-creation torque.
• Refining demand versus capacity growth~2.5mbpd demand growth versus ~1.2mbpd capacity growthOver the next three years.
• Refining forecast misses versus IEA estimates~0.36mbpd lower capacity additions and ~0.8mbpd higher demandAverage annual differences over the past five years.
• New refinery completion time10+ years outside China; ~7 years for brownfield expansionIllustrates slow supply response.
• Asia data-center power-demand growth~24% CAGRMorgan Stanley estimate for 2023-2030, above the United States and Europe.

Impact & implications

The report sees the themes as mutually reinforcing: energy-security policy supports dependable-power, grid, storage and fuel investment; constrained refining capacity supports stronger utilization and margins; and AI data-center expansion increases the need for reliable electricity infrastructure. Morgan Stanley frames these effects as a broad thematic opportunity set rather than a single-company call.