AI's energy demands are driving fossil fuel investments, as energy needs for AI training and data centers are poorly matched with the patterns of solar and wind
The developers may talk green, about small nuclear reactors& renewables, but it's show — What's actually getting developed for AI is a huge surge of fossil fuel power stations — www.ft.com/content/0f61... Margot Finn / @eicathomefinn : Short answer: no. Renewables don't provide energy that allows data centres to run 100% of the time; at current rates, data centre greenhouse gas generation is expected to double in the next 5 years. Carbon credit systems mask these effects (if you close your eyes, or opt to live under a rock).
Context & Ripple Effects
AI’s power appetite had already put data-center sustainability targets under pressure, with earlier coverage documenting the sector’s difficulty meeting its own climate commitments as AI workloads grew. This report identifies the operational constraint behind that tension: intermittent renewable output does not by itself fit continuous data-center demand.
The result places AI buildouts in a broader infrastructure bottleneck: demand growth is colliding with the availability, timing, and reliability of electricity, rather than simply with annual clean-energy procurement.
First-order effects
- AI training and data-center operators gain access to dispatchable power through new fossil-fuel generation, but at the cost of higher associated emissions than their renewable and nuclear messaging suggests.
- Corporate carbon claims face greater scrutiny where credits are used to offset, rather than reduce, the emissions created by round-the-clock computing loads.
Second-order effects
- Power developers and fuel suppliers gain a nearer-term demand channel, while renewable procurement alone becomes less sufficient for customers that need firm capacity.
- The mismatch raises execution risk for AI infrastructure projects: later coverage of developers turning to turbines and diesel amid grid delays shows how constrained grid access can push operators toward on-site fossil backup.
Third-order effects
- If this persists, AI capacity deployment will increasingly be governed by firm-power and grid-connection availability, making electricity infrastructure a binding constraint on compute expansion.
- The sector’s climate accounting may shift from annual offset-backed claims toward closer examination of when and where power is generated; whether that produces stricter standards depends on corporate and regulatory responses.
The trend: AI infrastructure is becoming utility infrastructure, with reliable power supply—not just chips, capital, or renewable contracts—shaping where and how quickly compute can scale.