Moët Hennessy AI aims to flag a wine fault before it shows

The Moët Hennessy AI project with chipmaker Analog Devices and the University of California, Davis has produced its first result: an early warning for fresh mushroom aroma, a wine fault that usually surfaces only after fermentation. The partners say their sensors flagged at-risk samples before conventional methods could, but they have published no figures to back the claim.

The collaboration was announced on 7 October. Analog Devices (ADI) supplies a sensing and machine-learning platform that grew out of research at MIT. UC Davis brings its Department of Viticulture and Enology. Moët Hennessy, the wine and spirits arm of LVMH, brings the samples and the production data.

The system reads volatile chemical signatures given off by vines, musts and wines. Instead of testing for a fixed list of known markers, the software decides which signals in a broad chemical profile matter most. The training set is a library of samples assembled by Moët Hennessy's Robert-Jean de Vogüé Research Center, near Épernay.

A warning that arrives before fermentation is only useful if someone can check how often it is right.

Why fresh mushroom aroma is hard to catch

In Champagne the off-flavour has been turning up in some wines since 2005, mainly in Pinot Noir and Meunier, according to a 2022 study in the journal Molecules co-authored by researchers from the Comité Champagne. It is linked to moulds that develop alongside grey rot and to C8 compounds such as 1-octen-3-one. The same paper notes that the defect is not detected in musts and appears at the end of alcoholic and malolactic fermentation, when the options for a base wine are far narrower.

A tool that reads the risk earlier would let a cellar isolate, treat or downgrade a lot before it reaches the blend. Manuel Reman, the Moët Hennessy executive committee member in charge of research and development, speaks of reading the "chemical language" of grapes and wine. Max Shulaker, who heads ADI's Health Solutions unit, talks of moving quality control "from reacting to anticipating".

What the Moët Hennessy AI claim still lacks

The announcement gives no sample size, no detection rate, no false-positive rate and no indication of how much warning the system provides. No peer-reviewed paper came with it, as the trade site Vinetur pointed out. The partners describe a decision-support tool for winemakers rather than a replacement, and set no date for commercial use.

Ben Montpetit, professor and chair of the UC Davis Department of Viticulture and Enology, frames the goal as giving growers and winemakers the right information at the right time.

After a 2026 harvest at a modern-era low, every lot Champagne can save counts. Whether these sensors can predict the fault outside one company's sample library is the part still to be shown.