The break, seen from the other slope

A recomposition is under way in Silicon Valley: companies with one to two hundred employees will soon be generating billions in revenue, at valuations running into the hundreds of billions. Mountain View sees this as an internal rationalization. Seen from Valais, it is also, and perhaps above all, a territorial dilution taking shape.

Something is going to happen in the coming years that has never been seen before. Companies with one to two hundred employees are going to generate billions in revenue, at stock-market valuations in the hundreds of billions. The movement is already under way in California; it can be read in the announcements of mass layoffs at the big tech companies, in share prices that climb just as headcounts fall, and in what operators themselves are saying, from inside the ecosystem, about what they see coming. I see in this far more than one news item among others: a recomposition of the grammar of skilled work. And what a canton like Valais makes of it, from the other side of the Alps, will determine over the coming decade whether this turns out to be a window or a sentence.

Setting the scene

Silicon Valley covers a hundred square kilometers (the size of the Creuse) and produces, on its own, eight hundred and forty-five billion dollars of value added a year. If the zone were a country, it would rank twentieth in the world. Its density belongs to a different order than that of any other technology hub on the planet: three hundred and twelve unicorns are active there, against twenty-five for the whole of France, and its GDP per employee runs to roughly three hundred and fifty thousand dollars, seven times the French average. None of this concentration is accidental. It is the product of a century of military investment, university policy, and an accumulation of capital that has been reinforcing itself ever since the end of the Second World War.

Seven companies from this zone, the Magnificent Seven, now account for thirty-two percent of the market capitalization of the five hundred largest American companies, against twelve percent ten years ago. Their rise says more than any other figure about what this decade has produced. These are less bubble numbers than tipping-point numbers.

Why this break is different

The American ecosystem has gone through five waves of innovation since the war: the personal computer, software, the internet, mobile, social networks. Each produced its winners and its losers, but all shared one trait in common: to seize the moment, a team of engineers, a modest amount of seed capital, and a product were enough. The sixth wave, the one now opening with generative AI, no longer obeys that grammar. It demands data centers built at continental scale, specialized semiconductors made by a handful of foundries, an energy supply capable of absorbing loads that rival those of entire states, and a software orchestration layer that only a few companies master. It is a bet on infrastructure, and that is what sets it apart from everything that came before it.

This difference has a direct territorial consequence. Infrastructure is not built in a garage. It calls for equity and debt that only a few players can absorb, and by its very nature it concentrates in the places that already have it: California, Texas, the state of Washington, and now a handful of Gulf emirates. In the near term, this concentration will only deepen, and the infrastructure layer will close around five to seven global players. Above it, though, something different is opening up, and that is where the movement becomes interesting for Valais.

What is shifting above the infrastructure layer

Above the infrastructure layer, the nature of skilled work is changing. For two decades, the economics of a skilled-services firm came down to a simple formula: a headcount, multiplied by the value produced per employee, with a diminishing scale effect past a certain size. The big firms won because they had the headcount; the small ones lost because they didn't. Generative AI breaks that equation. A team of five people, properly equipped, now produces what a team of thirty produced five years ago, and in some segments the ratio is even more brutal. The factor of four or five that I observe within my own group turns up, in other forms, in almost every field of intellectual work I see operating around me.

The consequence in Silicon Valley is immediate and brutal: the big tech companies are laying off staff en masse, and their share prices are climbing. The market is ratifying what operators themselves describe, namely that headcount has stopped being a reliable proxy for value produced; on some functions it has even become a liability. The prediction that opens this piece is therefore already under way, and what the players themselves are preparing confirms it quarter after quarter. In the years ahead we will see companies with one to two hundred employees emerge, capable of generating billions in revenue. Block, Microsoft, Salesforce, Amazon: every quarter brings its share of four-figure layoffs at companies that remain, all the while, profitable.

What is shifting, at bottom, is the nature of what we call capital. For two centuries, productive capital consisted first of hands, machines, and organizations that made them work together. In the new cognitive economy, it shifts toward imagination, experience, and the capacity to architect a problem complex enough that no machine can formulate it alone. That capital has left the hands. It now resides in the heads of people in their forties or fifties who have been through enough complex problems to recognize their underlying structures.

The same break, read in both directions

This is where reading it from the other slope becomes strategic. Seen from Silicon Valley, this break is an internal recomposition. Block lays off four thousand people in San Francisco, and the senior staff orchestrating what's left are themselves in San Francisco, Austin, or Seattle; the junior engineers who can no longer find their first job are Californian, and they will adapt, since the American ecosystem has a capacity for bouncing back that few European observers take seriously. That reading is correct, but partial.

Because the Californian slimming-down is not playing out only within Silicon Valley. Structurally, it is also a displacement outward, and that is the movement that non-metropolitan regions with a strong resident base of skilled human capital can capture. A Californian company with a hundred and fifty employees generating a billion in revenue does not need those hundred and fifty people gathered on Sand Hill Road. It needs a few dozen senior architects, paid at San Francisco Bay Area rates, and a substrate of operators capable of orchestrating the tools. That substrate can be located elsewhere: in Lisbon, in Tallinn, in Sierre, in Sion, or in Martigny.

What Valais can make of it

Beyond this exported substrate, the same grammar applies to local players themselves. A four-person tax firm in Valais now handles files it would not have touched five years ago. A valley medical practice gains access to document-analysis capabilities that no hospital library could match. A trust and fiduciary firm delivers to a Geneva client work that, until yesterday, only an international firm could have delivered. The bisse cognitif is working. It does not create the expertise, which Mountain View will always produce in greater quantity; it moves it. And that displacement, like the water in the bisse, looks minor right up until you look at what it makes possible.

What is still needed, at the far end of the bisse, are human assets capable of architecting and orchestrating. Capturing water where it is abundant, carrying it to the parcels that lack it, sharing it out according to transparent rules: that is the historical grammar of the bisse, and it is also the grammar of the competitive shift now beginning. Without resident senior architects, the water passes through without irrigating anything; without governance, it floods, or it is lost. That is why, to my mind, the priority is not so much rushing to train juniors in the tools — the market will take care of that — as it is training the seniors who will know how to orchestrate, and retaining, attracting, and passing on the skills that make that orchestration possible.

The window

This window will not stay open indefinitely. The European regions positioning themselves now will capture the movement; those that wait will watch comparable neighbors organize ahead of them and will inherit whatever is left over. Valais holds, through what its history has bequeathed it (living communal institutions, available energy, habitable geography, working bilingualism, a growing resident pool of skilled human capital), a rare combination of assets to seize it. That combination will produce nothing without explicit choices made in the decade now opening.

The same break can be read in two directions. Seen from Mountain View, it looks like a rationalization; seen from Nendaz, like a dilution. I believe it is both at once, and that the regions which understand this now — not in 2030, not in 2028: now — will turn it into a window rather than a sentence.