AI Spending Plateau in 2026: Carbon Removal Shifts to 'Prove and Learn'

The carbon removal sector is confronting a pivotal shift as signals of a potential AI spending plateau emerge, forcing a reassessment of the industry's growth narrative. According to a recent analysis on Marginal Carbon, the era of explosive voluntary demand for durable carbon dioxide removal (CDR) credits may be giving way to a more measured, compliance-driven market. The publication notes that total contracted tonnes may have peaked in 2025 or 2026 due to Microsoft's forward buying spree, and it could take years to reach 30 million tonnes contracted annually again.

This recalibration arrives as the broader climate technology ecosystem grapples with a 'rupture' in the liberal international order, as described by Quico Toro on One Percent Brighter. Toro argues that the post-1945 rules-based order is effectively dead, and with nations like Germany forced to spend 4% of GDP on defense, energy security will consistently trump emissions reductions. In this environment, the old 'speed and scale' mantra for carbon removal appears increasingly untenable.

The AI Spending Plateau Impact

The AI spending plateau is not just a funding issue but a signal that the entire ecosystem must mature. As AI demand drives unprecedented energy consumption, data centre operators are exploring direct air capture (DAC) and point-source carbon capture to offset their growing carbon footprints. A report from Noah Deich on Everything and the Carbon Sink highlights that the Department of Energy's DAC Hubs funding and CDR Purchase Pilot Prize could have been designed more effectively, but the concept itself is gaining traction.

Carbon Removal Market Shifts

Industry insiders are now advocating for a new strategic framework. Marginal Carbon's analysis suggests that the sector's new story should be 'prove and learn' rather than 'speed and scale.' Voluntary demand for durable CDR is expected to result in low tens of millions of tonnes delivered per year by the mid-2030s, but a near-term explosion is unlikely. The only real driver for compliance demand remains slow, and may be much smaller than many expect even over the long term.

Prove and Learn Strategy

Noah McQueen, in a detailed vision piece, emphasizes that carbon removal will not scale because any single system works perfectly. 'It will scale when the interaction between systems becomes coherent: when interventions, policy, funding, social legitimacy, markets, and natural constraints move in step,' he writes. This systemic view underscores that the AI spending plateau is a necessary correction, forcing the industry to prove its viability before scaling.

Compliance-Driven Market

The shift from voluntary to compliance-driven CDR is critical. As nations prioritize energy security, compliance demand may be smaller than expected. However, the need for gigatonne-scale removal remains urgent, as Quico Toro bluntly states that the 1.5 degree target is effectively dead.

Carbon Capture in Data Centres

Chris Goodall's Carbon Commentary newsletter further notes that integrating carbon capture into data centre operations could provide a dual benefit: reducing operational emissions while generating durable carbon credits for voluntary buyers. However, the technology remains expensive, and the economic viability hinges on policy support and falling costs.

Technology Integration

Data centre operators are exploring carbon capture to offset their carbon footprints. The Department of Energy's DAC Hubs funding is a step forward, but more effective design could accelerate adoption. The AI spending plateau may be a catalyst for this integration, as operators seek cost-effective solutions.

Economic Viability

The economic viability of carbon capture in data centres depends on policy support and falling costs. As the AI industry matures, the need for carbon removal will grow, but only if the technology proves its worth.

For the carbon removal sector, the path forward requires aligning these pieces. The AI spending plateau may be a necessary correction, forcing the industry to prove its viability before scaling. As Noah McQueen concludes, 'We need a shared vision for how these pieces align into something capable of reaching gigatonnes.' The next decade will test whether the sector can deliver on that promise.