Anthropic is reportedly inching toward public markets, and the framing writes itself: another San Francisco AI lab, another billion-dollar listing, another chapter in the Silicon Valley story you've heard since the iPhone launch. The Claude-versus-ChatGPT rivalry slots into a familiar plot. Two brilliant labs, a few square miles of Bay Area real estate, the future getting decided over catered lunches. It's a satisfying story because it asks nothing of you. You already know the cast. But the question hiding under the IPO chatter has less to do with which lab wins and more to do with where the thing they're selling actually gets built. The chips. The capital. The factories that turn a research paper into a product. Anthropic's road to the market runs through places that rarely make the headline.
The IPO coverage skips over something simple. An AI company can sit three blocks from its rival and still depend on a supply chain that touches half a dozen countries before a single model trains. The advanced chips come from Taiwan. The machines that print those chips come from a single Dutch firm. The capital flows in from sovereign funds and pension money with no particular loyalty to California. Read that Anthropic is stepping up its race with OpenAI and the contest looks local. The dependencies are not. Whatever valuation a banker assigns to either company assumes a global production system holds together, and that assumption gets almost no airtime. The story sold as a two-horse San Francisco race is a story about who controls the inputs, and most of those inputs sit very far from Sand Hill Road.
Mehran Gul's The New Geography of Innovation, a Financial Times Best Book of the Year, maps this gap between where technology gets credited and where it gets made. For fifty years, Gul argues, Silicon Valley held a near-monopoly on producing fast-growing, billion-dollar companies. The Apples and Googles weren't just American successes; they were proof of a place. That monopoly is coming apart, and Gul follows the pieces to where they've landed. The specifics give the book its bite. Samsung, the South Korean conglomerate, rivals Apple as the largest maker of smartphones on earth. Arm, founded in Cambridge, designs the chips inside more than ninety percent of the world's mobile devices.
Spotify came out of Sweden and runs global music streaming. TSMC in Taiwan and ASML in the Netherlands hold the semiconductor industry together at points so precise that a single factory outage can ripple through every gadget on your desk. Gul, who won the Financial Times and McKinsey Bracken Bower Prize, traces how three things scattered outward from California: technical skill, manufacturing capacity, and money. None of them moved together or for the same reasons. That split matters for the Anthropic question. A lab can keep its research talent in San Francisco while the physical basis of its product gathers somewhere else, in places that answer to different governments and different incentives.
The word I'd argue with is "breaking apart." The evidence Gul assembles shows something more lopsided than a clean dispersal. Capability has spread, but it has spread into a handful of new chokepoints rather than a flat, open field. ASML is one company. TSMC's most advanced fabrication sits on one anxious island. That isn't the end of concentration; it's concentration relocated and, if anything, made more brittle. Gul sometimes reads the spread as democratization when the same facts support a darker reading about single points of failure. The framework still earns its keep against the IPO news.
Gul hands you sharper questions than "who's winning." You start asking which inputs a company controls, which it merely rents, and what happens to a sky-high valuation when a rented input gets cut off by a trade dispute or a drought near a fab. The book predates the current AI funding frenzy, so it won't name Anthropic. It doesn't need to. The pattern it describes was already running before the chatbots arrived, and the AI labs are simply the loudest current tenants of a system built well outside their walls.
An IPO is the moment a private bet becomes a public price, and a price is a story everyone agrees to believe for a while. Gul's book is good company for the skeptic who wants to know what that story leaves out. Read it not to predict whether Anthropic lists this year or next, but to understand the physical world the listing depends on. The people in Bengaluru and New York filing the reporting are tracking the race. This book tracks the track itself, and that turns out to be the more durable thing to know.