The Yeast Bottleneck Blocking Cheap Biosynthetic CBG

September 6, 2026

Photo by cottonbro studio via Pexels.

There are two separate stories floating around about biosynthetic CBG, and they keep getting mashed into one. The first story is a real, documented piece of metabolic engineering: researchers have coaxed yeast into producing cannabigerolic acid (CBGA) by inserting plant genes into its genome, and they've published the titers, the bottlenecks, and the enzyme names in peer-reviewed journals. The second story is a commercial narrative -- the one that says fermentation is about to make CBG as cheap as citric acid, that Big Biotech and Big Cannabis are on the verge of a manufacturing breakthrough, that plant cultivation for minor cannabinoids is basically obsolete. That second story gets repeated in pitch decks and trade press far more confidently than the underlying science actually supports.

This piece is an attempt to pull those two threads apart. We'll walk through exactly what's been published about the enzyme that's currently choking the pathway, what a real commercial partnership between Cronos Group and Ginkgo Bioworks actually delivered on store shelves in Canada, and what it would take -- based on how comparable biosynthesis projects have actually played out historically -- to get from where the science sits today to something resembling commodity-priced CBG. None of this is a prediction that biosynthesis will fail. It's a case for reading the specific numbers before you believe the timeline.

Why Yeast Doesn't Want to Make Cannabinoids

Photo by Jonathan Borba via Pexels.

Brewer's yeast, Saccharomyces cerevisiae, is a remarkably obliging organism -- it's been fermenting sugar into alcohol for at least 7,000 years, and over the last four decades it's been re-engineered to churn out human insulin, hepatitis B vaccine antigens, and artemisinin precursors for malaria drugs. But none of that history gives it any head start on cannabinoids, because yeast has zero native machinery for the specific chemistry cannabis plants use. It doesn't naturally accumulate meaningful pools of geranyl diphosphate (GPP), the ten-carbon building block that cannabis plants use as one half of the CBGA molecule. It doesn't produce the fatty-acid-derived intermediate, olivetolic acid, that forms the other half. And it lacks several of the enzymes needed to string those pieces together and stabilize the result.

So engineering yeast for CBGA isn't a matter of tweaking an existing pathway -- it means importing an entire multi-step branch of plant biochemistry, gene by gene, into a cell that never evolved any of it. That's a fundamentally different order of difficulty than the insulin case, where scientists inserted a gene for a single, relatively small protein and let E. coli's existing translation machinery do the rest. Cannabinoid biosynthesis requires coordinating a whole assembly line: precursor supply, intermediate stability, and a final coupling step, all inside a cell whose native metabolism is actively competing for the same carbon and energy.

That final coupling step is where a prenyltransferase enzyme comes in, and it's the linchpin of the whole operation. In plain terms, a prenyltransferase is the molecular welder that attaches the GPP unit onto olivetolic acid to form CBGA -- the parent molecule from which THC, CBD, and other cannabinoids are derived. Without a prenyltransferase doing its job efficiently, you can have all the right raw ingredients sitting in the cell and still get no CBGA out the other end, because nothing is fusing them together. Get this enzyme working well, at scale, and you've effectively unlocked the whole cannabinoid product family, since CBGA is the shared precursor upstream of nearly everything else. That's exactly why so much of the published research over the past several years zeroes in on this one enzyme rather than the pathway as a whole.

The Specific Bottleneck: An Unstable Enzyme Named CsPT4

Photo by Ivan Aviles via Unsplash.

The specific prenyltransferase at issue in cannabis biosynthesis is called CsPT4, and a 2023 study published via ScienceDirect on engineered CBGA-producing yeast pinned down two linked problems holding back the whole system. First, the yeast's conversion of hexanoate (a cheap fatty-acid feedstock) into olivetolic acid was inefficient -- a lot of the starting material simply wasn't making it through that early step. Second, and more consequentially, CsPT4 itself showed limited activity and poor stability inside the yeast cell. When an enzyme like this degrades faster than the cell can replace it, or simply doesn't fold and function correctly in a microbial host it was never evolved for, the whole pathway backs up like a clogged pipe -- precursors accumulate upstream while the final product trickles out at a fraction of what the theoretical chemistry would allow.

That same 2023 study reported a benchmark CBGA titer of 510 milligrams per liter of fermentation broth, achieved using relatively cost-effective feedstocks -- glucose and hexanoate, rather than exotic or expensive precursors. That's a genuine improvement over earlier engineered strains, and it matters because titer is the single number that determines downstream economics. The more product you get per liter of broth, the less broth you need to run through extraction and purification to get a kilogram of finished isolate, and purification is typically where most of the cost in fermentation-derived compounds actually lives. A titer that's 10 times higher doesn't just mean 10 times more product from the same tank -- it can mean dramatically lower cost per kilogram, because you're not paying to filter, concentrate, and purify a mostly-water solution to extract a trace amount of target compound.

Still, 510 mg/L, while a real step forward, remains far below the titers achieved in mature industrial fermentation products -- citric acid and lysine, for comparison, are produced at titers measured in tens to over a hundred grams per liter, not milligrams. That gap is worth sitting with rather than glossing over. And critically, this isn't a solved problem sitting in an old paper waiting for someone to scale it up: a 2024 paper in Biotechnology Journal and a 2025 paper in the Journal of Agricultural and Food Chemistry both revisit CsPT4 stability as an open, unresolved research question, not a bottleneck that's since been cleared. If the enzyme-stability problem were fixed, you'd expect to see it show up as a much larger titer number in a more recent paper. As of the most recent published research, that number hasn't appeared.

Cronos and Ginkgo: What Actually Got Built, and Where It Stopped

Photo by Elsa Olofsson via Unsplash.

The commercial side of this story has actual documented milestones, and they're worth laying out plainly because they show both how far a real partnership got and where the public trail goes cold. Cronos Group and Ginkgo Bioworks signed a partnership in 2018 reportedly worth up to $100 million, with a stated goal of producing pure, single-molecule cannabinoids at a cost under $1,000 per kilogram -- a price point that would put fermentation-derived cannabinoids in direct competition with plant extraction economics. This wasn't a vague research collaboration; it had a specific cost target attached to it from the outset.

The buildout produced real regulatory and commercial results. Cronos Fermentation received both a Health Canada processing license and a CRA license in 2021, and began commercial CBG production that June. From there, the timeline shows a company actually shipping product: a CBGA equity milestone was announced in August 2021, followed by the SPINACH FEELZ Chill Bliss 2:1 THC to CBG gummy launching that October, and a 7:1 THC to CBG vape launching in January 2022. By the week ending June 11, 2022, that gummy had captured 2.4% market share within Canada's gummies category -- a modest number in absolute terms, but a genuine, verifiable commercial foothold for a biosynthetically-derived minor cannabinoid product, not a lab curiosity.

A THCV productivity milestone was announced in June 2022, extending the platform to another minor cannabinoid. And then the public record largely stops. There have been no scaled-CBG commercialization announcements from Cronos or Ginkgo since that point covering this specific partnership's progress. It would be tidy to call this a failure, but the honest answer is that it's an open question rather than a resolved narrative. Ginkgo Bioworks has separately faced widely reported cost overruns and financial strain across its broader foundry business in the years since, which could plausibly explain a quieter cadence on any single partner program without implying the CBG work itself failed technically. It could also reflect a strategic pivot toward other cannabinoids or priorities, or simply a change in what gets publicly reported once a product has launched and market share has been established. Neither Cronos investor relations nor Ginkgo have offered a public accounting of the current status of scaled CBG production, and that silence itself is the data point worth flagging -- not proof of failure, but not evidence of success either.

What It Would Actually Take to Get to Cheap CBG

Photo by Yuri Shkoda via Pexels.

It's tempting to assume that once a lab demonstrates a working pathway, getting to commodity pricing is just a matter of building bigger fermentation tanks. That's not how this works. Strain engineering, not reactor capacity, is the gating factor here -- a 100,000-liter bioreactor running an unstable, low-activity CsPT4 enzyme just gives you a much larger volume of the same disappointing titer. Scale-up amplifies whatever the strain can already do; it doesn't fix a biochemical bottleneck sitting inside the cell.

The most plausible path forward, based on how comparable industrial enzyme problems have actually been solved, is directed evolution and targeted enzyme engineering -- iteratively mutating CsPT4 to improve its thermal and oxidative stability inside the yeast host, the same general approach that took Taq polymerase and industrial amylases from fragile lab curiosities to robust workhorses over years of incremental refinement. That's a real, well-documented engineering pattern, but it's also a slow one; it typically involves many rounds of mutation, screening, and re-testing rather than a single fix.

The closest large-scale historical analog for the overall timeline is recombinant insulin. Genentech achieved lab-scale proof-of-concept for engineered E. coli insulin production in 1978, and Humulin didn't reach FDA approval and market until 1982 -- roughly a decade all told from early feasibility work to a commercial product, and that was with strong financial backing and a single, comparatively simple target protein. That's a useful rough yardstick for how long it can take to go from a working lab demonstration to a market-ready biosynthetic product, even under favorable conditions.

The conservative counter-case matters here too, though: cannabinoid biosynthesis is meaningfully more complex than insulin production. Insulin required inserting and expressing one protein. CBGA production requires multiple enzymes working in sequence, at least one of which is membrane-associated and prone to the exact instability problem the 2023 study identified, plus adequate precursor supply from two separate metabolic branches. That complexity is a legitimate reason the insulin timeline could turn out to be optimistic rather than representative. Layered on top of that is the broader pattern seen across synthetic biology projects generally: titer improvements tend to follow a slow-then-fast curve, with years of incremental single-digit gains followed by a breakthrough that unlocks a step-change. A jump from 510 mg/L to something an order of magnitude higher is plausible under that pattern -- but there's no publicly established, universally agreed titer threshold for cost parity with plant extraction, so any specific number attached to when that happens should be read as an estimate, not a fact.

Who Wins If Biosynthetic CBG Gets Cheap -- and Who Doesn't

Photo by Weedzard Cannabis Light via Unsplash.

If CsPT4 stability and the hexanoate-to-olivetolic-acid conversion problem both get solved to the point of hitting sub-$1,000/kg, the practical effect is that CBG isolate stops being a byproduct-scarcity product and starts behaving like a commodity formulation input -- priced and sourced more like a flavoring agent than a specialty extract. Right now, high-CBG cultivars exist largely because growers deliberately breed and harvest plants at a specific stage to maximize CBG before it converts into THC or CBD; it's a niche, labor-intensive cultivation strategy built around chemical scarcity. Cheap fermented CBG would undercut the economic rationale for that entirely, in much the same way synthetic vanillin -- now the overwhelming majority of the global vanilla flavoring supply -- reshaped vanilla bean economics and left natural vanilla as a smaller, premium-positioned niche rather than the default source.

The beneficiaries of that shift would most likely be beverage and topical brands that need consistent, high-purity minor cannabinoid inputs at predictable cost -- categories where batch-to-batch consistency matters more than any provenance story, since fermentation can in principle deliver a more uniform, standardized product than extraction from a variable plant crop ever will. That's the same logic that's made fermentation-derived vitamins and amino acids the default over plant- or animal-extracted versions in food manufacturing.

But it's worth being clear-eyed about the counter-case: biosynthesis isn't racing against a fixed target, it's racing against a moving one. Conventional plant breeders have continued improving CBG-dominant cultivar yields through ordinary selective breeding, the same slow, cumulative process that's driven yield gains in commodity crops for a century. If cultivation-side CBG yields keep climbing even modestly while biosynthesis works through its enzyme-stability problem, the cost gap fermentation needs to close gets reset upward before it's ever closed.

And there's a variable that has nothing to do with chemistry at all: regulatory classification. Fermented cannabinoids may end up treated differently than plant-derived ones under US and Canadian rules -- the regulatory pathways for a genetically engineered microbial fermentation product don't automatically map onto the ones built for plant-derived extracts, and that distinction has slowed other fermentation-derived food and supplement ingredients before. Even a fully solved enzyme-stability problem doesn't guarantee a fast path to market if the regulatory question is still unresolved when the science arrives.

If you want one concrete signal to watch for, rather than a vague sense of momentum, it's this: a published titer figure meaningfully above 510 mg/L, appearing alongside a named commercial partner announcing an actual production or supply deal. Either one alone is easy to overinterpret. A higher titer in an academic paper without a commercial partner tells you the science moved but says nothing about who's willing to build a plant around it. A commercial announcement without an accompanying titer figure -- which is largely what the Cronos-Ginkgo trail gave us after mid-2022 -- tells you a deal exists but not whether the core enzyme problem got solved or simply worked around at small scale.

Until both of those things show up together in the public record, the responsible read is that cheap biosynthetic CBG remains a plausible mid-term outcome, not a scheduled one. The science is real, the 2018 partnership was real, the shelf presence in Canada was real -- and CsPT4's instability, as of the most recent peer-reviewed literature, is still real too. Anyone putting a specific year on commodity-priced fermented CBG is filling in a gap that the published research hasn't filled in yet.