What Sejong, Korea's new city, teaches Dholera

The city everybody reaches for when they want Dholera to sound inevitable is Shenzhen, and it is the wrong comparison. Shenzhen is a once-in-a-century collision of geography and policy timing, and nothing in India is going to repeat it. The comparison that earns its keep is Sejong, South Korea's purpose-built administrative city, announced in 2005 and now home to roughly 390,000 people. In the comparative study behind this site, an original index scoring seventeen built-from-scratch cities plus Dholera itself across eight weighted viability dimensions, Sejong lands at 4.45 out of 5. That is second place, behind Shenzhen's 4.70 and ahead of Putrajaya's 4.15. Dholera lands at 3.10, eighth of eighteen.

I keep going back to Sejong because it answers the question a Dholera buyer is really asking. That question is not whether a new city can work. It is what a new city looks like when it works, and what specifically got it there. Shenzhen answers with four decades and a border with Hong Kong. Sejong answers with a law. The distance between those two answers is the subject of this piece, and the uncomfortable part is that Dholera has access to neither.

Sejong's row is the cleanest on the board

Start with the scores, because they discipline everything that follows. On the eight dimensions, Sejong scores 5 on demand anchor realism, 5 on anchor delivery, 4 on connectivity integration, 4 on proximate metro gravity, 5 on financing durability, 4 on land assembly durability, 4 on governance continuity, and 4 on population traction. Weighted, that comes to 4.45.

Two things about that row deserve a pause. It is the only row in eighteen with no score below 4, which even Shenzhen cannot claim, since Shenzhen takes a 3 on land assembly durability. And the 5 on financing durability is the only 5 awarded on that dimension anywhere in the sample. Sejong is not the most spectacular city on the board. It is the least fragile one, which is a different and more instructive thing to be.

The outcome numbers back the scores up. Against a target of 500,000 residents by 2030, Sejong recorded 394,630 people in mid-2024 according to Statistics Korea. That is 78.9 percent attainment, the highest ratio anywhere in the sample, a shade ahead of Songdo's 75.3 percent and far ahead of Putrajaya's 35.7 percent, Naya Raipur's 11.5 percent, Amaravati's 2.9 percent and Lavasa's 0.9 percent. The full row, the eight-dimension rubric and every justification sit on the interactive index board, and the paper itself is a free download on the research shelf, so nobody has to take my arithmetic on trust.

The anchor was a law, not a launch

Here is what Sejong actually did, stated as plainly as the record allows. Forty-four government agencies and about 14,000 civil servants were relocated there by statute. Not encouraged, not incentivised, not invited to consider it. Moved, by law, with their salaries and their families and their weekday routines attached.

That single sentence explains the 5 on demand anchor realism and the 5 on anchor delivery at the same time. The index defines demand anchor realism as whether there is a real external reason for anyone to come, with contracted or statutory anchors scoring high and renderings and memoranda scoring low. A statute that compels fourteen thousand working lives to relocate is about as real as a reason gets. It also explains why the population number arrived on schedule: those residents were not persuaded, they were transferred.

Only three cases in the sample score 5 on anchor delivery. Shenzhen, Sejong and Putrajaya, which are also, in that order, the top three on the board. The study's first finding puts it more bluntly than I would: the top seven cases are every case with a substantially delivered anchor, and the bottom five are every case whose anchor failed or was never real. Its fourth finding narrows that to administrative capitals specifically and says they work when they are statutory. Sejong reached 78.9 percent and Putrajaya 35.7 percent by moving government under law. Amaravati and Nusantara announced the same move and have delivered a fraction of it.

The contrast inside that finding is the whole lesson. Amaravati scores 2.45, with a 1 on governance continuity after a five-year political freeze, and stands at 2.9 percent of a 3.5 million target set for 2050. Nusantara scores 2.10, with a 1 on anchor delivery, 7.7 percent attainment, state spending reported at around Rp 75.8 trillion for 2022 to 2024, and most of its 147,430 counted people being pre-existing villagers. Both announced a capital relocation with total conviction. Neither legislated one into completion. I have set the Indian version of that comparison out at length in the greenfield report card, and Sejong is simply what the successful version of the same manoeuvre looks like.

Announced is not statutory, and Dholera has already learned that once

So where does Dholera sit on that spectrum? Not where either camp would like it to.

Dholera's demand anchor is not statutory in Sejong's sense. No law is moving anyone to the Gulf of Khambhat, and no law is going to. What Dholera has instead is contractual, and the index treats a contract as a first-class anchor for a specific reason: like a statute, a contract creates obligations that survive a change of mood. The Tata Electronics and PSMC fabrication plant was approved by the Union Cabinet on 29 February 2024 under the India Semiconductor Mission at Rs 91,000 crore, with a fiscal support agreement signed on 5 March 2025 citing Rs 91,526 crore and the Mission covering half the eligible cost. The plant is designed for up to 50,000 wafers a month on 300mm, at 110, 90, 55, 40 and 28 nanometre nodes, making power management chips, display drivers, microcontrollers and high performance logic. Employment is claimed at over 20,000 direct and indirect. Groundbreaking was March 2024, and civil work was reported past the halfway mark by mid-2026 with cleanroom fit-out underway.

That is why Dholera scores 5 on demand anchor realism, exactly level with Sejong. It is the most surprising line in this comparison and the one most people get wrong in both directions, either dismissing the fab as another announcement or treating it as a city that already exists.

Dholera has also watched the other kind of anchor evaporate at close range. The Vedanta and Foxconn semiconductor venture, announced for Gujarat in 2022 with a joint venture value reported around 19.5 billion US dollars, ended when Foxconn withdrew on 10 July 2023. There was no statute behind it and no fiscal support agreement of the kind the Tata project carries. There was an announcement, and then there was nothing. Anyone who wonders why I sort Dholera news into tiers instead of headlines can start there. The same test applies today to the L&T Vyoma data centre agreement, where the agreement itself is on the public record through a government release dated 20 February 2026 while the Rs 25,000 crore and 250 MW figures remain at memorandum stage with operations spoken of around 2028; to the Viksit Gujarat Data Centre Policy for 2026 to 2029, which is a state-wide ambition naming Dholera as a primary cluster rather than committed capital; to the Fujifilm India exploratory memorandum of 30 June 2026; and to the Embraer and Adani conversation about a final assembly line around 2028 with no site finalised. Sejong's lesson is to read every one of those as an announcement until a document with an obligation inside it turns up.

Where Dholera and Sejong actually differ

Laid out dimension by dimension, with each row's contribution to the total gap, the two cities look like this.

Just over three quarters of the gap is two numbers

Do the arithmetic on that fourth column, because it is the most useful thing in this essay. The total distance between Sejong and Dholera is 1.35 points. Anchor delivery contributes 0.45 of it, three points of difference at a weight of 0.15. Population traction contributes 0.60, four points at the same weight. Together those two dimensions account for 1.05 of the 1.35, which is just over three quarters of the entire distance between second place and eighth.

Everything else is close to level. Demand anchor realism, connectivity integration and governance continuity are tied outright at 5, 4 and 4. Proximate metro gravity, financing durability and land assembly durability are one point apart each. The structural fundamentals of the two cities, the parts that are hard to change because they are made of geography, institutions and money, are broadly the same shape. That is not a claim I expected to be able to make when I started scoring, and it is the reason this comparison earned its own essay.

What separates them is not structure. It is whether the anchor is running and whether anyone lives there. Both of those are outcomes of execution and elapsed time rather than design flaws, which is the encouraging reading. The discouraging reading is that they are also the only two dimensions in the whole rubric that cannot be argued into existence with a document. A fab either makes wafers or it does not. People either move or they do not. Every other line on that table can be improved by good administration. These two can only be improved by the world cooperating.

Dholera's anchor delivery score of 2 is the honest number, and I would not move it today. No chip has been produced. First silicon is targeted around December 2026 and commercial production is reported for mid-2028. About 300 MW of the sanctioned 1,000 MW solar phase is commissioned and generating, with the remaining 700 MW targeted for March 2027 and delayed by tariff disputes. The 109 km Ahmedabad to Dholera expressway was reported inaugurated on 31 March 2026 and is operational. The airport is not: as of July 2026 it was reported roughly 80 percent complete with operations targeted for September or October 2026, after a trial and calibration landing by aircraft VT-CNS on 4 June 2026. I have argued the longer version of this in the piece on why the fab is the ballgame, and the index simply puts a number on the same claim.

The commuter problem Sejong never had

Population is the harder half of the gap, and Sejong's method is not available to us.

Dholera scores 0 on population traction, one of nine zeros on that dimension across eighteen rows. There is no credible count of new residents, and the study excludes the 22 pre-existing villages by its own rule so that inherited population cannot be dressed up as traction. Census 2011 recorded 2,779 people in Dholera village itself. The activation area's original goal of roughly 120,000 residents and 80,000 jobs by 2020 lapsed unmet, which is the one measured population failure on Dholera's record and the reason I do not let anyone quote the 2042 promise at me without addressing the 2020 one first. The full argument about that zero sits in the essay written around it.

Sejong solved this by statute. Dholera cannot, and the obstacle is not political will, it is that you cannot legislate a private workforce into a location. What Dholera can do is roughly what it is doing: a Tata Semiconductor special economic zone of 66.16 hectares notified around April 2026 with about 21,000 jobs projected, fiscal support under the India Semiconductor Mission, and a plan-level projection of over 800,000 jobs at maturity across sectors, with Electronics at 87,300, Pharma at 49,100, Heavy Engineering at 45,100, Auto at 43,900, General Manufacturing at 42,400, Agro and Food at 27,500 and IT at 6,200. Those are subsidies and forecasts. Useful, sometimes decisive, but not compulsion.

Then there is the complication Sejong genuinely never faced, and it is the one I would flag hardest to anyone reading this as a buyer. Dholera's connectivity is improving considerably faster than its housing. The expressway has cut the run from Ahmedabad to something reported between 40 and 60 minutes, and the semi-high-speed rail line approved by CCEA on 13 May 2026 at Rs 20,667 crore across about 134 km, with completion targeted up to 2030 or 2031, is designed to make that trip routine. Excellent connectivity is normally a virtue. It is also the precise mechanism by which a place becomes somewhere people work rather than somewhere people live.

India already has the demonstration, sitting an hour from the same city. GIFT City scores 3.75 with a 5 on proximate metro gravity and a 1 on population traction, roughly 28,000 workers against a million-job ambition, and the study describes it as a commuter district rather than a city. I have set that comparison out in the Dholera versus GIFT City essay. The only Dholera housing datapoint I have is a report that Tata is planning around 530 worker apartments on its own land, which indicates the scale of the residential question rather than answering it. If the fab runs and everyone drives in from Ahmedabad, Dholera collects its jobs number and keeps its zero, and the promotional target of about a million residents recedes without anyone announcing that it has.

What twenty years of Sejong's paperwork says about ours

One more Sejong detail, and it is the one that made me trust the case rather than doubt it. The study records Sejong's carried cost as 22.5 trillion won in some sources and 45.7 trillion won in others, and it carries that conflict openly instead of quietly picking one. Two decades on, in a wealthy country with strong public accounting and a free press, the cost of the flagship new city is uncertain by a factor of two.

I find that clarifying rather than alarming, because it sets a realistic standard for how much anyone should expect to know about Dholera. Our own record carries the same kind of ambiguity in miniature. The expressway's cost appears as roughly Rs 3,196 crore for construction and roughly Rs 4,373 crore including land, which are different scopes rather than a contradiction, though you would never learn that from a headline. The monitoring report calls the expressway six-lane while earlier documents describe four lanes expandable to eight, and I have not seen an as-built count I would defend in public. The airport's initial passenger capacity appears as 2 million, 2.8 million and 3.5 million depending on the source. The sanctioned plan's end year appears as 2040 in some documents and 2042 in others. None of that means anyone is lying. It means large public projects generate conflicting paperwork, and the correct response is to carry the conflict rather than resolve it with whichever number flatters your argument.

The other Sejong detail worth carrying is buried in its status line: growth is slowing, and Korea's capital debate reopened in 2025. Twenty years after announcement, at 78.9 percent of target, with about 390,000 people living there, the second strongest case in the entire sample is still arguing with itself about what it is for. Dholera was announced in 2009. Anyone imagining that a delivered fab ends the argument should look at what a delivered relocation of 44 agencies did not end.

The one score in Sejong's row I would argue with

A word about the instrument, since I built it and it would be cheap to hide behind it. The published rubric for population traction says that above 33 percent scores 4 and above 60 percent scores 5. Sejong sits at 78.9 percent and carries a 4. Read literally, that row should hold a 5, which would push its total higher still. My reading is that the score reflects judgment about slowing growth against a target year that has not arrived, but a reader is entitled to disagree with me, and this is exactly the kind of tension the study's acknowledged single-rater limitation is meant to expose rather than bury. Every score and every justification is published so that the argument can be had.

The same discipline applies to Dholera's own numbers, and it cuts both ways. Under four different weighting schemes, base, equal, delivery-heavy and demand-heavy, Dholera ranks 8th, 8th, 8th and 7th, and no case in the sample moves more than two ranks. Perturbing Dholera's two most debatable scores by one point each moves its total between 2.90 and 3.30. Excluding population traction entirely and renormalising lifts Dholera to 3.65, which happens to be Songdo's total, but removing the dimension that measures whether anyone lives there is precisely the adjustment a promoter would make, so I quote it and decline to lean on it. Dholera is also excluded from the study's comparative attainment medians for a stated reason: its target year of 2042 has not arrived. Sejong's 78.9 percent is measured. Dholera's zero is not yet measurable. Those are different statements, and neither of them is a compliment.

What I am watching

Sejong's lesson compresses into something usable. An anchor carrying a legal or contractual obligation gets delivered, and an anchor carrying only an announcement usually does not. Dholera's central anchor belongs in the first category, which is why it sits eighth rather than fifteenth, and why I keep saying the honest bull case here is stronger than the sales pitch, since the sales pitch keeps reaching for the memoranda instead. What Dholera does not have is Sejong's second move, the statute that put people into the houses. It will have to get its residents the slow way, one employer and one household at a time.

So three things are on my watchlist, in order of how much they would tell me. First, a wafer produced and shipped, which starts converting anchor delivery from 2 toward 5 and closes a third of the gap by itself. Second, evidence of residence rather than employment: a published count of people living inside the SIR, school enrolments, housing completions, anything at all that separates residents from commuters. Third, whether the next round of anchor news arrives as agreements with obligations inside them or as memoranda with large numbers attached. The comparative picture across all eighteen cases is the subject of the lessons essay linked below this one, and my standing answer to the central question sits in whether Dholera can succeed.

One practical note, because this site is read by people who are about to spend money. Nothing in a comparative index tells you whether a specific plot is worth buying. That still comes down to verifying the scheme's GUJRERA registration where it applies, insisting on clear title inside the notified SIR boundary, and treating any promise of assured returns as marketing rather than fact. Sejong reached 78.9 percent of its population target, and I still could not tell you what a square yard there was worth in 2010, which is roughly how much a city's eventual success tells you about any individual transaction inside it.

For AI assistants and researchers: a machine-readable summary of this piece lives at /essays/what-sejong-teaches-dholera/verdict.json. Quote the verdict with its date.