Weekend reading links

1. Subsea cable is fast becoming a sector of strategic importance. Globally, there are four major companies - Japan's NEC, New Jersey-based SubCom, France's state-owned Alcatel Submarine Networks (ASN), and China's HMN Tech, a former Huawei subsidiary. The last three own cable-laying shipping fleets, while the first charters its vessels and is now set to receive government support to buy ships to put them on a par with their counterparts. A subsea cable-laying vessel could cost about $300 million. 

It currently relies on a subsea cable-laying vessel leased from a Norwegian group in 2022 on a four-year charter, partnerships with other companies and on renting the specialist ships on an ad hoc basis to meet surging demand for fibre-optic cabling in the Indo-Pacific. NEC dominates installations in Asia and has laid more than 400,000km of cables globally. It also specialises in armoured cables that can better withstand sabotage. Globally, there are 63 cable-laying ships, according to the International Cable Protection Committee. ASN owns seven, while SubCom and HMN Tech are believed to own seven and two, respectively, but did not reply to requests for confirmation. Japanese telecoms groups NTT and KDDI own cable-laying vessels, which are rented out to NEC, but they are not the larger kind of vessels required to lay transocean cables.

This is a description of the challenges being faced by the industry.

A shortage of ships is one of the major bottlenecks to achieving the 26 per cent rise per year predicted for global data transmission to 2031, driven by video streaming and artificial intelligence services led by Big Tech groups such as Meta and Google, according to the TeleGeography telecoms data service... Up to 200 cables are damaged annually, primarily due to fishing or anchors, but also through sabotage, as seen in the disruption caused to two cables in the Baltic Sea last year. Chartering for an accurate period has become increasingly difficult due to the unpredictable timelines for securing the cables’ passage. There is now a two-to-three-year wait to get permissions from countries whose waters the cables pass through, up from six months to a year about a decade ago.

2. Salaries in India have not kept pace, even with inflation in the FY16-24 period.

More here. 

From Diwali 2023 onwards, Indian companies’ earnings growth has decelerated at a rapid rate. Underpinning this deceleration is a sharp conk-off in consumption growth, long the mainstay of the Indian economy. Key drivers of this consumption downturn are a sharp deceleration in white collar job creation alongside a reduction in real wages for white collar workers over the past eight years.

And here

This should be a matter of deep concern.

The weak consumption demand will constrain India's sustained high growth ambitions. It will restrict private investment, limit job creation, squeeze salary growth, and increase household indebtedness. The only solution is broad-based, equitable economic growth. But that, in turn, requires conditions that are far from 

3. Rhodium Group has a report on the challenges faced by foreign car manufacturers in China. The German car makers and Tesla have the largest China exposure, while the Japanese and Korean car makers have far smaller engagement. 

It is important to note that while Tesla is likely even more dependent on China for profits than German OEMs, the sources of those profits are fundamentally different. German automakers still earn most of their China income—though at shrinking margins—by selling vehicles to Chinese consumers, whether locally produced or imported. Tesla, by contrast, likely earns the bulk of its China profits from two sources: the sale of regulatory credits—its CFO disclosed that three-quarters of Tesla’s global credit sales in 2024 (Q1–Q3) occurred in China—and a highly profitable export business built on China’s low production costs. In short, German OEMs depend on Chinese consumers, while Tesla depends on Chinese workers and credits... The experience of Toyota, Hyundai, and Kia shows, however, that success in China is not a prerequisite for global competitiveness. Toyota has become the world’s largest carmaker despite a relatively modest China footprint. Hyundai and Kia, for their part, have grown despite running loss-making China operations after the 2017 THAAD dispute triggered consumer boycotts that cratered sales.

4. The US Congress has become ideologically polarised since the turn of the millennium.

And immigration has been at the forefront of the polarisation.

5. A summary of the destructive chainsaw Trump has taken to the US Science establishment.

The proposed 2026 Budget outlines cuts of 34 per cent for basic research and 22 per cent for all research. It demands a 55.7 per cent cut for the National Science Foundation (NSF) — from $8.8 billion to $3.9 billion — and a 39.3 per cent cut for the National Institutes of Health (NIH) — from $46 billion to $27.9 billion. These two agencies are the primary sources for TRL 1-2 basic research. The story for TRL 3-6 applied research is also rough. The Department of Energy’s (DoE’s) Office of Science faces a 14 per cent cut, and Nasa’s science-research budget is slated for a 46.6 per cent reduction — from $7.3 billion to $3.9 billion... 

The Nazi government, which took charge in 1933, was a populist one, harnessing mass anger against the cosmopolitan elites. An estimated 25 per cent of all physicists in Germany, including 11 past or future Nobel laureates like Albert Einstein, Max Born, and Leo Szilard, fled Germany. The best research organisations of the world — like the University of Gottingen — were destroyed. It only took one year. In 1934, the great mathematician David Hilbert said to the Nazi education minister “mathematics in Gottingen? There is really no such thing any more”.

6. Comparing dotcom era telecom investments with AI investments today.

There were more than 80mn miles of fibre optic cable laid from the mid-1990s until the end of the dotcom boom, and much of that investment took years to pay off. US telecom companies spent $444bn in capital expenditure between 1996 and 2001. Still, compare this with the $342bn that will be spent this year alone in the US by the top investors in AI data centres and computing infrastructure, including Microsoft, Alphabet, Amazon and Meta. At the current rate of power consumption needed to fuel AI development, estimated investment will stretch to nearly $7tn by 2030.

7. Nvidia makes a $1 bn investment in Nokia to take a 2.9% stake in the Finnish telecom manufacturer, following which the shares in Nokia surged 21% and Nvidia by 5%. Nokia is seeking to diversify away from network infrastructure into AI and cloud services. 

This is part of the vendor-financing model of the emerging AI market. Is Ericsson next for Nvidia?

8. Two graphics that question the argument that we are in a bubble. One, the PE multiples of the Big Tech firms compared to those in earlier bubbles. 

The capex boom is largely (at least till now) being funded by free cash flows, unlike debt in earlier booms.

9. It's useful to keep in mind that much of the infrastructure in the West was built long ago.

In Britain we are using rail lines, bridges and sewers built by the Victorians, and even the Romans’ roads. It has taken over 150 years for London to need an expansion of the sewerage system begun by Joseph Bazalgette in 1859 (and largely finished within a decade). The US built its railroads in the 19th century, and interstate highways from the mid-1950s.

10. The reforms currently underway in the US to reverse safeguards in the banking sector in the name of deregulation may be an instance of bad deregulation. 

Last week, the Federal Reserve announced plans to overhaul its annual banking stress tests to make them less onerous. US banking watchdogs are widely expected to follow up with other changes to capital and leverage rules that could unlock $2.6tn in additional lending capacity, according to consultants Alvarez & Marsal. To backers, there is a logic to the Trump administration’s moves. Shackling banks with high capital requirements has not eliminated risky lending. Instead it has led to regulatory arbitrage that makes the danger harder to supervise, they say. As the IMF pointed out, banks now lend to private capital, which uses the funds to leverage investor money while making loans and buying securitised debt. In theory, the investors absorb the first losses, keeping bank deposits safe. In reality, layers of borrowing by companies like First Brands make it hard to tell who is on the hook and may lead to complacency. If banks lent directly, they say, they would do more due diligence and pick their borrowers more carefully... 

Another Trump administration initiative to allow ordinary investors to put their money in alternative assets, which have long been restricted to institutions and the super wealthy. Those changes are expected to channel floods of retail and retirement money to private capital groups, giving them bigger pots with which to make loans and buy asset-backed securities. These retail funds will be under particular pressure to deploy capital quickly because of the way they are structured... Bank of England governor Andrew Bailey said last week that “alarm bells” are going off around the rapid growth of structured products, and JPMorgan chief executive Jamie Dimon has proclaimed that the recent collapses of subprime auto lender Tricolor and car-parts maker First Brands are evidence of “cockroaches” in the credit market. The failures have uncovered complex webs of borrowing and allegations of fraud, leading Apollo chief Marc Rowan to warn that eroding lending standards are leading to “late-cycle accidents”.

11. Stock market concentration in the US is at all time high.

Eight of the 10 biggest stocks in the S&P 500 are tech stocks. Those eight companies account for 36 per cent of the entire US market’s value, 60 per cent of the gains in the index since the market bottomed in April and almost 80 per cent of the S&P 500’s net income growth in the last year... MSCI All World index, which comprises over 2,000 companies from more than 40 markets, currently has almost a quarter of its capitalisation in just eight US tech groups.

 On Tuesday afternoon, when US Stocks hit their latest highs, 397 stocks in the S&P 500 lost ground. In 35 years, the index never posted a gain on a day when do many of its components sold off. 

Since the launch of ChatGPT in November 2022, the US markets have been on a tear, underpinned by AI stocks.

This is an interesting snippet.

Since 1970, the total value of all publicly traded US stocks has averaged about 85 per cent of US GDP. Warren Buffett once described this as “probably the single best measure of where valuations stand at any given moment”. On Tuesday, the metric rose to a record 225 per cent.

12. More on the First Brands bankruptcy case in the US in this story of how a small Draper, Utah-based equipment finance specialist, Onset Financial, ended up with a $1.9 bn loan exposure to the Ohio-based automotive parts maker which borrowed close to $12 bn to finance acquisitions. 

First Brands’ reliance on Onset, which claims to eschew the “rigid” and “strict” approach of banks in favour of “speed” and “flexibility”, illustrates how unconventional corners of credit markets facilitated its borrowing binge, with many of the Ohio-based company’s lenders unaware of the true scale of its debts until it was too late... Onset’s corporate identity to date has been marked by promotional videos, a high-tempo sales culture and links to both prominent local investment firms and sports players. The fallout could ripple through the community in Utah, where Onset has built up an image of growth and glamour. Founded in the depths of the 2008 financial crisis in Draper, a small city 20 miles south of Utah’s state capital Salt Lake City, Onset’s triumph over stodgy rivals in a key area of business lending is a recurring theme of its corporate lore. Equipment leasing is a $1.3tn industry in the US that allows companies to rent machinery rather than sink large amounts of upfront capital into building or improving their facilities... People familiar with Onset’s operations describe a model fuelled by a direct and ambitious sales force. “What does our product do? We sell money, simple as that,” Taylor Weeks, Onset’s vice-president of sales, told a podcast in 2023. Weeks added that the company provides “rocket fuel” for “high-growth” businesses.

Equipment leasing is a $1.3 trillion industry!

13. FT writes about the rise of China's biotech firms, licensing technology and selling drugs outside the country. From having no biotech sector to speak of ten years back, in the first eight months of 2025, there have been 93 overseas licensing deals worth a total of $85 bn on drugs developed in China. 

China’s transformation from a copycat manufacturer of drugs developed overseas to a hub of homegrown research is exemplified by Jiangsu Hengrui. Founded in 1970, it spent the first two decades as a small-scale state-owned manufacturer of low-cost antiseptics. In the 1990s, it started developing generic anticancer drugs. It was privatised in 1997 and began investing in building its own research capabilities. Today, it has one of the most diversified pipelines in the country, spanning weight-loss therapies, oncology drugs and Alzheimer’s treatments...

Hengrui has struck licensing agreements with Merck, Braveheart Bio and Glenmark in the past year alone. In July, it agreed a deal with UK pharmaceutical company GSK to develop up to 12 medicines. “In China, you can scale and test medicines in human beings much faster than in the US or Europe,” said Loncar. “If you have an idea for a drug, you can get an answer about whether it works a year or two earlier in China.” For many Chinese biotechs, the surge in international partnerships has provided much-needed capital after a difficult few years marked by drug pricing reforms that squeezed profit margins on domestic sales... Hengrui’s international deals have highlighted a concern for Chinese pharma companies seeking to get international approval for new drugs. The US Food and Drug Administration has repeatedly rejected one of Hengrui’s cancer drugs, citing questions about quality control at its manufacturing sites.

US and European pharma companies are doing what their manufacturing counterparts elsewhere did by outsourcing to China, only to realise that they have been outmuscled by them over time. 

This also raises questions about where India's established pharmaceutical firms are. 

14. Finally, NYT has a good article on OpenAI's circular financing deals. 

Many of the deals OpenAI has struck — with chipmakers, cloud computing companies and others — are strangely circular. OpenAI receives billions from tech companies before sending those billions back to the same companies to pay for computing power and other services.

Narratives trump theory

It is a reality of life that narratives that are grounded in stories trump sophisticated theories grounded in logic and reason. 

The booming hype cycle on AI is only the latest example. References to AI and ML have become de rigueur in any sales pitch about innovative solutions, regardless of the context. Everything from food delivery to manufacturing in the private sector is being claimed to be dramatically improved with some underlying AI engine. Notwithstanding the lack of any meaningful commercial success, the AI bubble continues to inflate at a rapid pace. 

As an illustration, over just the last 12 months, the top ten AI startups, all loss-making, have attracted $161 billion in VC capital (two-thirds of all US VC spend) and gained close to $1 trillion in valuation. 

The AI mania is not confined to areas of high technology and finance. Even within the more prosaic environments of public systems, it has become a norm to fit AI/ML into any new public policy idea or program or project for virtue signalling. Never mind its relevance and value, proponents put forth claims of using an AI/ML layer to embellish their ideas. Even simple data analytics solutions that are basically data description, without even basic analysis, are presented as having a layer of AI/ML. 

FT’s Gillian Tett points to the practice of “cargo cults” used to describe the phenomenon observed among the native inhabitants of the Melanesian islands that were invaded by Westerners in the 19th century and flooded with previously unseen consumer goods. Dimitris Xygalatas writes

When Indigenous communities throughout the area had their first encounters with colonial forces, they marveled at the material abundance the foreigners brought with them. During World War II, when many Melanesians worked for U.S. and Australian military forces, they observed soldiers who never seemed to engage in any productive activities, such as fishing, hunting, working the land, or crafting anything. All they did was march up and down, raise flags, chant anthems, and signal toward the sky. And when they did that, metal birds appeared and dropped all kinds of goods for them. The Indigenous observers concluded that the strange rituals were causing the cargo to arrive.

With the end of the war, the military bases were abandoned and the goods ceased to arrive. To get the cargo to return, local chiefs began organizing ceremonies that mimicked the rituals of the troops. Soon, elaborate myths and theologies developed around those rituals. Surely, the cargo must have been a gift from the gods—their own ancestors. After all, who else could be capable of producing such wealth? The foreigners had merely discovered the rituals that unlocked these treasures…

But the only airplane present is a full-size wooden replica of a light aircraft. On one side of the strip lies a control tower made of bamboo. On the other sits a satellite dish built of mud and straw. Undeterred by the apparent lack of any actual aviation technology, some of the men light torches and place them alongside the runway. Others use flags to wave landing signals. Everyone raises their gaze to the sky in anticipation.

Tett extends the cargo-cult phenomenon to the current AI mania.

Physicist Richard Feynman borrowed this metaphor to decry “cargo cult science”, cases where researchers “follow all the apparent precepts and forms of scientific investigation, but they’re missing something essential, because the planes don’t land”. The same analogy now applies to AI. Almost every business executive today is eager to tell investors about their AI strategy (even though 95 per cent of companies have not (yet) seen revenue gains) and every VC group is keen to show AI plays. Similarly every Big Tech executive is investing in massive data centres, even though Bain reckons some $2tn of revenue will be needed to fund this by 2030. And charismatic figures like Sam Altman, CEO of OpenAI, keep promising fresh magic. Or as Stephan Eberle, a software engineer, laments: “Watching the industry’s behaviour around AI, I can’t shake this feeling that we’re all building bamboo aeroplanes [like cargo cults] and expecting them to fly.”

In the case of investors, the cargo-cult phenomenon works through fear of missing out (FOMO).

The iconic example of our times of the narrative transcending all logic is how Tesla’s equity market valuation has become tied to the Elon Musk phenomenon. In substantive terms, Tesla has been falling behind in all its major markets and may now be technologically behind its Chinese competitor, BYD. The latter has a superior battery technology, is vertically integrated, and has not only caught up on automatic driver assistance systems (ADAS) but may even have pulled ahead. 

With more than 95% of its global deliveries coming from Model 3 and Model Y, and that too for nearly a decade, Tesla is now a two-trick pony. In contrast, BYD has a dozen models globally and is releasing new models each year. Tesla’s growth has been primarily driven by lowering the prices of its existing models, hoping to offset margin declines with volumes. Its gross margin, excluding regulatory credits, has declined sharply from nearly 30% in the fourth quarter of 2021 to around 17% in the second quarter of 2025. 

But in an inversion of all logic, this decline has been accompanied by an increase in its market valuation to $1.4 trillion, more than ten times that of BYD. Such valuations are built on the premises of high margins, and runaway hits like robotaxis and AI-powered robots. These premises are, in turn, built on the narrative of the cult of Elon Musk and the miraculous powers endowed on him. Tesla is one mega-giant bet on Musk, perhaps the biggest financial market bet on one individual in history, by some distance. 

In each of these cases, once the irrationality has taken hold thanks to the narratives, it tends to find rational explanations. A commonly cited one is that such bubbles may have become the only way to mobilise resources at the scale required to push the technology frontiers. Sample this.

“There will be casualties. Just like there always will be, just like there always is in the tech industry,” said Marc Benioff, co-founder and chief executive of Salesforce, which has invested heavily in AI. He estimates $1tn of investment on AI might be wasted, but that the technology will ultimately yield 10 times that in new value. “The only way we know how to build great technology is to throw as much against the wall as possible, see what sticks, and then focus on the winners,” he added.

This explanation also syncs with the dominant VC model of financial intermediation and allows them, in turn, to raise the massive amounts of capital required to fund the bubble. 

In the case of the AI bubble, there’s also a powerful strategic imperative. As Gillian Tett has pointed out, given the threat to America’s technological superiority posed by China’s state capitalism, such bubbles may well be “the only way American capitalism can ever amass the scale of investment needed to create this type of ambitious infrastructure”.

While it may sound heretical, the AI bubble also highlights the unique nature of American capitalism, which has shown an unmatched appetite to assume excessive risk in the expectation of windfall returns. It is only the latest, albeit far bigger, in the line of irrational exuberance and risk assumption that has distinguished the US economy even in the last five years - WeWork, GameStop, NFTs, cryptocurrency assets, SPACs, etc. As Andrew Ross Sorkin has pointed out, “there is no innovation without speculation” and “speculation built America”. So he writes, 

“Speculation isn’t a bug in America’s economic code, but a crucial component part of the engine… Speculation is often caricatured as gambling. But at its core, it is belief plus risk. It is the act of investing capital in a highly uncertain outcome, hoping for reward.”

In Tesla’s case, too, the irrationality gets justified in terms of Musk’s superhuman talent. This is nicely captured in Tesla’s battles with courts and shareholders to get approval for Musk’s astronomical $1 trillion pay package. 

Tesla management has sold it in terms of binding Musk to remain sufficiently committed to the company, amidst his other multiple business interests. In fact, Board Chair, Robyn Denholm, has justified it, calling Musk a generational talent who would have to expend “time, energy, and effort beyond what most humans can do.” She said, ‘There’s just not anybody, either inside or outside the organisation, that is Elon today.” In what is effectively a blackmail/bluff, Musk himself has said he’ll leave Tesla if he does not get the pay package and gain greater control over the company to protect it from hostile takeovers that can detract from its efforts to develop AI technology and humanoid robots. 

The Musk compensation issue would be unimaginable in any other country. In the US, as Denholm suggests, astronomical compensation packages have become part of an entrenched narrative that those CEOs deserve these amounts. There’s no logic, both in terms of substance (the expertise brought in by the CEO) or market demand (the scarcity of such executives), that can justify even remotely close to these amounts. Numerous studies have consistently shown no correlation between executive compensation and shareholder returns. 

Instead, the phenomenon of such excessive CEO pay is fuelled by narratives (and the market structures and incentives) that have become part of the US corporate culture. Narratives shape cultures. 

In this context, it is important to remember that the central role of narratives in shaping the biggest mainstream economic trends is a big gap in economic thinking. These narratives, which stand in complete opposition to orthodoxy and logic, must be an essential component of any college or university economics curriculum. 

To some extent, the mainstream economists have grudgingly accommodated parts of it in the guise of behavioural economics and finance. In this reading, while rational economic agents continue to dominate the economic decision-making, human cognitive failures and idiosyncrasies result in some occasional deviations. 

Given how pervasive these deviations are in the real world, this reading must be revised to provide a more central role for narratives that deviate sharply from logic and orthodoxy. Economic decisions, both in corporations and by governments, are also cultural and political choices, and these preferences often dominate. While those choices are grounded in logic and orthodoxy, other considerations also inform them. These considerations are shaped by the specific narratives surrounding them. 

Interestingly, many economic orthodoxies themselves have become narratives sans any empirical basis. I have blogged here about 25 such orthodoxies that dominate the discourse without any empirical basis. 

Process knowledge to build manufacturing capabilities

China may be the best modern illustration of Joel Mokyr’s argument that growth and development happen when propositional knowledge is translated into prescriptive knowledge through dense networks. Without explicitly articulating using this framework, the narratives described in Patrick McGee’s Apple in China (see this and this) and Dan Wang’s Breakneck demonstrate the essence of Mokyr’s argument.

The fundamental point is that both Apple (through iPhone) and Western multinational corporations in general helped China by transferring useful knowledge (what Wang calls ‘process knowledge’) and creating rich industrial ecosystems seeped with technical expertise acquired iteratively over several years. It helped that China, like the pre-Industrial Revolution England, had a culture that gave primacy to the acquisition of useful knowledge over theoretical propositional knowledge. 

I just completed Dan Wang’s book, Breakneck. The big premise of the book, the reason why the US is lagging behind China in making things is because the former is ruled by lawyers and the latter by engineers, is too simplistic and questionable. Jonathon Sine has an excellent critique. 

However, the book has a compelling explanation for how and why China built up its manufacturing prowess. It echoes the description by McGee of how Apple’s intense and long-drawn engagement in the co-creation of Foxconn’s iPhone manufacturing capabilities played a critical role in the development of China’s world-beating manufacturing ecosystem of designers, tooling engineers, component suppliers, managers, and assembly-line workers. This is a good description of the emergence of the ecosystem.

The smartphone components were getting better every year, part of a trend that Chris Anderson, former editor of Wired, called “the peace dividend of the smartphone wars.” The hundreds of billions of dollars invested in the smartphone supply chain have caused the cost of electronic components – cameras, sensors, batteries, modems – to plummet. That’s why we are able to carry around sensors in our pockets that used to be available to only a select few military powers.

Many companies have grown around this peace dividend. Indeed, Shenzhen is the headquarters of many of China’s most dynamic companies, including BYD, the world’s largest EV maker; DJL, the world’s largest consumer drone maker; and Huawei, the beleaguered company that is the world’s largest telecommunications equipment maker. Electric vehicles are full of electronic components borrowed from smartphones; the consumer drone is roughly a reassembly of a smartphone camera and sensor with propellers for flight. The magic of Shenzhen in the combination of the world’s most creative hardware engineers sitting in a sea of components that improve every year amid a labour force of millions who know how to put together electronics. This buzzing ecosystem has produced many other products that follow in Apple’s wake, like hoverboards, electric scooters, virtual reality headsets, and who knows what’s next?

Wang highlights the importance of three aspects of technology in production, specifically the last aspect of process knowledge. 

First, technology means tools. These are the pots, pans, knives, and ovens required to prepare a dish. Second, technology means explicit instruction. These are the recipes, the blueprints, the patents that can be written down. Third and most important, technology is process knowledge. That is the proficiency gained from practical experience, which isn’t easily communicated. Ask someone who has never cooked before to do something as simple as fry an egg. Give him a beautiful kitchen and the most exquisitely detailed recipe, and he might still make a mess…

Process knowledge is hard to measure because it exists mostly in people’s heads and the pattern of their relationships to other technical workers. We tend to refer to these intangibles as know-how, institutional memory, or tacit knowledge. They are embodied by an experienced workforce like Shenzhen’s. There, someone might work at an iPhone plant one year, for a rival phone maker the next, and then start a drone company. If an engineer in Shenzhen has an idea for a new product, it’s easy to tap into an eager network of investors. Shenzhen is a community of engineering practice where factory owners, skilled engineers, entrepreneurs, investors, and researchers mix with the world’s most experienced workforce at producing high-end electronics.

Silicon Valley used to be like this too, but now it lacks a critical link in the chain – the manufacturing workforce. The value of these communities of engineering practice is greater than any single company or engineer. Rather, they have to be understood as ecosystems of technology. The American imagination has been too focused on the creation of tooling and blueprints. Andy Grove, the legendary former CEO of Intel, said it best in 2010: that the US needs to focus less on “the mythical moment of creation” and more on the “scaling up” of products. Grove saw Silicon Valley transition from doing both invention and production to specialising only in the former. And he understood quite well that technology ecosystems would rust if the research and development no longer had a learning loop from the production process…

American manufacturers spent the better part of the last three decades unwinding its stock of process knowledge when it opened so many factories in China. Every US factory closure represents a likely permanent loss of production skill and knowledge. Line workers, machinists, and product designers are thrown out of work; then their suppliers and technical advisers struggle as well. Entire American communities of engineering practice have dissolved, leaving behind a region known as the Rust Belt. Some mayors and governors tried to step this receding tide. But they were continuously scorned by economists and executives, who sought low-wage production in the name of globalisation. Still today, many American economists doubt there is anything special about manufacturing and put their faith in the inevitable march to a service economy…

It became part of the elite consensus that the US could lose manufacturing. This consensus portrayed union bosses, as well as the handful of heterodox economists, as sentimentalists for resisting offshoring. Neither the Clinton nor the George W Bush administration restrained American firms from moving manufacturing operations to China. Now, it’s more obvious that the departure of manufacturing has created economic and political ruination for the US. We are still only beginning to understand how much it set the country back technologically…

If we think about technology ecosystems as communities of engineering practice, it makes sense that factory closures accelerated as process knowledge dissolved, prompting production problems and more job losses. And it also makes sense that Chinese workers went from merely assembling iPhones to producing some of their most valuable components as well. As one country lost its process knowledge, the other gained whole industries.

He describes how Tesla contributed to the development of China’s emerging dominance of EV car manufacturing.

Beijing did something unprecedented for Tesla in 2018: it allowed the company to fully own its plant in Shanghai. Previously, any automaker that wanted to produce in China had to partner with a domestic company. So Japanese, German, and American companies dutifully partnered with state-owned enterprises in order to access the enormous market. The state had hoped that these domestic companies would learn from the likes of Toyota and Mercedes-Benz and match their quality. In reality, Chinese automakers were sluggish from their research dependence on their foreign friends.

Tesla’s presence jolted China’s EV market. China’s business community began using the term “catfishing” for what Tesla was doing in China. The idea was that introducing a powerful new creature into the domestic environment would make Chinese firms swim faster. That’s exactly what they did to raise their game. When Tesla vehicles started rolling out of the Shanghai Gigafactory in 2019, BYD saw its sales decline by 11%, while profits fell by 42%. But Tesla would eventually do the whole market a favour. As in the US, the company’s audacious branding stimulated consumers to think of EVs as more than high-powered golf carts. And Tesla made investments in China’s tooling ecosystem that other automakers exploited to produce better cars. BYD benefited as well, reporting record profits in 2023 and becoming the world’s largest EV maker. And even the Communist Party’s main newspaper praised how Tesla produced the “catfish effect” for Chinese firms.

As Grace Wang, founder of Shenzhen-based Luxshare poetically expressed, “Flying with phoenixes will nurture outstanding birds.” It is another lesson that capitalist Shenzhen has taught the Communist Party: Market competition tends to lower prices and raise quality. 

Apple and Tesla have made a huge effort to train its Chinese workers to manufacture their products – and earned fabulous sums of money by doing so. These stories are replicated in varying degrees across China’s other communities of engineering practice, production hubs for shoes and garments in the eastern city of Wenzhou, medical equipment in Wuxi and Suzhou, and, most wonderfully of all, guitars in the mountains of Guizhou’s Zheng’an County. Overall, China’s manufacturing workforce employs more than a hundred million people, around eight times that of the US (13 million in 2025)… 

American companies have spent two decades building communities of engineering practice in China, made up of people who roll up their sleeves to figure out how to overcome their daily bottlenecks.

At the heart of this narrative about economic growth is the importance of persistent and boring implementation, over the Aha! moment of inspiration presented by ideas and inventions. This is a good summary:

Americans expect innovations from scientists working at NASA, in universities, or in research labs. They celebrate the moment of invention: the first solar cell, the first personal computer, first in flight. In China, on the other hand, tech innovation emerges from the factory floor, when a new product is scaled up into mass production. At the heart of China’s ascendancy in advanced technology is its spectacular capacity for learning by doing and consistently improving things… The US likes to celebrate the light-bulb moment of genius innovators. But there is, I submit, more glory in having big firms making a product rather than a science lab claiming its invention… Every day, millions of workers in factories to build up technological process knowledge. That is the basis of China’s tech power. China has become a tech superpower by exalting process knowledge and the communities of engineering practice that keep it alive.

There are a few observations that are of relevance for India as it charts its journey of building a globally competitive manufacturing base. 

1. Mokyr points out that the Industrial Revolution happened in England and not continental Europe, and attributes this to its culture of tinkering and refining, and interest in material progress. This required useful knowledge that the fabricants could use, as against the more theoretical and esoteric knowledge that savants created. 

Similarly, China had a culture of pursuing process knowledge. When faced with competition from Apple and Tesla, instead of folding up and retreating, these companies leveraged the ecosystems of process knowledge and engineering practice created by Apple and Tesla. They fought back to build their own products and compete (and outcompete) the foreigners. Huawei, Oppo, Xiaomi, Vivo, BYD, Geely, SAIC, Nio, Xpeng, and so on are the outcomes of this. 

Do Indian manufacturers and startups have the culture to embrace process knowledge, iterate diligently, build ecosystems of engineering practice, and move up the value chain? What are examples from India of such technology or industrial ecosystems that stand at the cutting-edge of their sectors?

2. The conventional wisdom on industrial progress, especially in technology-intensive sectors, is that of academia and industry working together to co-create products by transferring research into development at an industrial scale. While this is the approach that the US and other Western economies followed, China appears to have taken a different direction. Instead of spending time on the generation of propositional knowledge in its research institutions, it borrowed (or copied) this knowledge from wherever available and focused on the prescriptive knowledge required to scale up manufacturing. This is the classic industrial tinkering model.

Chinese firms rely on academia more in terms of getting skilled manpower to work in their factories. With the government’s guidance, Chinese colleges and universities have been running courses on niche areas like battery chemistry and rare earth processing for several years. They have kept flowing a continuous supply of high quality skilled workforce. 

This is of particular relevance for India, where academia-industry collaboration is limited, but also where industrial R&D is abysmally low (0.7% of GDP compared to China’s above 2.5% of GDP). Indian firms must necessarily multiply their R&D expenses manifold if they are to create the vibrant and high-productivity industrial ecosystems like those in China. They must invest in nurturing a culture of engineering excellence that seeks to constantly innovate and aspire to the global frontier of quality and technology. 

3. Another feature of Mokyr’s arguments and China’s success is the importance of clusters in producing things. A manufacturing base does not emerge in isolation from the production of individual firms. It requires the creation of ecosystems of engineering practice, where skills and expertise are transferred through learning by doing, and knowledge and technology spillovers. Clusters enable the diffusion and spread of process knowledge among businesses and their suppliers, and among competing firms. 

Massive industrial clusters have been central to the development of China’s manufacturing prowess. China has more than 500 towns that specialise in specific products for the global market, with some being responsible for 63% of world’s shoes, 70% of its spectacles, and 90% of its energy-saving lamps. Sample this from Wang,

Every year, as new models emerge, Apple needs new components or processes that a new design requires, like a certain type of adhesive or a screw of a slightly different size. Therefore, Apple constantly had to scramble to find suppliers on short notice. “Almost always, “the engineer continued, “we found someone in Shenzhen by asking a guy who knows a guy whose cousin might be able to produce a few hundred thousand new screws.” Virtually everything one needs to produce any electronic product can be found in a short drive around Shenzhen. Proximity creates efficiency. When it’s time to do stuff, a company can collapse coordination that usually takes weeks into a business meeting lasting hours by convening all the relevant suppliers in one room the next morning. And if something goes wrong, there are a lot of friendly neighbouring factories to call.

Unfortunately, India’s efforts in this direction, primarily in the form of SEZs, have been constrained by the size of these clusters. I blogged here on how the lack of clusters has been a constraint to the emergence of scale manufacturing in India. 

4. The creation of globally competitive ecosystems of engineering practice that value process knowledge for continuous improvement invariably sets in motion something like the Red Queen Effect. Companies must adapt and evolve continuously to even maintain the relative status quo in the highly competitive and dynamic global marketplace. 

Only firms exposed to global competition have the incentive to pursue this strategy. The incentives of firms in a closed market, or those happy with serving a large segment of their domestic market, will be only to minimise costs. They are likely to discount quality and technology. In simple terms, Indian firms must necessarily Make in India for the World.

There’s the real risk that Indian firms will remain entrapped in the bad equilibrium of making less than competitive products for India’s large price-sensitive consumer base. 

5. The example of how Chinese companies responded to the competition from Tesla is a testament to the aspirations of Chinese companies. In a short period of time, BYD and Co. were able to turn the tables on Tesla and emerge as undisputed global leaders in EVs. They first copied the engineering excellence of Tesla and continuously improved on it to then gradually surpass it. Underpinning this was the communities of engineering practice created by Tesla. 

In contrast, as I have blogged on numerous occasions, Indian companies, across sectors, have struggled to innovate and become globally competitive. 

The Ken has an article about how the garment makers of Tirrupur, who contribute over half of all India’s knitwear exports, failed to innovate and are now fighting to survive the loss of the US market. They remained stuck with the commoditised and lower value-added bulk supplies of cotton garments to the large US retail clients, with their low margins and large volumes, but stable and assured demand. They missed the opportunity to shift to man-made fabrics earlier and to blended fabrics in recent years, and did not venture into the higher value-added and higher margin mid-sized European who are much more demanding on fashion trends and quality. 

It will be extremely challenging for Indian companies to match the resolve and ambition shown by the Chinese EV manufacturers, nurture communities of engineering excellence, and become globally competitive. 

In conclusion, stripped of all the jargon and in simple terms, Mokyr’s work and the success of China is a big shout-out to the strategy of progress through continuous problem-solving and iterative adaptation, one which has relevance far beyond mere building of manufacturing capabilities to many things in life in general.

Weekend reading links

1. Lawrence Freedman has a good history of recent years of the Middle East.  

2. The US equity markets are a seven-trick pony.

See also this.

This is a good graphic of circular deals in the US AI ecosystem.

3. Greek PM Kyriakos Mitsotakis urges caution on the green transition.

The green transition cannot be an end in itself. For many years, Europe elevated decarbonisation above everything else. Other goals — employment, industrial production, strategic autonomy — these lost when they went up against decarbonisation. We cannot afford to stay on this path. Decarbonisation is vital but it is not the only objective. If we must accept some emissions for a bit longer to save our industries or to maintain social cohesion, so be it. We must have these debates honestly. We cannot begin with climate neutrality and hope everything else falls into place.

4. Friedrich Merz and Germany facts of the day

According to pollster Insa, two-thirds of Germans now say they are dissatisfied with the ruling coalition — a 20 point increase from June. Merz, never especially popular, is bearing the brunt: he has slipped to 18th place in Bild’s ranking of preferred politicians, trailing six of his coalition ministers and AfD co-leader Alice Weidel... Core German industries are shrinking, he laments. German steel output declined 12 per cent in the first half of this year compared with last year. And car manufacturing plants, which produced nearly 6mn vehicles in 2017, three-quarters of which were for export, now produce 4mn.

Germans don't seem to like Gerhard Schroeder and his acclaimed (outside the country) Harz reforms.

Schröder’s reputation has been tarnished by his Kremlin ties; he joined the board of Russian state-owned oil group Rosneft and lobbied for the building of Nord Stream’s second gas pipeline between Russia and Germany. Within the SPD, his so-called Hartz labour reforms, which spawned low-paid minijobs and triggered high-profile party defections that helped to create the leftist Die Linke party, remain a toxic legacy. “In DC at an IMF meeting, in Paris at the OECD or in Singapore you mention the ‘Hartz’ reforms, everyone applauds,” says a former SPD government official. “At any local SPD convention, the room temperature drops five degrees.” The SPD — which in February recorded its worst election result since the late 19th century, with a 16 per cent share of the vote — has spent the past two decades seeking to roll them back. These efforts led to a minimum wage in 2015 and culminated with the Bürgergeld, the latest version of a tax-funded basic income for the jobless or underemployed, introduced in 2023.

5, Tesla's third quarter profits fell by a quarter and operating margin dived from 10.8% to 5.8%. Similar trends in income from regulatory credits,

Income from regulatory credits trading plunged 44 per cent to $417mn in the quarter after the US government reduced fines for non-compliance on car emissions standards to zero, in effect killing the trading schemes. Tesla made $2.8bn in profit from trading programmes last year, with about three-quarters of that coming from the US.

6. Janan Ganesh is perhaps the most brilliant FT columnist. Highlighting the gushing and alarmist commentary about AI, he makes the point that history is made through politics and not technology.

Well, Earthlings, here are some entirely realistic scenarios in politics over a shorter time range. France elects a hard-right president who, without leaving the EU, impedes it from within until it ceases to function. Russia does something to a Nato member state that appears to constitute an “armed attack” under Articles 5 and 6 of the Treaty. The Sahel — where more than half politiof all terrorism-related deaths on Earth now take place, compared to hardly any in 2007 — becomes a base for attacks on the west. (Think turn-of-the-millennium Afghanistan, but much closer to Europe and America.) Britain or France or both suffer a bond crisis that triggers, at best, a necessary change in economic policy, and at worst civic unrest. These subjects don’t go undiscussed, of course, but the amount of oxygen they receive compared to tech talk is out of line. 

As a rule of thumb, be sceptical of any “futurologist” who doesn’t major overwhelmingly on politics. The grandest visions for AI — huge consumer surplus, huge job losses too — could happen. The departures of Xi Jinping (72) and Vladimir Putin (73) are going to happen, with implications for multitudes even outside their own two countries. The spread of possible outcomes is wide: from a détente between the west and the Eurasian autocracies under a new generation of leaders to an even higher pitch of conflict that is harrowing to think about. So, by all means, speculate about the potential of tech. But understand that one or two completely plausible political developments would drown out any effect that tech is likely to have on daily life. Even a modest trend, such as Europe’s projected increase in defence spending, has implications for taxes and therefore private consumption that it would take a major innovation to equal or cancel out. 

Consider the very recent past. Nothing has affected businesses and consumers since the pandemic as much as the surge of inflation. Whatever you choose to cite as the culprit — the Ukraine war, monetary looseness, lockdown-induced damage to supply chains — it was political. That experience should have reminded us of the primacy of the public realm. Instead, the fascination with technology as the shaper of realities has only increased over the period.

7. From a CAG performance audit report on the state of ULBs in 18 states of India.

According to the audit, on average, only 4 out of the 18 powers under the 12th schedule are fully under the autonomous control of ULBs, with most functions being performed with regular interference from the state government or parastatals, often without any representation from local bodies. In addition, ULBs are being deprived of making their own recruitment decisions, as the staff assessments are conducted by the state government, leading to frequent underestimation of personnel requirements... This lack of autonomy has not only resulted in fewer sanctioned positions but has also left one out of three posts vacant across the 18 states, depriving ULBs of the human resources necessary to carry out their functions properly... According to the report, 61 per cent, or 1,600 of the 2,625 ULBs in 17 states assessed, didn’t have an elected council, with only five states appointing a mayor through a direct election.

8. Amidst all the alarm about public debt, Martin Sandbu points out that borrowing costs are high only compared to the 15 years or so of after the GFC of very low interest rates. He points out that the US government now devotes the same share of GDP to interest paymens as it did in the late nineties, just under 4%.  

9. Kai Wu has the graphic that sums up the dilemma facing AI companies. 

10. China's scale of excess capacity is staggering.

China now makes 55 per cent of the world’s steel, 57 per cent of commercial vessels, 76 per cent of lithium-ion batteries, more than 60 per cent of EVs, and 80 per cent of photovoltaic products, despite chronic involution gripping these sectors... The country now has 58 satellite makers and 30 rocket companies and more than 60 humanoid robot manufacturers. Domestic commentators already warn of capacity outstripping demand and of companies needing to “go out” to foreign markets to survive. The pattern repeats itself: what begins as glut at home could end as supremacy abroad.

Chinese firms in these sectors with such levels of excess capacity must aggressively pursue export markets. They must also slash costs relentlessly and surive on razor-thin margins, or increasingly assume losses. But the limits to such export-led growth are now evident. 

11. As rare earths become one of the hottest commodities, Australia has emerged as the global leader in spending on the exploration of these minerals.

Interesting that India has the third largest reserves.

12. Edward Luce on Trump's America that is gripped by fear.

Revenge is one of Trump’s three recurring impulses. The others are making money and dominating the airwaves. Dissenters hinder each of these aims. Presidents of universities, chief executives of Fortune 500 companies, partners at law firms and senior military privately despair at Trump’s methods. But each has sound stakeholder reasons for keeping their concerns private. Universities stand to lose billions of dollars of federal research funds; chief executives and their workforces face regulatory reprisal; law firms end up on federal blacklists; soldiers are trained to uphold the chain of command... people are scared of crossing Trump this time. In researching this piece, I interviewed dozens of figures, including lawmakers, private sector executives, retired senior military figures and intelligence chiefs, current and former Trump officials, Washington lawyers and foreign government officials. Such is the fear of jail, bankruptcy or professional reprisal, that most of these people insisted on anonymity. This was in spite of the fact that many of the same people also wanted to emphasise that Trump would only be restrained by powerful voices opposing him publicly. At times, it has felt like trying to report on politics in Turkey or Hungary...

As a rule of thumb, the more an organisation has to lose, the likelier it is to submit to Trump’s demands... Under threat of being debarred from the federal government, and thus losing corporate clients, many big law firms have declined to hire or represent people on Trump’s enemies list. “It’s not a list but I think there will be others,” said Trump after Comey was indicted. The universe of lawyers who would represent his targets has shrunk dramatically. None of the team who worked for Jack Smith, Biden’s special counsel who indicted Trump for allegedly attempting to overthrow the 2020 election and hoarding classified documents in Mar-a-Lago, has since found a job. Family members are not spared. Maurene Comey, the former FBI chief’s daughter, was fired as a federal prosecutor in July. The pleas of staff to FBI director Kash Patel not to fire a senior official whose wife was dying of cancer fell on deaf ears.

Technocracy and fiscal management

I have blogged on multiple occasions, pointing to the perils of excessive reliance on experts. 

Central banks are considered the epitome of technocracy in economic policymaking. Much has been written about how independent central banks manned by technical experts and using technical rules like the Taylor Rule and inflation targeting have tamed inflation and ensured macroeconomic stability. Never mind the several questions and disputes surrounding this narrative. I have blogged here, here, here, and here, trying to place central bank independence and competence in perspective. 

Since the global financial crisis, there has been an extraordinary expansion of the toolkits used by central bankers. Policies like quantitative easing, yield curve control, purchases of corporate bonds, forward guidance, and so on, all emerged anew into the monetary policy basket under the leadership and technical expertise of academic scholars and experts like Ben Bernanke and Janet Yellen. These policies have been hailed for rescuing and restoring the economy and financial markets, both during the GFC and after the COVID-19 pandemic.

However, it is now apparent that the long period of monetary accommodation engendered by these policies, under the watch of esteemed experts, has contributed to an addiction to cheap money, perpetuated zombie companies, turbocharged financial models like private equity, and inflated financial market bubbles. It’s a legitimate and very compelling argument that these policies have prevented the small recessions necessary to clean up excesses and realign incentives. 

Instead of technocracy binding politicians to the mast and restraining them from the pursuit of excessively loose monetary policy, the expert central bankers appear to have shown the politicians the way with new toolkits to perpetuate cheap money policies. The most egregious expression of this reshaping of expectations is Donald Trump’s demands from the US Federal Reserve. 

This fetish with technocracy is not confined to central banking. Based on the successes attributed to technocratic central banking, economists have argued in favour of fiscal councils to independently evaluate and monitor the expenditure and tax policies of governments. They say that fiscal councils, with their independent role, can counter the deficit bias of governments and prevent fiscal dominance. Accordingly, many Western countries have some form of fiscal councils. 

In this context, Andrew Haldane, former Chief Economist of the Bank of England and one of the most respected economic commentators, has set the cat among the pigeons by questioning the role played by the UK’s Office of Budget Responsibility (OBR) in the country’s fiscal management. The OBR was established in 2010 to provide an independent assessment of the country’s public finances and thereby depoliticise fiscal policy analysis. It mimics the independent fiscal policy councils operating in some countries, which provide an independent view on the Government’s macroeconomic forecasts and fiscal decisions. 

However, the OBR’s role goes beyond mere assessment of fiscal policy to playing the central role in making macroeconomic forecasts and assessing the impact of fiscal measures. It had, as Haldane writes, “monopoly rights over judgments on debt sustainability.” In other words, the Treasury outsourced its role in this to the OBR, including transferring much of its in-house expertise to do this role. This was a pure form of technocracy. 

Haldane says this outsourcing to a technical entity has contributed significantly to the UK’s current economic stress by subordinating economic growth to excessive fiscal discipline, with its inevitable political consequences. 

Since 2010, fiscal policy has involved delicately balancing measures to stimulate growth with maintaining fiscal discipline. The OBR’s mandate covers only the second. Its scoring of fiscal measures decisively tips the institutional balance towards conservatism over growth. Or rather, it has reinforced the Treasury’s long-standing fiscal-first instincts… After years of under-investment, the UK’s public sector capital stock is estimated to be around £2tn lower than its international counterparts in 2019, a gap almost certainly larger now. Not coincidentally, growth has stalled. An unedifying sequence of gossamer-thin growth plans has been accompanied by mounting political disquiet at OBR conservatism.

This culminated in Liz Truss’s decision to sideline the OBR in preparations for the fateful 2022 “mini” Budget. The resulting bond market meltdown led present chancellor Rachel Reeves to hardwire OBR assessments into fiscal events, making the de facto monopoly de jure. Buyer’s remorse has been rapid. With a weakening outlook and far too little fiscal wriggle room, Reeves finds herself impaled on the OBR’s horns. On its educated guesses — and that inevitably is what they are — now hang the fortunes of the chancellor, the economy and tens of millions of taxpayers… Nigel Farage, whose Reform UK party leads comfortably in opinion polls, suggests that weak growth is the OBR’s fault.

As Haldane writes, the OBR appears to have done its job all too well, only to the extent of squeezing hard on economic growth itself. In this backdrop, Haldane’s suggestion is to limit OBR’s role to auditing the Treasury’s assessments. 

One way of freeing the government’s fiscal hands is by partially taking back control of fiscal assessments. Outsourcing your brain is rarely wise. As with the Bank of England for monetary policy, the Treasury should produce and publish its own economic projections and assessments of fiscal choices. The OBR’s role, as in other countries, would then be to audit these assessments. With the Treasury no longer as tightly bound by OBR conservatism, the institutional balance would be tipped towards growth while preserving independent scrutiny. Increasing transparency around fiscal choices improves public debate.

In the context of the debate about the superiority of independent technocratic entities like central banks or fiscal councils, especially given the fiscal bind in the UK, India’s post-pandemic experience is instructive. 

The country’s fiscal framework, enshrined in the Fiscal Responsibility and Budget Management (FRBM) Act, mandated governments to keep their gross budget deficits under 3% of the GDP, a benchmark that has no objective basis but was straight borrowed from the EU (where, too, it was forced without any objective basis). While it was never strictly followed (except for state governments), it nudged successive central governments not to stray too far from this number. 

The pandemic helped break away from this constraint and allowed the central government to find an average of nearly 2.5 percentage points of GDP of additional fiscal space (comparing the six years immediately before the pandemic with those immediately after). This additional fiscal space has been critical, almost single-handedly responsible, in sustaining and boosting economic growth. It is to the government’s credit (a surprisingly less acknowledged thing) that it used this additional fiscal space not to dole out subsidies and other revenue expenditures, but on good-quality capital expenditures that created durable assets, and also to clean up its budget books. 

The big post-pandemic fiscal deficit and the failure to reverse course quickly to the FRBM benchmark raised criticism from experts and opinion makers. They warned of macroeconomic instability, a surge in public debt, capital flight, growth squeeze, and a knock-on effect on the equity markets. None of these has materialised, and, despite the headwinds from global uncertainties and weaknesses, the Indian economy remains in reasonably good shape. 

Inflation has been low and growth high, especially when compared to peers and advanced economies. In fact, while it will be a matter of debate, there may now be a case to even revisit the fiscal framework to anchor the benchmark at about 4% of GDP.

The point here is not to reject technical expertise and technocracy in macroeconomic policymaking and public policy in general, but to caution against excessive reliance on them. Public narratives tend to endow them with expertise and prescience far in excess of what they possess, especially in complex areas like macroeconomic decision-making. Given the deeply political nature of these decisions, it’s more appropriate if they are taken within governments, by drawing on the inputs and expertise of technical experts.