Global operational carbon capture capacity reached roughly 50 million tonnes per annum by end-2024, according to the IEA's CCUS Project Database, while announced project pipelines multiply faster than new commercial ground-breakings. Reuters framed the resulting tension as "edges forward despite cost challenge" in its reporting on the sector.

That framing captures the industry's split identity in 2025: technology works, deployment economics mostly don't. Point-source capture on power, cement, steel, hydrogen and gas-processing flue streams is commercial, with more than two decades of operating history at facilities such as ExxonMobil's roughly 7 Mtpa LaBarge operation in Wyoming and the SaskPower Boundary Dam unit in Saskatchewan. Direct air capture (DAC) is leaving the demonstration phase, with the first commercial-scale plants beginning injection operations.

What does the cost gap look like?

Reported cost ranges for point-source capture cluster between $40 and $120 per tonne of CO2 separated, varying with flue-gas CO2 concentration, plant integration and proximity to storage geology. DAC sits materially higher: operators and the IEA have published figures between $250 and $600 per tonne at first-of-a-kind facilities, with stated learning-curve targets pushing toward $100 to $200 per tonne at second-generation plants.

Those costs intersect carbon prices that remain well below capture economics outside of regulated regimes. EU ETS allowances hovered in the €60 to €90 per tonne band through 2024-2025, supporting point-source capture on high-purity industrial streams but leaving DAC dependent on subsidy. California's LCFS credits and the federal Section 45Q tax credit fill the gap in the US.

How does 45Q shape the buildout?

The IRA-era 45Q schedule remains the single most material policy lever. Statutory values start at $60 per tonne for industrial point-source capture with permanent storage and rise to $180 per tonne for DAC paired with geological sequestration across vintages from 2026 through 2032. Transferability of the credit, codified in a 2018 amendment, is what made bankable project finance structures possible for facilities whose captured volumes exceed the developers' own tax appetites.

Which projects decide what happens next?

Two DAC facilities are the markers the industry will read most carefully over the next 18 months:

• 1PointFive (Occidental subsidiary) Stratos, a 500,000 tonne-per-year DAC plant in West Texas designed to inject CO2 into the Permian basin via Class VI wells; first CO2 injection was reported in 2024, with nameplate ramp expected through 2025-2026.
• Climeworks Mammoth, an Iceland-based DAC and mineralization facility with nameplate capacity of up to 36,000 tonnes per year; commissioning and first operating-year cost disclosures are the test.

What does the regulatory calendar change?

The binding near-term milestones are storage-permitting throughput and cross-border transport availability. In the US, the EPA's Class VI injection-well permit queue still runs multi-year reviews; agency throughput has expanded, but the queue continues to dictate which announced projects convert to final investment decisions. In Europe, the Northern Lights cross-border CO2 shipping and receipt terminal in Norway took its first commercial cargo in 2024, opening transport optionality for emitters without onshore storage access.

The deciding market marker is the EU Carbon Border Adjustment Mechanism's full-rate phase-in from 1 January 2026. As embedded emissions in imported cement, steel, fertilizer and aluminum face CBAM-related carbon costs at the border, the implicit carbon price for those same industrial streams inside the bloc — and for their US and Asian competitors selling into the EU — moves closer to capture-economics parity.

For CCUS to convert its announced pipeline into operational capacity at the rate implied by net-zero pathways, the sector needs both DAC cost disclosures from the first commercial plants and EPA Class VI permit throughput to clear the queue without further slippage. Either slipping by 12 months resets the deployment curve; both slipping compounds it.