AI is ushering in a new development trap for the Global South
A decade ago, French professor Thomas Piketty became the zeitgeist global economist with the debut of his magnum opus, Capital in the Twenty-First Century. It’s an exhaustive and exhausting book, one that posited a simple formula: r > g, namely, that inequality grows when the rate of return to capital (investments in stocks and other securities) exceeds economic growth more broadly (a partial proxy for returns to labor). Devilishly simple but magisterially argued (with perhaps too many allusions to the novels of Jane Austen), the book generated an entire cottage industry of criticism and remains a totemic work for different strands of politics.
Outside its iconic formula though, Piketty repeatedly points to the winner-takes-all nature of the present economy as one of the world’s greatest challenges. Top returning funds have better access to the best investments, securing their status for years to come. Top companies have better access to the top talent (for the simple reason that they can pay better), while the dynamism of top national economies like the United States attracts the best talent from around the world. Winner-takes-all plus time is a potent combination in the asymptote.
One of Piketty’s most striking points (and one that I still think about a decade later) is the importance of crises in resetting that concentration of wealth, status and capital. War, economic depressions, and massive technological changes can all mangle wealth and upend the ingrained Pareto distribution. If you believe in Steven Pinker’s thesis that the world is getting less violent, happier and safer over centuries, then the world is also getting more unequal as those wealth concentrations have time to accrue without chaos. For the Global South, both the crises and the concentration are disastrous.
So-called development traps have been discussed for decades, but what’s striking is how the traps keep evolving. A few weeks ago, I wrote in “brAIn drAIn” about how developing nations are struggling to find pathways to prosperity thanks to automation. But the story is much broader than just automation: the challenge is that the advantages of scale and incumbency — that winner-takes-all dynamic highlighted by Piketty — are challenging the conceit that smaller and more disruptive countries have a shot at rising up the developmental ladder through strategic economic competition.
In fact, it’s a pattern that’s increasingly prevalent these days in the startup world as well. No startup can defeat all of Google from search engines and Android to YouTube, just as no small emerging market has a hope of displacing the United States or China (Google, reaching more than $300 billion in revenue in 2023, outranks the GDP of all but 46 countries including Portugal, Peru and Greece). Nonetheless, the scale here is so gargantuan that there still remains small opportunities aplenty.
Kagi, a subscription search engine that offers endless features for power users, competes with Google by just narrowing its focus to customers willing to pay to customize their experience. Similarly, the island of Anguilla now makes more than 10% of its GDP by selling .ai domain names (it owns the two-letter country code, similar to Tuvalu owning .tv and Libya .ly). Yes, it’s a small niche, but Anguilla at roughly 16,000 citizens doesn’t need a big niche to succeed (it’s about a tenth the population of Williamsburg, Brooklyn).
Unfortunately, a patchwork of small niches doesn’t construct an advanced economy anymore than a startup can hope to cobble together a dozen products into a competitive business. To succeed at scale means finding large niches and conquering them. Obvious and increasingly impossible, not only because of the entrenched winner-takes-all dynamic in place, but because those winners are also never slowing down.
Take EVs. One of the big international debates recently has focused on China and whether it is overproducing electric vehicles (it’s a subject so interesting that our Riskgaming consultant Ian Curtiss is building a game on this very subject — stay tuned for that release in the weeks ahead). The Biden administration last month raised tariffs to 100% on Chinese-made EVs to counter the country’s low export prices, while Europe is leaning in the same direction even as it debates how to respond. The implications are massive: global auto sales measure in the trillions of dollars, a far cry from the paltry millions available from AI-inflected domain names.
Per Piketty, massive technological change is a rare crisis that can upend global wealth concentration. Indeed, that is what has been and is taking place. In lieu of GM and Ford, which have pulled back from EVs due to anemic domestic sales, China has scaling advantages that allow it to make these cars cheaper than anyone else. It’s already won, and now it’s taking it all.
Take one of its leading manufacturers BYD: the company just announced last week a hybrid powertrain that can reach distances of 2,000 kilometers (1,250 miles) and will be featured in two new sedans costing under $14,000. There’s practically nothing on offer from American, Japanese and European auto manufacturers that even comes close to that distance-to-price ratio.
BYD is hardly an overnight success story, as Kevin Xu chronicled on Interconnected earlier this year. Decades of battery research has given it a scaled advantage in EVs that other auto manufacturers struggle to compete with. Essentially, China took advantage of a massive gap in the market for more affordable and electric vehicles, and now dominates it much as it does the solar panel market.
Unlike traditional development paths though, China didn’t give up the lower-cost manufacturing while advancing to the cutting-edge. It increasingly dominates both, and there is no niche open for others to swoop in and seize. It’s nearly impossible to envision an emerging market competitor from India, Latin America or Africa competing with BYD in EVs anytime in the next two decades. Competition is all but over.
Electric vehicles are complex machines, but so are other electronics. The appliance revolution in the West after World War II would eventually become one of the most lucrative segments for the East Asian tigers, with companies like LG and Samsung eventually taking the lead in everything from dishwashers to televisions. Decades ago, these products had the advantage of being labor intensive to manufacture, while also not requiring extraordinarily advanced skills to build. That matched the labor forces available in these post-war, developing countries.
That’s just not the case anymore, where these same appliances are now produced in advanced factories with heavy automation designed to churn out tens of millions of machines a year at a cost efficiency that precludes any manufacturer that isn’t similarly scaled. Take Lux’s own startup Impulse Labs, which is upgrading the cooktop for the twenty-first century with better induction technology, software and design. This is the future of cooking, but it requires a manufacturing sophistication that only a handful of companies globally can do at scale with minimal defects and at a price that consumers are willing to pay. Growth for these appliance companies comes from expanding sales to the burgeoning middle classes of the Global South, which is one reason why South Korea just signed an expansive trade agreement with the United Arab Emirates this past week.
Switching from EVs and appliances to software only makes the story worse, of course. SaaS over the last decade offered an historically unusual financial alchemy: relatively low upfront costs (“lean startup”) coupled with a scalable and disruptive business model that offered an early path to compete against incumbents. Dozens of competitors duked it out to win against a smattering of incumbents.
AI is the complete opposite, and a return to the norms of tech innovation where a handful of competitors are well-funded to fight. To compete in AI today, investors need to pour massive dollars into founding rounds in order to offer engineering teams the salaries, compute resources and infrastructure necessary to reach the frontiers of science. We recently backed the founding round of Xaira Therapeutics, which raised $1 billion to build an AI platform for therapeutics discovery. A billion dollars for a brand-new company is extraordinary, but it’s the only path to building a company of enduring value in the space.
Or take our recent investment in Sakana AI. Sakana is turning into one of the sovereign AI champions for Japan, a country that’s heavily invested in securing a perch in the next generation of top tech companies. Japan has market leaders from semiconductor materials to consumer electronics, but it has failed to build a tech giant on par with Google or China’s Tencent. With top VC investment and heavy support from central government bureaus, the nation’s leaders hope that it can incubate a local champion with global reach.
Again and again from hardware to software, we see massive niches where the upfront costs preclude competition from all but the incumbent winners. What then becomes of the Global South? For countries looking for paths to prosperity, all roads seem to lead to thickets of failure.
Take India’s commitment to building up an indigenous chip manufacturing cluster. That’s a major goal of prime minister Narendra Modi, who just secured a third term as head of government despite a surprise setback in parliamentary elections this week. Tata Group and several other Indian conglomerates announced large commitments to a new series of chip fab clusters earlier this year, with a focus on legacy chips that are easier to manufacture and are diversely used across appliances and automative. It’s the traditional path to prosperity: start with the cheaper and less advanced end of the manufacturing spectrum and assiduously work up the ladder.
There’s just one problem: China is unwilling to cede market share in the legacy chip sector even as it has ambitions to manufacture leading-edge chips. In fact, not only is it not intending to give up — it’s doubling down. This year, China will produce more new capacity in legacy chip manufacturing than the rest of the world combined, forcing margins lower and throttling competing countries. India will struggle to even get its first foothold on the ladder, let alone climb up. China is already the winner here, and now it just keeps on winning.
The challenge for the Global South remains unresolved, which leads back to Piketty. Ten years after Capital in the Twenty-First Century, Piketty’s more recent work has focused on how to ameliorate the vast inequalities of the winner-takes-all economies that have come to dominate modern capitalism. Much of his proposals can be summarized as mechanisms of redistribution, like a more national version of universal basic income (UBI). It’s not an approach that resonates with me, but the fundamental challenge he originally identified remains the same. Network and scaling effects are closing down competition in more and more industries, and that’s limiting opportunities for billions of people globally to lift themselves out of poverty and into the middle class. How long will their patience last?
The Orthogonal Bet: Novelist Robin Sloan’s Love for Books with Maps on the First Page
In this episode of The Orthogonal Bet mini-series we’re doing on the Riskgaming podcast, our scientist-in-residence Sam Arbesman speaks with Robin Sloan, novelist and writer and all-around fun thinker. Robin is the author of the previous novels, Mr Penumbra’s Twenty Four Hour Book Store and Sourdough, which are both tech-infused novels, with a sort of literary flavor mingled with a touch of science fiction. That’s why Sam was so excited by Robin’s brand new third novel Moonbound, where he goes for broke and writes a sprawling science fiction tale set in the far future.
In this episode, Sam explores how Robin built this far future and how he thinks about world-building, an exercise regimen for your imagination, science fiction and fantasy more broadly, and of course, novels with maps. And Lord of the Rings obviously makes an appearance as well.
But Moonbound also touches on AI in some really thoughtful and thought-provoking ways, and Robin has also been an early experimenter and adopter of language models. They get into all of that too, talking about AI, the nature of creativity, storytelling, and so much more.
Lux Recommends
- Sam enjoyed Maggie Appleton’s lecture on the future of software in “Home-Cooked Software and Barefoot Developers.” She emphasizes that with AI helping more and more people develop software, we will see more opportunities for people to build smaller programs just for themselves and friends, loosening the constraints of Big Tech and ushering in an era of a more open and friendly software world. I hope she’s right.
- I enjoyed Burkhard Bilger’s essay in The New Yorker on “Piecing Together the Secrets of the Stasi,” which includes trying to reassemble millions of shredded documents using AI. “In the past five years, he said, MusterFabrik has used its scanner and a newly designed puzzler program to help reconstruct fragments of a Roman mural, documents from a Jewish community center in Buenos Aires which was destroyed by a bomb in 1994, and the papers of the polymath Gottfried Wilhelm Leibniz. But its most ambitious project, the one likeliest to serve as a model for reconstructing the Stasi files, is in Cologne.”
- I track many things, but the machinations of Russia’s central bank is not one of them. So I found this compact profile of governor Elvira Nabiullina to be fascinating, particularly in the context of Giuliano da Empoli’s novel The Wizard of the Kremlin which I recommended a few weeks ago.
- I enjoyed this piece by Dan Brooks in Defector on why Culture Needs More Jerks. “The importance of having antagonists to the experience of loving a particular artist can be seen in music fandoms that are legitimately misunderstood, such as Juggalos or people who listen to Phish. While different in their aesthetics, these two groups are both made stronger by the ridicule they get from people who do not—and I deploy these scare quotes as hard as I physically can, at the risk of finger cramping—‘get it.’”
- Finally, Silicon Valley is increasingly finding itself enmeshed in that swamp known as DC, and Politico has a good profile of what firms like Y Combinator are trying to accomplish. “But with about 70 percent of the companies [Garry Tan]’s firm supports being built on AI, including the new models powering tools like ChatGPT, the VC titan is also focused on promoting a laissez-faire AI ecosystem and pushing back on fears that the technology poses existential threats to humanity. It’s a stance that, as Washington weighs how or whether to regulate AI, pits him against some of the dominant tech firms advocating for licensing regimes and other safety-minded restrictions on its development.”
That’s it, folks. Have questions, comments, or ideas? This newsletter is sent from my email, so you can just click reply.
