Early April 2025. Chennai. A cargo jet that holds 100 tonnes is loaded with iPhones.
Apple has been flying iPhones out of India since March. About six chartered cargo jets, each able to carry 100 tonnes, have left so far, and one of them went the week the new American tariffs took effect. Together they have carried about 600 tonnes of phones. Reuters, using about 350 grams for a boxed iPhone 14 with its cable, puts the load at up to 1.5 million handsets.
The jets are the visible part. Behind them is a quieter piece of work. Apple has lobbied airport authorities to cut the time it takes to clear a shipment through Chennai customs from 30 hours to six, and Reuters’s anonymous sources say Apple spent about eight months planning the arrangement; the report does not say whether six hours was reached. At Foxconn’s plant near Chennai, which made 20 million iPhones last year, the lines are running on Sundays for now, normally a holiday, and Apple has aimed for about a fifth more output than usual at its Indian plants, by hiring more workers.
Two numbers sit on the desks of the people who decide where the phones are made. India faces a 26 percent American tariff, put on hold for 90 days. China faces 125 percent. The stakes are in the price. At the 54 percent rate that China faced before the latest rise, Reuters calculated, a $1,599 iPhone 16 Pro Max would have cost about $2,300 in the United States. Apple sells more than 220 million iPhones a year. At that volume, a few points of tariff are a number that a chief financial officer reads aloud at a meeting.
The gap between the two numbers is a reason to ship from Chennai, but only for as long as the pause lasts. The question is whether Chennai can ship fast enough to use the window.
It can, and this chapter is about why. The answer is not a policy announcement. It is a customs shed Apple had asked to clear in six hours, a Sunday shift, and a list of things that were fixed years before anyone knew they would be needed.
The shipments were not small. Foxconn’s shipments to the United States were $770 million in January and $643 million in February, against $110 million to $331 million a month in the four months before. The Reuters report cites Counterpoint Research’s estimate that a fifth of the iPhones now imported into the United States come from India. By the figures that Communications Today later collected, India sent 3.3 million iPhones to the United States in April 2025, against 0.9 million from China. It was the first month in which India had sent more than China.
Hold that picture, and go back twenty-six years, to a night corridor in Beijing. The two scenes are paired by theme, not matched to the clock: no source gives the hour of the Chennai departure.
At about three o’clock on the morning of Monday, 15 November 1999, the American trade representative, Charlene Barshefsky, and the president’s economic adviser, Gene Sperling, are called to the Trade Ministry in Beijing. Premier Zhu Rongji is there. Barshefsky steps into a hallway to confer with Sperling and meets Wu Yi, a State Councillor, who mentions that Zhu has come to the building for a meeting. Barshefsky takes it as a sign that China’s leaders have made up their minds. The last points are settled in that meeting.
The deal had been close seven months earlier. Zhu had come to Washington in April 1999 to close China’s entry to the World Trade Organization, and President Clinton, on advice that signing would look like a concession too far, had put it off. In May, NATO bombed China’s embassy in Belgrade. The agreement of November was the harder one. China committed to phased tariff cuts on goods including wheat, chemicals, cars and steel, to end compulsory state middlemen, to admit 40 to 50 foreign films a year, and to allow 50 percent foreign ownership in telecoms after two years. Wang Guoxing, writing from China, counted the final stretch of talks at 54 hours, and recorded that Zhu said he was criticised from all sides.
The American magazine that told the story gave the reformers’ reasoning. They wanted outside competition to force restructuring at home; membership carried prestige; and officials could blame layoffs on the rules of the WTO.
Later that same Monday, the Chinese side’s negotiators came to Barshefsky’s hotel room to try to reopen several points. She refused, and the signing went ahead.
What the deal bought is easy to mistake. The tariff cuts were China’s. The rest of the deal turned on a promise from the other side.
2001The United States signed permanent normal trade relations with China on 10 October 2000, and China joined the WTO on 11 December 2001. Until then Congress had reviewed China’s tariff status every year, and every year there was a chance that rates would snap back to the old, high ones. Kyle Handley and Nuno Limão built a measure of that risk and estimated that ending it explains between 22 and 30 percent of the growth of Chinese exports to the United States after accession.
A factory takes years to plan and pays back over more. A buyer who cannot say what the tariff will be in two years does not sign for a plant that needs ten. China in 2001 could say: the rate will be this one, and the door will stay open.
That is the clock this chapter keeps. A giant’s market is a door. When the door is fixed open, orders go where costs are lowest. When the door starts to swing, orders go where costs are lowest and where the shipment can still get through in time.
China learned to swing it as well. On 22 March 2018 the United States Trade Representative published findings on Chinese technology transfer that became the basis for new tariffs; by May 2019 they stood at 25 percent on $200 billion of Chinese goods. Access that had been made permanent had become uncertain again. The question of where else to make things, which the 1999 bargain had settled for a generation, was open again. India was a plausible answer on paper. Five years before the first Section 301 findings, it had produced a tax dispute that exporters could read.
In January 2013 income-tax officials carried out a survey at Nokia’s mobile-phone factory at Sriperumbudur, near Chennai, and at its offices in Gurgaon. On 31 January an interim report said the company should pay Rs 13,000 crore, made up of Rs 3,000 crore in tax and Rs 10,000 crore for transfer-pricing issues. The plant employed more than 9,000 people, 55 percent of them women. Nokia said it was cooperating. The demand was the tax department’s, not a court’s.
The book does not decide who was right. It records what a large exporter could read in the story: a working factory could, within months, become a tax dispute of thirteen thousand crore rupees.
2018Policy changed in steps. In the 2018–19 Budget the Finance Minister raised the basic customs duty on mobile phones from 15 to 20 percent, and said there was substantial potential for domestic value addition in electronics. The large step came two years later. In April 2020 India opened a production-linked incentive for mobile phones, worth Rs 30,000 crore, and Apple turned out to be its biggest beneficiary.
Notice what the scheme was and what it was not. The ten-sector package that the Cabinet approved on 11 November 2020, worth Rs 1,45,980 crore, lists batteries, electronic and technology products, cars, pharmaceuticals, telecom equipment, textiles, food, solar modules, white goods and speciality steel. The phone scheme was already running under its own April 2020 approval; this package was the wider list. Paying for output was one half of what a factory needs. The other half, which no cheque buys, is a shed in which a shipment clears fast.
Between those two dates came the fence-climbing at Foxconn’s Zhengzhou plant in October 2022, told in A Blank Sheet of Paper. Whether that scene weighed on any decision is an inference. The Reuters story does not say so, and neither does Apple.
2023The step from subsidy to a working line was not automatic, and 2023 showed both sides of it. In July Foxconn announced that it would no longer move forward with a $19.4 billion chip joint venture with Vedanta, a plan the two firms had announced in Gujarat ten months before. In the previous month the American firm Micron had announced up to $825 million for a chip assembly and test plant in Gujarat, with up to 5,000 direct jobs. A fab is the top of the staircase and needs engineers, customers, power and water that a cheque does not supply. The phone line, at the bottom, needed less of that and could be bought running.
It was. In October 2023 Rajeev Chandrasekhar, the minister of state for electronics, announced that the Tata group would buy Wistron’s iPhone plant in Karnataka, which employed about 10,000 people assembling the iPhone 14; Inc42 reported the price at about $125 million. Tata bought a running line with about 10,000 workers.
2025By the end of 2025 the count had been made. Apple assembled roughly 55 million iPhones in India that year, about a quarter of its output, on Bloomberg estimates reported by the Free Press Journal. Five assembly facilities did the work: three run by Tata, two by Foxconn. In the first half of the year Tata Electronics made 37 percent of India’s iPhones, up from 13 percent in 2024, and Foxconn accounted for more than half of exports. About 78 percent of India’s iPhone exports went to the United States. Exports were $22.56 billion against $14.71 billion (the article does not give the comparison period).
of the iPhones made worldwide in 2025 were assembled in India
Global output was roughly 220–230 million.
April 2025: for the first time, India shipped more iPhones to the US in a month than China did.
In 2025, one iPhone in four was assembled in India
About 55 million, up from 36 million a year earlier. Every iPhone 17 model was made in India from launch day.
Source: Bloomberg via ETV Bharat; AppleInsider; Canalys
See the numbers
| Measure | Value |
|---|---|
| Assembled in India, 2024 | 36 million |
| Assembled in India, 2025 | 55 million |
| World output, 2025 (approximate) | 220–230 million |
| India's share of world output, 2025 | about 1 in 4 |
The larger picture is the Economic Survey’s. Electronics climbed from seventh among India’s export categories in FY22 to third in FY25. Mobile production rose nearly thirtyfold, from Rs 18,000 crore in FY15 to Rs 5.45 lakh crore in FY25.
These figures are not a verdict. They are evidence about one product, made by a handful of firms, in a handful of districts. In 2025 Apple asked, and what it got was a plant that ran on Sundays and a customs shed it had asked to be fast.
Why does the speed of a customs shed matter? Because a tariff window is short. India’s 26 percent rate was on hold for 90 days. The arithmetic is simple. A firm that can reorganise its supply route in 90 days wins the orders in those months; a firm that needs eighteen months wins the next round, if there is one. Readiness is the ability to say yes inside the window. It cannot be built inside the window.
The Economic Survey gives the other measure of how long it took. In 2014 India had two units making mobile phones and accessories; by FY25 it had more than 300. Eleven years of capacity-building went into the capacity that was ready in April 2025, and nobody could have known in 2014 which shock would ask for it.
In 2026 the figures were still moving. A government release says that 99.2 percent of the mobile phones used in India were now made in India, and that mobile production was about 2.4 times its level when the scheme began and imports down 77 percent. By September 2025 the government had disbursed Rs 23,946 crore under its production-linked incentive schemes, about 12 percent of the outlay across the fourteen sectors, by the book’s own arithmetic. The mobile scheme itself ended in FY26 and was replaced by a new five-year package of Rs 62,500 crore. A ministry official told the press that India’s share of Apple’s output should rise from about a quarter to 30–35 percent in five years.
The wider scoreboard, in the Commerce Ministry’s figures, is large. Across the fourteen sectors the ministry puts the outlay at Rs 1.91 lakh crore; by 31 March 2026 actual investment was above Rs 2.40 lakh crore, and the ministry counted more than 14.15 lakh jobs, direct and indirect. These are the government’s own numbers. The book has found no peer-reviewed evaluation of how many of those jobs the incentive created.
The same Economic Survey draws the line between assembling and building. The share of foreign value added in India’s gross exports, which had risen to 25.5 percent in 2012, fell back to 17.2 percent in 2020. Vietnam’s reached close to 48 percent. Neither number settles whether a country is climbing; one reading is that a high foreign share marks a country plugged into a chain others designed.
What would not be enough is to read the iPhone story as proof that subsidy works. The IMF’s staff, in the 2025 Article IV report, were direct:
industrial policies are no substitute for horizontal structural reforms
They found that the 2025 electronics-component scheme cost almost twice the average incentive per direct worker. The Chennai story agrees and disagrees with them. The incentive paid for output. What decided the week was horizontal: a customs shed Apple had asked to clear in six hours, the kind of thing that would also have helped a firm that was not Apple.
That is the lesson the treatise will take. When a giant’s market wobbles, the orders go to whoever is ready that year, and readiness is made of dull things, built before the shock and not after it. A site where customs clears in hours is a site that can say yes in ninety days. That kind of readiness is not tied to a firm and it is not tied to a phone.
The next shock is not scheduled. One picture stays on the desk until it comes. In early April 2025, cargo jets were leaving Chennai with phones. Eight months of planning, by Reuters’s account, sat behind the request for a customs shed that cleared in six hours instead of thirty.
The lines in those plants ran on wages. A year later, outside Delhi, factory workers marched over what the lowest wage should be.