Gigatonne-scale CO2 removal
Climate targets need billions of tonnes of CO2 removed from the air per year, but engineered removal is only about 2 million tonnes.
Open in the interactive tree →Options include forests and soils (cheap but reversible), bioenergy with capture and storage, direct air capture, enhanced rock weathering and ocean-based methods. Almost all pathways that limit warming to 1.5-2 °C assume several gigatonnes of removal per year by mid-century.
As of now
According to the State of CDR report (3rd edition, 3 June 2026), the world removes about 2.2 Gt CO2 per year, roughly 5% of annual emissions, and 99.9% of that is conventional land-based removal such as forestry; novel methods deliver only about 0.002 Gt per year but are growing about 40% annually. Direct air capture still falls far short of its claims: Climeworks' Orca averages roughly 600 t a year against 4,000 claimed. Stratos (500,000 t/yr design) is planned to start at the end of 2026.
What is missing
- Large amounts of cheap clean energy, since capture needs heat and electricity
- Cost: direct air capture is above US$1,000 per tonne and needs to fall to a few hundred
- Permanent geological storage capacity, pipelines and permits
- Trusted measurement and verification of how much CO2 stays removed
- Long-term policy and paying customers for permanent removal
Becomes possible once solved
- Offsetting residual emissions from aviation, cement and agriculture
- Net-negative emissions to bring temperatures down after overshoot
- Synthetic fuels made from CO2 in the air
- Less dependence on perfect emission cuts to hit climate targets
Open steps
- Cheaper, tougher CO2 sorbents High AI leverageFind materials that capture CO2 from humid air with low regeneration heat, high uptake and years of lifetime.
- Measuring rock-weathering removal Medium AI leverageMeasure how much CO2 enhanced rock weathering really removes from fields, given variable soil, weather and mineral content.
- Storage site screening and monitoring Medium AI leverageFind geological formations that can hold CO2 for centuries and detect leaks from seismic, well and pressure data.
- Capture plants on variable clean power Low AI leverageDesign capture plants and heat sources that run on variable wind and solar power with low energy use per tonne.
- Ocean CO2 removal verification Medium AI leverageProve how much CO2 ocean methods such as alkalinity addition take up, in a mixing ocean where the signal is tiny.
Where AI could help
Medium AI leverage. AI speeds the search for cheaper sorbents and storage monitoring, but gigatonne scale needs clean energy, pipelines, storage permits and paying buyers.
- Screen MOFs and other sorbents for CO2 uptake in humid air with learned atomic models instead of costly DFT
- Model weathering and soil chemistry to measure how much CO2 enhanced rock weathering really removes
- Tune capture-plant heat and power use against variable wind and solar supply
- Find and monitor geological storage sites from seismic and well data
Shown so far
- In April 2025, MOFGen, an agentic AI system of language, diffusion and quantum-chemistry models, generated hundreds of thousands of candidate MOFs, and five were synthesised in the lab. source
- In August 2025, Meta and Georgia Tech's ODAC25 dataset covered about 15,000 MOFs with CO2, H2O, N2 and O2 adsorption and trained machine-learned potentials for direct-air-capture sorbent screening. source