Things You Can Drop on Your Foot

Is a venture market that shrinks 15.5% in the biggest funding year on record in trouble? Swiss startups raised CHF1.25B in the first half of 2026. Next to $412.7B in the US and €44B in Europe, the answer looks obvious. It is the wrong answer. The American number is one trade: 86% of it went to AI, and three firms, Andreessen Horowitz, Thrive and Founders Fund, took 48.1% of everything venture managers raised. In Europe, 60.3% of the €44B is the same trade, and its two biggest rounds, Nscale and Isomorphic Labs, sit in London. Brussels is drafting a Savings and Investments Union. The money has a UK postcode.

What a venture market buys matters more than what it spends. New York and London are buying an outcome, the model that wins, at a price that only holds if it does. The Swiss top ten bought parts. Kandou, a spinout of EPFL, Lausanne's federal engineering school, took $225M for copper interconnects between processors and memory, sold to whichever model wins. Terralayr, out of Zug, the low-tax canton, took €192M to build grid batteries in Germany. CeQur took $100M for an insulin patch worn four days without tubing. SWISSto12 got €73M for small geostationary satellites from ESA, which Switzerland co-founded in 1975.

Polares does mitral valves. Acutronic builds the rigs that test missile guidance. One pure software company made the list, BLP Digital, $50M from Goldman. Everything else you can drop on your foot. Hardware alone pulled CHF324M, 60% above the previous record. Switzerland's AI exposure is copper.

Biotech fell 74% to CHF184M, and molecules are the one thing here nobody can drop on a foot. Zurich had a bad half while the canton of Vaud, home to EPFL, took three of the four hardware rounds in the top ten. Nobody in Lausanne sent flowers. Foreign investors supply 88% of Swiss late-stage funding above $100M. The second pillar, Switzerland's compulsory occupational pension system, sits on more than CHF1T and would rather buy an office block in Zurich than a Series D in Lausanne. A federal growth fund would solve the wrong problem. The money is already in the country. It declines to move.

Strategic investments doubled to 14, ENGIE and Johnson & Johnson among them, and strategics buy parts, not outcomes. Slow. Copper is slow too.

CHF1.25B is where Switzerland stood in 2023 and 2024, and none of it gets handed back when the AI trade reprices. No state champion, no sovereign fund chasing the consensus, 26 cantons competing for the same spin-offs and a pension system too cautious to join the party: the portfolio that falls out is one no ministry would have dared to design. Chips, batteries, insulin and satellites, things that still exist after the next earnings call. Small countries that refuse to buy the winner end up owning what the winner needs.