Strangest Patents: Climeworks Built a Cathedral. The Patent Office Just Approved a Paint.

The first thing the US Patent and Trademark Office issued on April 28, 2026, in the world of carbon dioxide removal was not a billion-dollar plant. It was a paint.

Patent number 12,612,521 describes a coating made of metal silicate particles, a binder, pigment, and solvent, dried onto walls to capture carbon dioxide from ambient air. The most plausible candidate for those metal silicate particles is olivine, a green magnesium-iron silicate spat out by quarries as a byproduct of nickel and chromite mining, which slowly mineralizes CO₂ into carbonate when exposed to humid air. A pound of crushed olivine can bind a pound of carbon. The same chemistry is at work in The People’s CO₂, a Rice University spinout led by chemical engineer Tanya Rogers; per Rice Engineering magazine, that company’s olivine wall paint sequesters roughly 40 to 50 percent of its mass in carbon over one to two years and was applied last year at Space Center Houston and around Kemah, Texas. No fans, no sorbent regeneration loop, no power bill.

On the same morning the paint was issued, the patent office also approved a “carbon capture adapter” — patent number 12,611,630 — designed to clip onto “high air flow locations,” explicitly including parts of a vehicle. And on April 21 it approved an ocean direct air capture system (12,607,400) that hangs two concentric tapered tubes vertically in the sea, using nothing but the temperature gradient between surface water and the deep to drive air through the filter and generate the electricity that runs it.

Read in isolation, each of these is mildly weird. Read together with everything the office has issued since September, they describe a quiet rout.

The megaplant has had a rough year

Climeworks’s Mammoth facility in Iceland was supposed to be the proof point for the dedicated direct air capture plant: 72 modular collector containers, geothermal power, 36,000 tonnes of CO₂ removal a year. In its first full year of operations, 2024, it captured 105 tonnes total, according to reporting by CleanTechnica and CNN. That is less than three one-thousandths of nameplate capacity. CNN later confirmed Mammoth had only twelve of those seventy-two containers actually working, with filter problems inherited from 2023 prototypes. In May 2025 the company laid off 106 people, around 22 percent of staff, blaming policy uncertainty and slower-than-expected project ramps.

Occidental’s Stratos facility in the Permian Basin, slated for one million tonnes per year using Carbon Engineering’s hydroxide-calcination loop, is still the standard-bearer for the megaplant theory of DAC. But Stratos is the exception that confirms the rule. According to a 2024 review in MRS Energy & Sustainability and reporting in CleanTechnica, the most mature solid-sorbent DAC modules still need 2,000 to 3,000 kilowatt-hours of energy per ton of CO₂, and credits sell for between $600 and $1,000 per ton. That is not a cost curve. That is a wall.

If you are an inventor sitting in front of that wall, the obvious move is to stop competing with it. And that is exactly what the patent office has been approving.

Eleven patents, one design pattern

Between September 2025 and the end of April 2026, USPTO issued 52 patents whose titles describe carbon capture or CO₂ capture systems. At least eleven of them — more than a fifth — share a specific engineering idea that does not appear in the Climeworks playbook: do not build a dedicated DAC plant. Find something that is already moving air, water, or heat for some other reason, and bolt the capture step onto it.

Some of the eleven are obvious once you read them.

Patent 12,515,163, granted January 6, describes a direct air capture system in which the blades of a working wind turbine are perforated. As the turbine spins to make electricity, air gets sucked through the holes into ducts inside the nacelle, across amine-based adsorbers, and back out. The same blades that turn wind into watts also become an enormous CO₂ scrubber. No new fans. No new shaft. The energy penalty is hidden inside an asset that already exists.

Patent 12,601,327, granted April 14, traces back through a 2021 provisional application and Lawrence Berkeley National Laboratory work to a 2023 ChemRxiv paper modeling a “hybrid geothermal cooling tower direct air carbon capture process.” The patent inserts a DAC filtration stage into the condenser fans of an existing geothermal power plant. The fans are already there; the waste heat is already there; the electricity is already there. The capture is a passenger.

Patent 12,607,400 takes the same parasite logic to the open ocean. Two concentric tapered tubes in the water column. Air gets pulled down through one tube, expanded and warmed back up the other, and the temperature differential between surface and depth is enough to drive the fans, generate electricity, and pull CO₂ across the sorbent — borrowing energy from a thermal gradient that has been running for free since the Eocene.

Patent 12,551,845, granted in February, is a residential olivine box: an emission source flows through channels of olivine, the silicate mineralizes the CO₂, no separate amine regeneration loop required. The carbon-capture paint is the same chemistry, smeared onto a wall.

Then it gets stranger. Patent 12,569,808, granted March 10, proposes inoculating polar sea ice with selectively bred extremophile algae as a “carbon-sequestering vaccine,” claiming the ability to restore atmospheric CO₂ to pre-industrial levels within a decade. Patent 12,496,551, granted in December, makes a near-identical claim using algae sprayed into tropospheric clouds. Patent 12,435,601, from October, describes a mobile offshore DAC plant mounted on a jack-up rig — the same hardware the oil industry uses to drill in shallow water, repurposed to extract carbon instead.

Every one of these patents is doing the same thing under the hood. They are getting out of the business of building a dedicated machine to move air, regenerate sorbent, and produce concentrated CO₂. They are renting that machine from something else: an HVAC duct, a turbine blade, a cooling tower, an ocean column, a sheet of sea ice, a wall.

So what

The “so what” matters most for two readers.

For the climate-investment side: the megaplant thesis assumes one large CapEx asset can amortize the energy penalty across millions of tonnes per year. The parasitic thesis assumes the energy penalty becomes invisible because the host machine is already paid for by some other revenue stream — a wind farm’s PPA, a geothermal plant’s electricity sales, a city’s coat of exterior paint. If the parasitic thesis is right, the carbon-removal market does not consolidate into half a dozen Climeworks-sized companies. It fragments into thousands of tiny add-on plays attached to existing infrastructure owners, who collect the credit revenue as a side hustle.

For the corporate R&D side: the action has visibly moved out of pure-play DAC startups and into adjacent industries that already own air-moving or thermal infrastructure. A wind turbine OEM that ships blades with engineered porosity is in the carbon-removal business, as is a paint company that ships olivine-loaded coatings. Neither has to build a Mammoth.

There is a real skeptical edge here, and it is worth keeping in mind. Olivine paint takes one to two years to fully saturate. Algal-vaccine claims of restoring pre-industrial CO₂ within a decade are physically implausible. Many of these patents will turn out to be cranks, vaporware, or rounding errors. But the pattern in the issued claims is not a vocabulary trend. It is a coherent technical move, repeated in eleven distinct chemistries and at radically different scales. Climeworks built a cathedral and the cathedral is mostly empty. The patent office is filling its filing cabinet with paint.

Method. Counts come from US utility-grant patents indexed by USPTO bulk grant XML, filtered by titles containing “direct air capture,” “carbon capture,” or “CO2”+”capture” between Sept 1, 2025 and Apr 28, 2026. The “parasitic” count of 11 is a manual classification by the author against full abstracts; reasonable readers could include or exclude one or two patents at the margin. Climeworks Mammoth performance and layoff figures are from CNN (May 2025) and CleanTechnica (May 2025). DAC energy and cost numbers are from peer-reviewed surveys, including MRS Energy & Sustainability (2024). The People’s CO₂ paint figures are from Rice Engineering magazine. The geothermal cooling-tower patent traces back to a 2023 ChemRxiv preprint by Lawrence Berkeley National Laboratory. Patent assignee data was not available in our local index for grants issued after the early-2025 cutoff in the standard PatentsView dump; named inventor and company attributions in the post are sourced from the news and academic record, not the patent face sheet.