The AI Trade Cracks, and Switzerland Quietly Wins October

I noticed something odd in the October performance tables this week.

Swiss blue chips aren't supposed to lead anything. The SMI is a defensive, slightly boring index, dominated by pharma and packaged food. And yet it's up 4.6% this month while the Euro Stoxx 50 is down 3.2%. That's a gap of almost 8 percentage points, the widest since March 2020.

March 2020. That's the comparison. The last time Switzerland beat the euro area by this much, the world was shutting down and everybody wanted somewhere safe to hide.

What's actually inside the index

Here's the part most coverage skips. The SMI's biggest weights are Nestlé, Roche, and Novartis. Those three names alone account for roughly 40% of the index. None of them are AI plays. Nestlé sells coffee and pet food. Roche and Novartis sell drugs people need regardless of what Nvidia's data center order book looks like next quarter.

Then you get UBS, Zurich Insurance, Swiss Re, and Givaudan. Banks, insurers, flavorings. The closest thing to a technology exposure in the whole top ten is ABB, and even that's mostly electrification gear for factories and data centers.

Now look at what the Euro Stoxx 50 is built on. ASML, SAP, Siemens, Schneider Electric, Infineon. ASML makes the machines that make AI chips. Schneider builds the power equipment that keeps data centers from melting. SAP has spent the year selling an AI story to investors. When doubts about AI capital spending creep in, those four get hit first. Switzerland barely feels it.

Bloomberg's read is that three forces are pushing investors to cut risk in Europe: rising borrowing costs, growing doubts about the AI trade, and political turmoil in France and Spain. That's a reasonable list. Bund yields have climbed, the French budget fight has kept the OAT spread wide, and Madrid has had its own mess to manage.

But the mechanics matter more than the headlines. This isn't a rotation into Switzerland. It's a rotation out of anything that requires a story about 2027 to justify today's multiple.

The crypto connection nobody's drawing

If you've been watching the AI-adjacent crypto tokens, you've seen the same trade unwind in miniature. The decentralized compute names, the GPU marketplace projects, the ones that pitch themselves as picks and shovels for machine learning, all of them trade on the same thesis as ASML and Schneider. Except they trade with far more use and far less revenue. When the AI capex narrative wobbles, they don't wobble. They drop.

The question worth asking: if the AI capex cycle is as durable as proponents claim, why is the market suddenly paying a premium for chocolate and chemotherapy? That's not a rhetorical flourish. It's a signal about duration risk. Investors are shortening their time horizons.

And the franc is doing quiet work here too. The Swiss National Bank has been cutting rates while the ECB moves slower, and the currency stays strong in a risk-off tape. In dollar terms, the SMI's October gain looks even better.

My honest take

I'm not entirely convinced the AI bubble fear is the whole story. Granted, some of this is just rate math. Higher borrowing costs punish long-duration growth stocks everywhere, and Europe has more of them in its main index than Switzerland does.

But the composition of the move tells you something. Roche and Novartis aren't exciting. Nestlé has had a genuinely rough two years. Investors aren't buying them because they love the businesses. They're buying them because the earnings show up on schedule. That's what a market does when it stops trusting forecasts.

History suggests otherwise on one point, though. In March 2020, Switzerland's relative outperformance lasted about six weeks before the rebound made defensiveness look silly.

So what should you actually watch? Three things. ASML and SAP's Q3 guidance, because both have to defend AI spending to a more skeptical room than they faced in July. The French OAT-to-Bund spread, which is the cleanest read on whether European political risk is calming. And whether the SMI's lead holds into November. If it fades fast, October was a hedge, not a thesis.

Color me skeptical, but a defensive rotation that's driven by fear of a bubble usually ends when the bubble stops inflating. Not before.