Case 1 · Reports 01 Performance and 03 Economics
Artemisinin is the core of the main malaria treatment and normally comes from a plant, sweet wormwood. Between 2004 and 2013 a Berkeley lab, the company Amyris and the drug maker Sanofi taught yeast to make its precursor. The science worked. The factory stopped two years later.
Figure 1 · Two routes to the same molecule
Artemisinin is the malaria drug. A plant makes it; yeast makes something close to it and chemistry finishes the job. Both routes sell into the same market, which is the whole story of this case.
The plant route
The yeast route
Artemisinin · C₁₅H₂₂O₅
Both routes end at this one molecule, so a buyer compares them on price alone.
The orange bond is the oxygen–oxygen bridge. It is the part that acts on the malaria parasite, and the part that has to be installed chemically after fermentation — which is where 45 to 60% of the material is lost.
Each number belongs to a different report. Together they explain the outcome. Any one of them alone would mislead.
Rise in yeast titer of artemisinic acid, from 0.1 g/L (2006) to 25 g/L (2013).
Report 01 · Ro 2006, Paddon 2013Overall yield of the chemistry that turns the yeast product into the drug. Bench route 40%, Sanofi route about 55%.
Report 01 · Kung 2018Design capacity of Sanofi's plant in Garessio, Italy. About a third of world demand.
Report 01 · PATH 2014, Peplow 2016Sanofi's no-profit, no-loss price. Plant artemisinin sold below $250/kg in 2014–15.
Report 03 · Peplow 2016The titer climbed in steps, and each step came from a specific change to the strain. A good TRY report shows this path and names the scale every number was measured at.
| Year | What changed | Result | Scale | Source |
|---|---|---|---|---|
| 2006 | Mevalonate pathway plus two plant enzymes (ADS, CYP71AV1) put into yeast | 100 mg/L acid | lab | Ro et al., Nature |
| 2012 | New yeast background, full pathway upregulated; switched to the simpler intermediate amorphadiene | >40 g/L amorphadiene | fermenter | Westfall et al., PNAS |
| before 2013 | Earlier attempts at commercial acid titers | 1.6 g/L acid | not stated | as cited in Paddon 2013 |
| 2013 | Added a plant dehydrogenase and a second cytochrome; lowered CPR expression | 25 g/L acid | 2 L fed-batch | Paddon et al., Nature |
| 2013 | Acid converted to artemisinin by chemistry, four steps | 40% overall | bench | Kung et al. 2018 |
| 2013–18 | Sanofi photochemical route, semi-batch | ~55% overall | up to 60 t/yr | Kung et al. 2018 |
The 25 g/L result came from 2 L fermenters. Sanofi then spent 2008 to 2013 turning the chemistry into a factory process. Actual tonnes produced per year were never published; only plans and capacity were.
Figure 2 · Seven years of strain work
Each dot is a published titer. The axis is logarithmic, because a linear one would flatten the first two into the baseline.
| Milestone | Year | Artemisinic acid |
|---|---|---|
| Ro et al., first engineered yeast | 2006 | 0.1 g/L |
| Earlier attempts at commercial titers | not stated | 1.6 g/L |
| Paddon et al., final strain | 2013 | 25 g/L |
Figure 3 · What actually reaches the drug
Fermentation output is not product. The chemical step after it sets how much of the titer survives.
Reaches artemisininLost in the chemistry
"Industrial-level" has no number. "Achieves" hides the scale. The sentence skips the chemistry step, where up to 60% of the material is lost.
Scale, number, unit, source, and the next loss in the chain, all in one paragraph.
Sanofi's break-even was about $350–400/kg. The plant crop's price moves with the harvest, so the comparison is not against one number but against a range that swung sixfold in a decade.
Figure 4 · The price it had to beat
Break-even is a line. The competitor is a range that moves with the harvest.
| Year | Plant artemisinin | Against break-even | Source |
|---|---|---|---|
| 2005 peak | $1,100/kg | yeast route wins | PATH 2014 |
| 2007 low | $180/kg | yeast route loses | PATH 2014 |
| 2011 average | $530/kg | yeast route wins | PATH 2014 |
| 2014–15 | under $250/kg | yeast route loses | Peplow 2016 |
True in 2011, false in 2015. A single comparison point hides the risk that ended the project.
Break-even, the competitor's full range, and which side of it the product sat on.
Sanofi's buyers judged 25 g/L multiplied by 40–55%, not 25 g/L. For us: lumen signal, then what crosses the membrane, then what the shell sample shows. Each step gets its own number or the words "not measured".
"2 L fed-batch" is in every strong sentence above. Our equivalent is the loop volume, operating mode (A, B or C) and cross-flow setting for each run.
When the BOM exists, compare cost per run against the full range of what a farmer pays for the incumbent, not one price. Until then, the claim ledger says "not yet established".
Brackets are filled from the Bioreactor Report builder. If a bracket cannot be filled, the sentence says so.
Numbers checked against the abstracts or full text listed. Peplow figures are from the Scientific American reprint of the Nature article. Gates funding totals differ by source ($42.6 M in 2004, $53.3 M by 2013, $64 M per Peplow), depending on what is counted.