If a YouTuber can reverse-engineer Coke, is your trade secret safe?

For 139 years, the Coca-Cola formula has been the gold standard of trade-secret-based intellectual property. A trade secret so guarded it became modern mythology. But this January, a YouTuber with a borrowed mass spectrometer technology offered a recipe that he says is a near replica that one can make at home.

Zach Armstrong (aka “LabCoatz”) notes how to make enough cola flavoring for 5,000 liters of cola. An earlier video from 2019 reached similar conclusions, but working from historical Pemberton documents, the 1886 recipe that’s been publicly available, that video didn’t use lab technology to reverse engineer the beverage. The testers in the video agreed that their beverage decidedly did not taste like Coke.

But LabCoatz claims to have gotten very close, even claiming to have “solved” the reverse engineering problem. At one point in the video, he says: “This is the mass spectrum of Coca-Cola, and this is the mass spectrum of one of my replicas. They are nearly identical.” The beverage also did well in taste tests. “I wasn’t expecting it to taste so much like Coke,” said one tester. “If you sold this on the shelves, I’d buy it,” said another.

Attempts to reverse engineer famous cola are not new, and Coca-Cola will be fine. The brand, the distribution network and a regulatory moat around coca leaf imports ensure that. But for R&D leaders relying on trade secrets to protect proprietary formulations, the video is a warning shot: the analytical barrier that once made “complexity” a viable defense has collapsed.

Armstrong notes that as recently as 2016, analysts assumed reverse-engineering a complex flavor profile would require the resources of a major pharmaceutical company. “Guess this was before science YouTubers got serious,” he says. His collaborators: two YouTuber friends with mass spectrometers, one of whom is a college professor.

Armstrong details a year of failures using Gas Chromatography-Mass Spectrometry (GC-MS), the standard tool for chemical fingerprinting. The machine kept missing the formula’s “dry, astringent” mouthfeel because of a fundamental methodological error: looking for volatiles where the answer lay in non-volatiles. The missing link turned out to be wine tannins, compounds that don’t vaporize in the gas phase of a GC-MS run and therefore remain invisible to the machine.

“Coca leaf extract is essentially a type of tea, and teas contain tannins,” Armstrong said. “These tannins possess a dry, astringent taste… And since they’re non-volatile, they don’t usually show up on gas-based mass spectrums, explaining how I missed them for so long.”

Armstrong also leaned on published literature, including a 2014 study in the Journal of Agricultural and Food Chemistry that listed measured concentrations of flavor compounds in Coca-Cola. The peer-reviewed data gave him targets; the mass spectrometer told him when he’d hit them.

Coca-Cola never patented its formula because doing so would require disclosure. Armstrong points out the consequence: “Unless I try marketing my own product under the Coca-Cola name, there’s really nothing they can do about it.”

The lesson is that “complexity” alone no longer constitutes a moat. For formulation-dependent industries, the question becomes: what can’t be detected?