Lenovo CFO Cites Power Scarcity as AI Site-Selection Driver, Moves Into Middle East
Lenovo CFO Cites Power Scarcity as AI Site-Selection Driver, Moves Into Middle East
Grid power — not capital or chips — is becoming the decisive constraint in where AI data centres get built.
Power grid access has emerged as the binding constraint for AI infrastructure deployment, Lenovo's chief financial officer said on Bloomberg's Odd Lots podcast, with the bottleneck directly shaping where the Chinese technology company is placing its capacity bets — including in the Middle East.7
The CFO's framing marks a shift in the stated logic of AI infrastructure planning. For most of the past two years, the dominant constraint narrative centred on compute: GPU lead times, advanced chip export restrictions, semiconductor yields. The Lenovo executive was direct about what now comes first. "Seeing bottlenecks across the board," he said, citing land and power — "not just the transformers, but actual power from the grid" — as the variables now driving site decisions.7
The Middle East's attraction in that context rests on energy availability. The region holds substantial generation capacity backed by cheap domestic gas and has been expanding grid infrastructure without the decarbonisation constraints that complicate new power supply in Europe or parts of the United States. A Foreign Policy analysis published on June 24 (2026-06-24) described a broader economic realignment underway, with trade between China and the Arab world growing from roughly $36 billion in 2004 to nearly $400 billion by 2024.6 Technology infrastructure investment flows are starting to follow that trajectory, and power affordability is a material reason why.
The power constraint Lenovo's CFO flags is not unique to one company or one region. A joint report by Bain & Company and Standard Chartered found that data centres, electric vehicles and green industrial parks could generate more than 100 terawatt-hours of incremental power demand in Southeast Asia alone by 2030.4 The same analysis estimated an $18 billion annual shortfall in grid investment by 2035 — a gap that will directly determine where AI workloads can physically run, regardless of where the capital would prefer to locate them.2
The International Energy Agency made the same structural point in its World Energy Outlook 2025, released on Wednesday (2026-05-20). Despite measurable improvements in energy efficiency per individual AI task, aggregate power demand from data centres continues to rise as the volume of AI workloads scales faster than per-unit efficiency gains.5 The IEA described the resulting energy security outlook as more complex and fragile than ever before — language that extends well beyond the data centre sector but is directly relevant to it.5
For European markets, the implications carry a specific weight. Montel senior analyst Huangluolun Zhou has described AI-driven demand growth as rapid and localised, concentrated around specific data centre clusters rather than distributed across regional grids.1 That load profile creates pressure on transmission nodes and substation capacity rather than system-wide supply — a different problem from a diffuse demand increase, and harder to solve through conventional grid expansion alone.
A BloombergNEF report found that the data centre buildout required to support AI is likely to extend the operational life of fossil-fuel generation assets.3 The arithmetic is straightforward: if new electricity load arrives faster than new clean capacity can be commissioned, the marginal kilowatt-hour comes from existing thermal plant. The question for markets is where that thermal load lands and which fuel prices it supports.
Lenovo's CFO has provided one answer from the demand side: where the grid can actually deliver power at scale. ICE Endex TTF front-month stood at €41.08 as of Friday's close (2026-06-27), a level that keeps European data centre power costs structurally elevated relative to energy-surplus alternatives elsewhere.7 Whether Middle East grid capacity proves durable enough to support decade-long infrastructure commitments — against both technical and geopolitical uncertainty in the region — is the trade-off that site-selection planners cannot yet price with confidence.