Sometime in the early 2000s, two New Zealanders went looking through a
German microbe library for a bacterium that could eat smoke. They found it
filed under an unglamorous origin: Clostridium autoethanogenum, a strain
first enriched from rabbit droppings and coaxed to grow on nothing but carbon
monoxide, described in a 1994 paper in Archives of Microbiology that almost
nobody outside anaerobic microbiology had read. The bug did one useful trick.
Feed it the poisonous gas that pours out of a steel furnace, and it exhales
ethanol.

Twenty years later, that rabbit-gut bacterium is the engine of a public
company that has been granted 130 US patents, is running commercial plants on
three continents, sells the alcohol in Gucci perfume, and is also nearly out
of cash. The company is LanzaTech, and it is the clearest case on the site of
a pattern worth naming: the money can be draining out of a startup while its
patent office keeps issuing at record pace. The portfolio is the asset that
outlives the stock price.

The number

Across every spelling and subsidiary of the name, LanzaTech holds 130 US
utility grants, the first issued in 2011. The filings did not taper as the
company matured. They accelerated. 2023 was the record year with 18 grants.
2025, only half over in this data, already shows 15. Whatever is happening on
the balance sheet, the invention is not slowing down.

Zoom out to the field. In the patent neighborhood that covers fermenting
single-carbon gases into products, LanzaTech owns 38 grants since 2015. The
next name on the list is Coskata, with 7, and its sibling Synata Bio, with
another 7. Those two are the same lineage: Synata bought the assets when
Coskata, an earlier gas-fermentation hopeful, ran out of runway. The only
other serious presence is Calysta, and it is playing a different game
entirely, feeding methane rather than steel gas to a different microbe to make
fish-feed protein. Strip out the defunct and the adjacent, and LanzaTech has
roughly five times the US patent depth of anyone still standing in its exact
lane.

Read the claims, not the label

A patent count is easy to fake into a story. The test that matters: if you
delete the phrase “gas fermentation” from every one of these documents, do
they still describe the same invention? They do, because they all describe the
same physical thing. One anaerobic bacterium, one gaseous meal of carbon
monoxide and carbon dioxide and hydrogen, and a growing catalog of exits
engineered into its metabolism.

Read them in sequence and you watch a product ladder being built rung by rung.
Thirty-nine of the grants concern ethanol, the original trick. But a 2023
grant describes converting that ethanol into para-xylene with high selectivity
over the other aromatics you would normally get, and para-xylene is the direct
precursor to terephthalic acid, which is one half of PET, which is the plastic
in a water bottle and a polyester shirt. A 2024 grant makes liquefied
petroleum gas, propane and butane, from the same alcohol. Another engineers
the bug to secrete terpenes and farnesene, molecules that go into fuels and
fragrances. A 2025 grant has the organism producing 2-phenylethanol, the
compound that gives roses their smell. And a July 2025 grant describes
microbes engineered to continuously co-produce not just chemicals but
proteins and amino acids and microbial biomass from those same C1 gases. Food,
in other words, brewed from flue gas.

That is not a keyword cluster. It is one microbe being taught, pathway by
pathway, to turn pollution into an ever more valuable set of things. The
engineering DNA is shared all the way down: the same host, the same feedstock,
the same anaerobic reactor with its custom sparger designs (several of the
grants are just about how to get gas bubbles into liquid efficiently, the
unsexy problem that decides whether any of this works at scale).

Where the smoke actually becomes stuff

None of this would matter if it lived only on paper. It doesn’t. In Hebei
Province, China, a joint venture with the steelmaker Shougang has been running
a commercial plant at the Jingtang mill since 2018, pulling waste gas off the
furnaces and turning out roughly 16 million gallons of ethanol a year,
according to LanzaTech and its partners. In Ghent, Belgium, ArcelorMittal, the
largest steelmaker outside China, spent 200 million euros on a plant called
Steelanol that captures blast-furnace gas and, per the two companies, can make
80 million liters of ethanol annually. The first barge shipped in December
2024. They sell the product under the brand Carbalyst.

From there it climbs the ladder into things you can buy. The fashion retailer
Zara released party dresses made from captured-carbon polyester. The running
brand On has worked with LanzaTech on the EVA foam in shoe midsoles. Coty
replaced the alcohol in Gucci’s Where My Heart Beats fragrance with the
steel-gas version. And a spinoff, LanzaJet, opened a plant in Soperton,
Georgia in early 2024 that the Department of Energy calls the world’s first
commercial facility making jet fuel from ethanol, designed for nine million
gallons of sustainable aviation fuel a year. The molecule that started in a
rabbit’s gut now has a path into a jet engine.

The catch, and the reason this is a story

Here is the tension that makes LanzaTech worth writing about rather than
celebrating. The company went public through a SPAC in early 2023. By April
2025 its auditor, Deloitte, had attached a going-concern warning to the annual
report: substantial doubt about the ability to keep operating without raising
significant new money. Cash fell to $19.6 million by the end of September 2025,
down from $43.5 million nine months earlier, per its quarterly filing. Revenue
is shrinking, not growing. In August 2025 it ran a 1-for-100 reverse stock
split to hold onto its Nasdaq listing, and took a $60 million lifeline loan
from Brookfield. This is a company fighting to survive the year.

So who cares about the patents? Anyone doing corporate development at a steel
company, an oil major, a chemicals giant, or a consumer-goods conglomerate
with a decarbonization mandate and no biology bench. The uncomfortable arithmetic
of a going-concern startup is that its most durable asset is the pile of
grants, and that pile becomes available, through licensing or acquisition or
distress, precisely when the company is weakest. LanzaTech has spent fifteen
years and a lot of other people’s money answering a specific, hard,
capital-intensive question: how do you get a finicky anaerobe to reliably eat
industrial exhaust and secrete something you can sell? The answer is now
written down in 130 enforceable documents. Whoever ends up holding them will
not have to isolate their own bacterium from a rabbit.

Fermentation is one of the oldest technologies humans have. For nine thousand
years the input was sugar and the output was beer. The recombination LanzaTech
patented is almost embarrassingly simple to state and brutally hard to do:
keep the fermenter, swap the sugar for smoke. The adjacent possible was sitting
in a German microbe library the whole time. The open question in 2026 is
whether the company that found it will still be around to collect, or whether
the smoke-eating patents will finish the job in someone else’s hands.

Method note. Patent counts come from 9.3M US utility grants sourced from
USPTO bulk grant data, searched for gas-fermentation and single-carbon
microbial-conversion language and grouped by assignee; each company’s total
combines variant spellings and subsidiaries (LanzaTech files under several
corporate names, which are consolidated here). The 130-grant figure covers all
years through the data’s cutoff of mid-June 2026; the yearly counts and the
product-category tallies (39 ethanol grants, and the smaller sets on
para-xylene, liquefied petroleum gas, terpenes, protein, and 2-phenylethanol)
were each re-run against the same corpus. Field-share comparisons use grants
issued from 2015 onward. Company financials are drawn from LanzaTech’s SEC
filings and press releases; the origin of Clostridium autoethanogenum is from
its 1994 species description in Archives of Microbiology and contemporaneous
reporting by CNBC and the University of Queensland; plant and product details
are from LanzaTech, ArcelorMittal, LanzaJet, the US Department of Energy, and
trade coverage in WWD and Trellis. A patent portfolio is a record of what a
company has protected, not a guarantee that any product ships or that the
company survives to ship it.