Oxford Research Finds 35-50% of Planned AI Data Centres Behind Schedule, Driving Shift to On-Site Gas and Off-Grid Generation
A white paper from the University of Oxford's Smith School of Enterprise and the Environment concludes that between 35% and 50% of announced AI data centre capacity is running behind schedule, with power grid constraints pushing operators toward on-site gas turbines, fuel cells and battery systems as power grids struggle to keep pace.
The paper, titled The Great AI Infrastructure Buildout: Impact on Power and Commodity Markets, was authored by Dr Antif Ansar, programme director of the Oxford Programme on the Sustainable Future of Capital Intensive Industries; Dr Stephen Lezak; and Dr Guy Wolf, global head of market analytics at Marex, the diversified financial services firm that funded the research. It has undergone at least one internal referee review and may be submitted for academic journal publication.
Markets pricing announcements, not delivery
The central finding is that commodity and equipment markets — copper, transformers, electrical steel — are reacting to "lofty buildout claims that may materialise years late or not at all," creating a risk of short-term overpricing followed by oversupply. "Markets appear to be pricing in the announcement curve, not the delivery curve," said Ansar.
Lezak characterised the gap bluntly: "Media and markets are expecting a rapid, exponential ramp-up of Al infrastructure. The evidence tells a different story: a rollout that is delayed, lumpy, and constrained by physical realities." The paper uses the term "bragawatts" for announced capacity that is inflated or postponed.
Wolf added a financial warning: "If enterprise AI adoption does not scale as expected, the result could be stranded assets or sharply lower utilization."
Grid delays drive gas and off-grid generation
Because grid interconnection queues and transmission upgrades are not keeping pace, operators are increasingly turning to on-site generation. The paper details a shift toward gas turbines, fuel cells and battery systems. By 2028, off-grid generation for data centres could account for a "substantial uptick in natural gas usage not captured in official power sector statistics," the authors write.
That dynamic raises local power costs and increases natural gas consumption, complicating both energy market planning and climate accounting.
Economic case for scale remains unproven
The authors conclude that the economic justification for the current scale of buildout has not been demonstrated. "The winners will be those who balance ambition with realism — investing in the future of AI without losing sight of the practical constraints of the present."
The research forms part of an ongoing programme on the sustainable future of capital-intensive industries funded by Marex, which provides clearing, execution, market-making and hedging services across global commodity and financial markets. Marex's client base includes major commodity producers, consumers and traders, banks, hedge funds and asset managers — parties with direct exposure to the infrastructure and energy markets analysed in the paper.
Publication status
The paper has been reviewed by at least one internal referee. The authors note the work may be submitted for publication in academic journals. The views expressed represent those of the authors and do not necessarily represent those of the Oxford Smith School or any other institution or funder.
Sources
- 35-50% of planned AI data centres are behind schedule (smithschool.ox.ac.uk)
- Knowledge Base: The Studies and Books Behind the Articles (owners-council.com)