A new startup called Ctrl-S is racing to buy up intellectual property and experimental data from failing direct air capture companies before that knowledge vanishes for good. Founded by Jason Hochman, who spent four years running the Direct Air Capture Coalition, the company aims to build a licensable library of DAC innovations that surviving companies, energy majors, and even AI-driven materials science firms can tap into, essentially creating a knowledge rescue operation for a sector that’s hemorrhaging capital and talent.
Why it matters
DAC investment has cratered more than 60% from its 2022 peak. The Trump administration has axed tens of millions in federal funding. Customers are choosing cheaper climate interventions. The result: dozens of startups that made real technical progress are running out of cash. When a startup dies, its patents might linger in a database somewhere, but the experimental data, the engineering notebooks, the hard-won lessons about what didn’t work? Those walk out the door with the last employees. Ctrl-S is betting that preserving this institutional knowledge is both a public good and a viable business.
The details
Hochman officially stepped down from his advocacy role at the Direct Air Capture Coalition to launch Ctrl-S, named after the keyboard shortcut for saving a file. The company’s model has three revenue streams: annual subscriptions for library access, licensing fees for specific IP, and longer-term royalties tied to technologies developed from those licenses.
The founding team is small but credentialed. Nicole Williams, formerly the IP lead at Climeworks (the largest operating DAC company in the world), handles the intellectual property strategy. Silvan Aeschlimann, previously a DAC and carbon removal lead at the clean energy research nonprofit RMI, rounds out the core group. Phil De Luna, former chief scientist at DAC developer Deep Sky, sits on the expert review committee.
Ctrl-S is currently closing its first two acquisitions, with what Hochman describes as a “handful of others in the pipeline.” He’s fundraising a $2.5 million seed round. If the model works for DAC, he plans to eventually expand into other categories of climate tech facing similar consolidation pressures.
The diligence process is where things get interesting. The team evaluates acquired IP on technoeconomics, life-cycle carbon analysis, and commercial potential. Critically, they assess not just whole systems but individual components: sorbents, air contactors, regeneration processes. And they explicitly distinguish between startups that failed because their technology didn’t perform and those that were simply victims of a brutal funding environment. Both categories have value. The companies with broken tech provide what Hochman calls “negative learnings,” saving future developers from repeating expensive dead ends.
Who would actually buy this?
Hochman envisions several customer types. The most obvious: surviving DAC companies that might discover a defunct competitor had already solved a problem they’re stuck on. Why reinvent a sorbent formulation if someone already spent three years and $10 million optimizing one?
Then there are the deep-pocketed energy companies. Occidental Petroleum, which has bet heavily on DAC through its 1PointFive subsidiary, is the kind of player that could license pre-vetted technology rather than developing everything in-house.
The most creative angle is cross-industry application. De Luna pointed out that DAC components like air contactors could be repurposed for separating gases in other contexts, such as extracting hydrogen that’s trapped underground mixed with natural gas and other impurities. A technology designed to pull CO₂ from ambient air at 420 parts per million might find a second life in industrial gas separation where concentrations are far higher and economics are more forgiving.
And then there’s the AI play. Materials science companies training machine learning models are hungry for experimental data. Most available datasets skew toward successes. What they lack, Hochman argues, is the failure data: the sorbent compositions that degraded after 50 cycles, the contactor designs that couldn’t handle humidity, the regeneration temperatures that proved uneconomical. For training robust AI models, knowing what doesn’t work is arguably as valuable as knowing what does.
The patent troll question
There’s an obvious historical parallel, and Hochman is surely aware of it. In the early 2000s, after the dot-com bust, former Microsoft executives founded Intellectual Ventures to buy up IP from failing tech startups. The company became one of the most controversial entities in Silicon Valley, widely accused of patent trolling: acquiring patents primarily to sue other companies for infringement. Edward Jung, an Intellectual Ventures co-founder who now advises the firm, told Heatmap that actual uptake of the IP by larger companies was less than hoped. But he’s more optimistic about the Ctrl-S thesis, arguing that AI’s ability to rapidly evaluate complex technical options makes dormant IP more actionable than it was 20 years ago.
The Intellectual Ventures comparison cuts both ways. IV did fund a successful incubator called IV Labs that spun out Terrapower, the advanced nuclear company. But the patent trolling reputation was toxic. Whether Ctrl-S can thread that needle, generating revenue from IP without becoming predatory, will depend heavily on how the licensing terms are structured and whether the company prioritizes deployment over litigation.
What to watch
This is a pre-seed company with zero completed acquisitions and no paying customers yet. The $2.5 million target is modest, which is appropriate given the stage, but it also means the library will be small for a while. The real test is whether the surviving DAC companies and energy majors actually open their wallets for access. There’s a difference between everyone agreeing this knowledge should be preserved and anyone being willing to pay a subscription fee for it. The negative-learnings-for-AI-training angle is genuinely novel, but that market barely exists yet. And the specter of Intellectual Ventures will follow this company until it proves its intentions through action. Still, in a sector where billions of dollars of R&D risk evaporating into nothing, the basic impulse behind Ctrl-S is hard to argue with. The question is whether good intentions can become a sustainable business before the knowledge it’s trying to save is already gone.
Source: Heatmap News