{"text":[[{"start":4.95,"text":"Despite what you may have heard about an inflating market bubble, the US stock market isn’t rising. The once-hot S&P 500 has been cutting a herky-jerky path sideways for months. What happened? Back in May, the Magnificent Seven big tech stocks (Alphabet, Amazon, Apple, Meta, Microsoft, Nvidia, Tesla) surrendered the market leadership they’ve held for years. For a few weeks, a group of semiconductor stocks including Micron and Broadcom took up the baton, but recently they have faded, too. This week, tech stocks’ back-and-forth volatility has been especially intense, and a big AI-driven hedge fund, Situational Awareness, blew up. "}],[{"start":45.75,"text":"It is tempting to say that markets are going mad, as they occasionally do. Amid the excitement of the AI revolution, tech stocks have become creatures of hype and price momentum. Now they are wobbling, and a panicked sell-off seems possible."}],[{"start":61.05,"text":"There is always plenty of irrationality in stock prices, and valuations are frighteningly high at the moment. But the recent changes in market leadership do not reflect the madness of crowds. Instead, the market is struggling — as rationally as could be hoped — to answer a hard question: what is the competitive structure of the AI industry, or indeed of an economy where AI is everywhere? "}],[{"start":86.3,"text":"For investors, the competitive structure of industries, and the competitive position of companies within them, is by far the most important consideration."}],[{"start":94.89999999999999,"text":"Consider the current industrial moment. We might call it the “increasing returns era”, after a paper the economist W Brian Arthur wrote 30 years ago. Arthur argued that unlike traditional production industries, “knowledge industries” — we’d call them digital industries now — enjoy increasing returns to scale. They have high product development costs but near-zero unit costs, high switching costs and powerful network effects (that is, adding customers makes their products better). Arthur’s paper proved prescient, predicting the ever-larger geysers of cash that would issue from the likes of Microsoft, Alphabet, Meta and Amazon. "}],[{"start":133.29999999999998,"text":"Knowledge businesses, in short, have very high barriers to competition once they reach a certain size. In operating system software, internet search or social media, even the best-funded competitor cannot dislodge a scaled-up incumbent. And barriers to entry determine the pattern of stock market outcomes. Most of the excess returns generated by stock markets over the long term come from a small coterie of companies with wide moats. Economic analyst Hendrik Bessembinder has shown that half of the net wealth creation in US stock markets over the last century flowed from just 46 public companies, out of a total of almost 30,000. Looking at his list of the greatest wealth creators, the knowledge companies dominate the top spots, and all the winners have high walls around them. "}],[{"start":178.25,"text":"Not many investors active today will remember a time before the increasing return era and the dominance of knowledge companies. Microsoft was already one of the biggest companies in the US in 1995. The possibility that AI might change the rules is hard to conceptualise. But we’d better try, because knowledge is exactly what AI threatens to commoditise. "}],[{"start":199.6,"text":"The threat operates at two levels. The first is that some knowledge businesses will be made obsolete by AI tools. This possibility is what has rocked the shares of leading business software companies, the so-called “SaaSpocalypse”. "}],[{"start":213.54999999999998,"text":"The second, far more disruptive level is that AI businesses may themselves have low barriers to entry — because the technology, while capital intensive, does not generate differentiated or defensible intellectual property. Frontier AI models will converge to near interchangeability, switching costs will be low and network effects will be minimal. What is more, hardware depreciation and energy costs mean unit costs may not come down as far as they have for software or internet services. Unless model builders resort to collusion, intelligence will become a commodity, almost like electricity — which would be a bonanza for the economy, but a massive disappointment for investors. There are no electric utilities on Bessembinder’s list of big wealth creators. "}],[{"start":260.65,"text":"Things might turn out differently. But the market takes the possibility of an AI industry with low returns very seriously. That is why US tech stocks shudder when low-cost Chinese competitors release competitive models or chips. And it is why the semiconductors were unable to assume market leadership when the Mag 7 flagged: if Alphabet or Microsoft can’t make outsized profits selling AI services, chipmakers’ margins will have to fall, too. Markets, which for years have depended on the Mag 7 for leadership, will have to look elsewhere."}],[{"start":292.25,"text":"The big US tech companies are — or possibly were — the best businesses in history. What markets are weighing now is the possibility that the era of increasing returns is over, and new forms of competition are on the way. Investors already recognise that tech companies are riskier than they were a few years ago, and are re-pricing them accordingly. And they are recognising that pre-digital forms of competitive advantage, from brands to distribution networks to manufacturing knowhow, may become relatively more valuable. Expect both patterns to continue. "}],[{"start":327.6,"text":""}]],"url":"https://audio.ftcn.net.cn/album/a_1785507149_4773.mp3"}