Vines, Cheese, Alpine Pasture

The three preceding chapters laid out an inheritance: communal institutions descended from Törbel and from further back still, an intangible heritage partly listed with Unesco, a linguistic landscape of rare singularity. This inheritance forms a kind of capital. But capital lies dormant until something puts it to work. The chapters that begin here move down into the trades where that capital takes shape, where it becomes economic value, or fails to. It is in the vineyard, in the cheese dairy, on the alpine pasture, in medical and consulting practices, in the hotels of the valleys that the thesis of this essay holds up or falls apart. If tradition really is the infrastructure of modernity, and not merely its charming relic, it has to produce measurable effects in these places.

In these concrete settings a transformation is under way that goes almost unnoticed: the arrival of artificial intelligence in trades handed down for centuries through gesture, eye, and habit. The cognitive bisse from the first chapter — the traditional irrigation channel that serves here as a figure for a flow of knowledge running from the high ground down to the fields — is already producing its effects here. Upstream, it gathers the expertise, the documentary resources, the technical capacities of urban hubs, and makes them available close to the terroirs and to the practitioners who had gradually been cut off from them.

Three trades concentrate the economy of Valaisan authenticity: the vine, the cheese, the alpine pasture. Small in scale, remarkably diverse, precarious in their transmission, bearing a territorial signature that proves itself in the glass or on the plate.

The vine: a genetic and fragmented inheritance

Valais is by far Switzerland's leading wine region. Nearly five thousand hectares are planted along the right bank of the Rhône²⁵, from Martigny to Loèche, a third of the entire national vineyard area²⁶. Some fifty grape varieties are grown, about ten of them indigenous²⁷: Petite Arvine, Cornalin, Humagne Blanche, Humagne Rouge, Amigne, Rèze, Païen — varieties found almost nowhere else on earth. Europe's highest vineyard, at Visperterminen, rises above a thousand metres²⁸. Behind these figures lies a less flattering reality: twenty-two thousand owners share some eighty thousand parcels across sixty-five wine-growing communes²⁹, making this one of the most fragmented vineyards in the world.

This fragmentation is both the glory and the weakness of Valaisan viticulture. Its glory, because it guarantees a diversity of terroirs, microclimates, and practices that few regions can claim. A Cornalin from Salquenen does not taste like one from Saillon. The Petite Arvine from Fully carries a salinity that the one from Chamoson does not quite share, because the bedrock differs: the soils of Fully extend the granite base of the Mont Blanc massif, while those of Chamoson rest on alluvial fans of another nature altogether. This fineness of calibration, which no industrial standardisation could ever reproduce, is what justifies the best prices.

Its weakness, because fragmentation complicates everything. The vignerons-encaveurs — winegrowers who vinify their own grapes into their own wine — number only a few hundred. Alongside them survive tens of thousands of owners of small plots, often just a few ares, passed down from generation to generation. A whole category of Sunday winegrowers — people with a second occupation, retirees, heirs — are now reaching the age at which succession becomes urgent, and a new generation is not always waiting in the wings. The cooperatives, which pool the grapes of small owners, remain a pillar of the sector but face the twin pressure of falling Swiss consumption and competition from foreign wines. In recent years the canton has gone through a crisis of overproduction that has forced the declassification of several million litres into ordinary table wine, amid the discreet silence of an industry that does not like to advertise its troubles.

It is into this two-speed landscape — a few hundred artisanal elite producers on one side, a long tail of fragile smallholders on the other — that AI is making its entrance. And it does not enter the same way depending on where one stands.

What AI can do for the vine

For quality vignerons-encaveurs, AI arrives as one more assistant, one that does not change the nature of the trade but widens its possibilities. It helps anticipate harvest dates by cross-referencing fine-grained weather data with maturation history, plot by plot. It supports precision viticulture by interpreting satellite or drone imagery to map water stress, incipient fungal disease, and variations in vine vigour. It eases the documentation of origin: every bottle can, at negligible cost, carry a rich digital record of its plot, its vintage, its winemaking conditions. It speeds up administrative procedures, multilingual communication with foreign clients, the drafting of technical sheets. For estates that remain artisanal by choice, this is a valuable multiplier that does not put their identity in question.

There is more to it, and this is where the competitive shift described in earlier chapters gains real weight in the vineyard. For two decades, Valaisan winegrowers have watched their margins squeezed from two distinct directions. From below, foreign wines produced at far lower cost — from southern Italy, Spain, South America, and more recently Eastern Europe — have taken over the entry and mid-range segments at prices Swiss producers could not match. From above, the big international houses and their marketing teams have captured visibility in guides, competitions, and the specialist press, deploying resources few Valaisan estates could hope to equal. In between, the Valaisan winegrower held their ground on what they did better than either side: the singularity of their grape variety, the precision of their terroir, a direct relationship with their customers.

Generative AI, well handled by a vigneron-encaveur who knows how to orchestrate the tools, changes this equation. The tasting note for each cuvée, in three or four languages, which once would have required an agency's help, can now be drafted in-house in a few hours per vintage. The entry file for an international competition — translated, formatted, illustrated — comes within reach of a five-person operation. The estate's digital presence — a multilingual website, regular editorial content, replies to inquiries from abroad — once dependent on costly outside providers, becomes sustainable for a single part-time staff member. Marketing for the estate, a cost long impossible to justify for a mid-sized operation, becomes economically viable at a quality level comparable to the major houses. With comparable resources, the balance of power between a fifty-hectare estate in Burgundy and a fifteen-hectare estate in Valais shifts, because the gap in human resources now weighs less heavily on content production and command of an international presence than it once did.

For fragile smallholders, the stakes are different, and more troubling. AI can help them: a plant-health diagnostic tool that works from a photo of a leaf, an assistant that drafts declarations to the cantonal viticulture office, an interface that tracks price and stock trends. It can also hasten their marginalisation if the larger players claim it first. The risk is not that AI will destroy the Valaisan vineyard, but that it will widen the gap between an equipped elite and a mass of downgraded amateurs, and that the transmission of small plots will break down even more abruptly than it would have without it.

Public policy has a role to play here. Making pooled, accessible AI tools available to smallholders, designed around their actual circumstances, is not a technological whim. It is a condition for the wealth of the Valaisan vineyard — which owes as much to its elite as to its long tail — to survive into the next generation. The canton, the cantonal viticulture office, the Interprofession de la vigne et du vin, and Valais research institutions (HES-SO Valais, Agroscope) have an obvious field of action here, one that only needs organising.

The productivity gains these tools offer do not switch on simply because the tools exist. They depend, upstream, on a feel for terroir that is not taught in any school: a fifty-year-old vigneron-encaveur knows, plot by plot, what the subsoil yields there, at what moment phenolic ripeness sets in on a given slope, and how to tell these facts to a Japanese sommelier or a New York buyer in a way that makes them desirable. This dual mastery — of the terroir and of its telling — is what AI extends without replacing. A digital assistant run by a young staff member produces competent technical sheets, but cannot pick out the vintage detail that tips a tasting one way or the other. The transformation of the Valaisan wine trade will turn less on training new entrants, who will pick up the tools quickly, than on the canton's capacity to make experienced vignerons-encaveurs the architects of their own strategy for communication, sales, and succession.

The special case of indigenous grape varieties

Another matter, quieter but likely decisive over the long run: the digital valorisation of indigenous grape varieties. Petite Arvine has grown from under forty hectares in 1991 to around two hundred and fifty hectares today³⁰. It stands as a model case of revival: a variety that had nearly disappeared and reclaimed its place through the patient work of winegrowers and oenologists, and through growing international recognition. Cornalin, Humagne Blanche, Amigne, and Rèze have followed comparable trajectories.

That revival is not complete. For it to continue, the indigenous grape varieties of Valais need to be documented in fine detail — not merely through administrative technical sheets, but through everything that makes up their genetic, sensory, agronomic, and historical singularity. In the age of AI, this documentation can become an asset in its own right. Imagine a structured corpus, open or semi-open, bringing together for each indigenous variety the full body of accumulated knowledge: DNA profiles, ampelographic descriptions, sensory analyses, vintage data, historical references, winegrowers' testimonies. Such a corpus cannot be built in a year, but it is technically within reach. It would become, internationally, the reference that any sommelier, any serious enthusiast, any oenological AI model would turn to when discussing Valaisan wines. Through the sheer rigour of its documentation alone, Petite Arvine becomes more citable and more recognisable than comparable varieties that never underwent this work.

This is an investment in cultural sovereignty in the strictest sense: deciding how our grape varieties are represented in the world's wine culture, rather than letting that representation be built elsewhere, without us.

Alpine cheese: a territorial economy

Let us turn to the high pastures. Valaisan alpine cheese, best embodied by Raclette du Valais AOP — a protected designation of origin, comparable to an appellation — rests on a dense territorial chain that few cantons can match. The sector today counts around three hundred and forty milk producers, twenty-five valley-floor dairies, some fifty alpine dairies³¹, and several certified maturing cellars. Close to eight hundred people work in it, equivalent to roughly five hundred and seventy full-time positions³². The sector's direct value added reaches fifty-two million francs a year³³ — not a major economic weight at cantonal scale, but a real one in the economy of the mountain valleys where it takes place.

This sector carries a historical particularity that ties it directly to Chapter 4 on the commons. Summer dairy farming remained for a long time, and in some alpine pastures still remains, under the governance of the bourgeoisies and the consortages — traditional cooperative bodies that manage shared grazing land and irrigation. The high pastures, the collective processing of milk, the making of the wheels, the upkeep of alpine chalets, and the allocation of shares among rights-holders were managed for centuries by alpine cooperatives whose workings follow the principles examined earlier. Valaisan alpine cheese is therefore not merely a product. It is, inseparably, the product of an institution. And it is precisely this bond between product and institution that gives it its value, distinguishing it from a comparable industrial cheese made elsewhere under a more permissive set of standards.

Raclette du Valais AOP, listed on the federal register since 2007, gives this singularity legal codification³⁴. Producing the milk, processing it, and maturing the cheese must all take place exclusively within the canton. The milk must be raw, processed in artisanal dairies, traditionally in copper vats. The cheese must rest on boards of untreated spruce and age at least three months if destined for melting. Every wheel carries a casein mark ensuring traceability. And only cheese made with milk from the high pastures may bear the designation "fromage d'alpage": the roughly seven hundred tonnes produced annually by the alpine herders³⁵ during the few weeks of summer transhumance.

What AI can do for the dairy sector

On the operational side, AI can support production. Sensors on cows to anticipate illness and heat cycles. Real-time monitoring of milk quality. Grazing-prediction models that help manage herd rotation according to grass growth and weather. Automatic logging of milking and cheesemaking operations to meet the AOP's traceability requirements without adding to the daily workload. For alpine dairies that often run on two or three people through the summer season, where the decisive judgment calls still rest on the cheesemaker's eye over the copper vat, this kind of support is more a relief than a transformation. It frees up administrative time so that time for the craft itself can be preserved.

On the valorisation side, AI can do a great deal more, and this is where the competitive shift plays out for the sector. Raclette du Valais AOP faces constant competition from other raclette cheeses made in France, Germany, or elsewhere in Switzerland, under looser standards and at far lower production costs. Legal protection under the AOP label is indispensable but not sufficient: most consumers cannot tell a raclette from Valais apart from one made in the lowlands with no restrictions whatsoever. The price premium meant to justify the origin is constantly under threat from the genericisation of the common name.

AI will not solve this problem on its own, but it can powerfully equip the defence of the Valaisan signature. A territorial platform that made the living memory of each alpine pasture accessible — transcribed interviews, interactive maps, documented recipes, lineages of cheesemakers, correspondence between pastures — would turn Raclette du Valais AOP from a simply labelled product into one that is narrated and traceable back to its plot of origin. Every wheel could carry, within its casein code, a digital record tracking its path from pasture to plate. This quality of storytelling and traceability is exactly what competing industrial cheeses cannot reproduce without faking it, because they possess neither the supporting institutions, nor the embodied memory, nor the living practices that give it substance.

Production costs will always be higher in Valais than in the lowlands, and that is entirely as it should be: it is the price of quality. But the value the consumer perceives can be considerably increased through equipped, accessible, verifiable storytelling. This narrative capacity was not economically feasible five years ago; it is today. The effect is strongest in direct sales and with recommenders; in large-scale retail, where the buyer weighs volumes and margins, the storytelling feeds more indirectly into the AOP brand that justifies a place on the premium shelf. As with the vineyard, it depends on cantonal coordination between the Interprofession Raclette du Valais AOP, the dairy sector, cantonal authorities, public operators, and the political will to fund a long-term narrative infrastructure.

The alpine pasture as a system

One thing remains to be said about the alpine pasture itself as a system, apart from what is produced there. In its traditional form, the Valaisan alpine pasture is far more than high-altitude grazing land: it is an arrangement spanning village, mayen, and alpage, the three tiers that still structure the spatial organisation of many valleys. The winter village. The mayen in spring and autumn, where the family moved with its animals as the snow melted and the grass grew back. The alpage in high summer, where herds were entrusted to a collective herder. This arrangement is no longer fully alive — many mayens have become second homes, many alpine pastures have been consolidated or abandoned — but it remains legible in the landscape and in memory.

AI will not bring the three-tier system back to life. What it can do is help document it before it fades away. A detailed map of the canton's mayens, cross-referenced with their historical uses, current condition, and ownership, would become a valuable tool for land management. Documentation of alpine know-how — herd management, cheesemaking techniques, transhumance calendars — would ensure these skills are not lost along with the generation that still holds them. Analysis of environmental data — vegetation, climate, wildlife, erosion — would support the sustainable management of high pastures in a climate that is transforming them quickly.

All of this calls for political choices. None of these projects can be carried out at the scale of a single herder or a small bourgeoisie. They require cantonal coordination, partnerships with research institutions, and the willingness to fund work whose returns are not measured in immediate turnover but in heritage built for the long term. A canton is, in principle, better placed than a private actor for this kind of work, because its own time horizon matches that of these projects: long, patient, oriented toward heritage.

An equipped economy of authenticity

The vine, the cheese, the alpine pasture share a single economic peculiarity. Their value is built not on volume but on signature. No one will pay for a Valaisan Cornalin the price of a great Burgundy if it is merely one Alpine red among others. No one will pay a fair price for an alpine cheese if it is merely one raclette among others. Signature — that which makes a product recognisable, traceable, rooted in place — is the principal economic lever of these sectors, and probably the only one that holds up against international competition.

Today this signature is maintained chiefly through living traditions, protected designations, and the patient work of practitioners. In the age of AI, it can be maintained even better, through digital documentation and valorisation that makes visible what used to remain implicit. In these age-old trades, the competitive shift has nothing to do with accelerated production — the vine grows at its own pace, the cheese matures at its own speed — but everything to do with a multiplied capacity for storytelling and commerce. A family winegrowing estate that used to produce twenty thousand bottles a year without knowing how to speak to a Japanese sommelier or a New York buyer can now carry out that commercial work with the same staff it always had. An alpine dairy that never made itself visible beyond its regular customers can now document and tell the story of its lineage, its terroir, its gestures, and capture some share of the consumers willing to pay well for authenticity when they know how to recognise it.

These effects do not occur on their own. As elsewhere, it is the experienced practitioner who orchestrates the tools; without that person, AI produces generic, undifferentiated communication. With that person, it multiplies what made these trades irreplaceable in the first place.

This equipped signature requires choices about governance: who holds the corpora, who makes them accessible, under what conditions, on what legal basis. It requires partnerships between sectors, research institutions, cantonal authorities, and digital platforms. It requires a vision — the idea that a vineyard, a dairy sector, an alpine system are not merely economic sectors but elements of territorial identity that deserve to be thought through together. That vision has not, to date, been explicitly formulated at the cantonal level. It exists in scattered form among the interprofessions and at Promotion économique Valais, and would benefit from being drawn together.

Over his copper vat, eye fixed on the curdling of the cheese, the alpine cheesemaker will go on judging alone the moment to gather the curd. The gesture does not change. What changes is everything that comes before it and everything that follows it: the documentation of the lineage, the traceability of the wheel, the storytelling that reaches all the way to a plate in New York. The canton that manages to equip everything surrounding the gesture, without touching the gesture itself, will have understood what AI can do, and cannot do, for its age-old trades.