🇺🇸 I've been wondering why Silicon Valley keeps falling back on software after trying at least twice to reindustrialize America. The answer, I believe, is the cost of capital in a financialized economy. In today's America, return-hungry investors have calibrated their expectations on tech: gross margins above 80%, near-zero marginal cost, light balance sheets, and exits within 5-7 years. By contrast, manufacturing comes with heavy upfront capex, long lead times, inventory risk, and gross margins that tend to sit at 15-20%. With hurdle rates of 25-30%, a factory rarely clears the bar. As a result, capital flows to software, financial assets, and buybacks. 🏗️ The first attempt to reindustrialize began in April 2020, when Marc Andreessen published "IT'S TIME TO BUILD", blaming America's shortages of masks, ventilators, and housing on a failure to build. By 2022, a16z had turned the essay into its American Dynamism practice, calling on Silicon Valley to fund defense, aerospace, supply chains, and advanced manufacturing. At this point, I assume many LPs wanted to know where the margins were, and why a VC fund was considering paying for machine tools and test ranges. 🤖 The second attempt came with the blessing that was generative AI. The same industrial companies could now present themselves as AI-native robotics, autonomous defense, or smart factories, legitimately priced at tech multiples. Some raised at 30 to 50x revenue. The problem is, this might create a valuation problem down the road. Hardware costs fall along Wright's Law through cumulative production over many years. But a company priced like tech has to deliver tech-like cash flows to grow into its valuation. If it can't, down rounds and recapitalizations are bound to follow. I think Silicon Valley is already seeing the writing on the wall. As it rediscovers permitting delays, supply chain bottlenecks, and thin margins, the pitch is gradually moving back to code and the "agentic economy." A company such as Instinct has nothing to do with reindustrialization. The irony, of course, is that code still runs on atoms. Agents need GPUs, data centers, transformers, and gigawatts of power. Hyperscalers spend hundreds of billions a year on infrastructure, far ahead of what AI software earns. Investors expect software returns from something that increasingly resembles a utility buildout. I conclude that past a certain level of financialization, private capital markets alone struggle to reindustrialize a country. Hurdle rates are too high and horizons too short. As a result, building physical businesses requires a speculative narrative for private investors or the government as a buyer (or both). And because the plane can't really land from a purely financial perspective, physical ambitions tend to drift back towards software stories. Meanwhile, Elon Musk stands as the rare exception of a CEO persisting in the narrative of building tangible things. 🚗 Back in 2021, I argued that Tesla was a car manufacturer whose returns to scale could only go so far. Musk knew it perfectly, so he bet on the narrative: the market for electric cars would grow exponentially and Tesla would lead it. His storytelling became a pillar of Tesla's valuation, giving a mere carmaker access to capital at tech multiples. (That was before the rise of BYD and the demise of Tesla as the leading EV builder in the world.) 🏛️ Then he went after the government as a buyer. He bought Twitter in 2022, turned it into X, and used it to enter the political arena, all the way to his stint in the Trump administration in 2025. The goal clearly was a closer relationship with Washington. 🚀 Now his attention has shifted to SpaceX, which combines all three ingredients: reindustrialization powered by AI (rockets, Starlink, and now xAI), a speculative narrative (Mars), and the government as anchor customer (NASA and the Pentagon). Musk is (too) often praised as an industrialist or a first-principles manager. I think his main quality will prove to be his absolute lucidity about the state of financial markets in America./.

To clarify, I think the difference between the US and any European country, from a tech company-building perspective, is as follows: 1️⃣ The US runs a huge trade deficit with the rest of the world, which creates a mirror effect: much of the world’s capital ends up in dollars and wants to be invested in the US. The corollary is that any company that's in the US and raises in dollars becomes a magnet for capital. This, not the notaries, explains why US companies have an easier time raising funds than German ones. 2️⃣ The current techno-economic paradigm, the one of semiconductors, computing and networks, has a feature familiar from software economics: you need very high up-front investment to reach scale and start generating effectively unlimited returns at near-zero marginal cost. This rewards any ecosystem where risk seed capital is abundant. In other words, it rewards the US, and Silicon Valley in particular. 3️⃣ It obviously helps that an entire segment of the financial services industry, venture capital, has emerged since 1979 to channel part of that abundant capital towards innovative companies. In fact, that may be Silicon Valley’s main contribution to the American company-building machine: the flywheel through which it attracts talent, which can then raise capital relatively easily, which in turn attracts more talent. 4️⃣ More specifically, US venture capital is funded by what I call the “overflow mechanism”: domestic capital, such as endowments and pension funds, is crowded out of safe assets by foreign capital and has to move up the risk curve to find proper returns. Sure, favorable regulation and a degree of home bias reinforce this effect: it is easier for a US LP to fund a US GP. Still, the end result is that a fully functioning venture capital industry exists at scale only in the US, and it is a by-product of the US acting as a magnet for foreign capital. 5️⃣ The US has plenty of legal and regulatory problems comparable to those damned German notaries and the mandatory role they play in forming companies. But everyone ignores the US problems because capital is abundant, making it possible to pay lawyers to deal with all that and leave founders largely unbothered. Note that the US stands out as the country with the most lawyers per capita in the world. It’s not even close. 6️⃣ America’s advantage will disappear if one of two things happens: (i) The trade deficit disappears, leading to a reversal of capital flows. That would also essentially mean that the dollar ceases to be a reserve currency and Americans have to pay more for less, that is, a hit to their living standards; or (ii) We shift to a new techno-economic paradigm, one that does not reward massive up-front investment as much as semiconductors, computing and networks do. That would typically happen if the next paradigm has more in common with industrial and manufacturing economics than with software economics... and it definitely looks as though we’re heading that way. 7️⃣ So perhaps the mistake is to assume that Germany is bad at building companies simply because it is less well suited to the current paradigm. If the US trade deficit eventually unwinds and capital starts flowing back towards Europe, while the next techno-economic paradigm rewards industrial depth, manufacturing capability and the ability to turn capital into physical assets rather than software-like returns, German founders may find themselves with the home advantage. I'm 100% sure the notaries will still be there, but they will matter a lot less. Cc @christianmiele @MangesiusH @patrickc @michaelxpettis