Can Dholera actually succeed? The bull case, the bear case, and the three things that decide it

Can Dholera be successful is the question underneath every other Dholera question, and it suffers from a definitional swindle on both sides. Sellers define success as inevitability, quote the 2042 brochure, and declare it arrived. Skeptics define success as Manhattan by Tuesday, observe correctly that the plains are still mostly plains, and declare the whole thing a mirage. Both are arguing with a definition, not with the evidence. So let me fix the definition first and then argue both sides as hard as I honestly can.

Success, for a greenfield industrial city, means three stacked achievements: anchor industry operating, supporting economy forming around it, and a resident population that lives there because working there makes it rational. Dholera does not need to beat Ahmedabad to succeed. It needs to become the place where a specific slice of Indian manufacturing physically happens, with a town attached. That is the claim on trial.

The bull case, at full strength

First, the anchor is real and mostly irreversible. The Tata Electronics fab is Rs 91,000 crore of committed capital with Cabinet approval from February 2024, a signed Fiscal Support Agreement from March 2025, and civil work reported past halfway with cleanroom fit-out underway in mid-2026. Fabs are the least footloose factories on earth: the building is the machine. Once tool move-in begins, the plant operates or the write-off is measured in tens of thousands of crores. Committed capital of that shape is how you distinguish this cycle of Dholera enthusiasm from the previous ones, and I treat the fab as the single load-bearing fact of the whole thesis.

Second, the infrastructure sequence ran in the correct, unusual order: trunk before towers. The DMU ledger records activation-area trunk works complete, built on Rs 2,784.83 crore of approved packages, and the sanctioned plan details what that bought: roughly 72 km of internal roads, a 50 MLD water treatment plant, separate recycled-water and effluent networks, underground power ducting, and an operations centre in the ABCD building actually running city systems. India has many townships with towers and no infrastructure. It has very few zones with infrastructure waiting for towers. The second kind ages better.

Third, connectivity is arriving in the right sequence too: the expressway is open and reported at roughly a forty-minute to one-hour drive from Ahmedabad, the airport is at roughly 80 percent with a trial landing done, and the semi-high-speed rail was Cabinet-approved in May 2026 at Rs 20,667 crore targeting 2030-31. Each layer compounds the others: the expressway alone rewired the labour-shed math by putting the site inside a commutable radius of Ahmedabad's established labour market.

Fourth, the policy tailwind is unusually specific. Gujarat's 2026-29 data-centre policy names Dholera as its primary cluster, the state budget carries a dedicated Rs 610 crore Dholera line, and Cabinet cleared two further semiconductor units for the state in May 2026. Governments change and policies fade, but a fab, a runway, and a rail alignment do not un-happen.

The bear case, at full strength

First, the base rate is brutal. Greenfield cities on this subcontinent have a long record of infrastructure arriving and life declining to follow. The comparison essay on Amaravati and Naya Raipur exists because both built impressive bones and neither has organic economic gravity to match, and a 2017 national-press review of the wider corridor programme could fairly describe it then as a tale of delays. Dholera's own record includes an airport that has been about to open since roughly 2010 and activation-area population targets for 2020 that lapsed unmet. The plains have heard confident dates before.

Second, the population question is genuinely open. The 2011 census counted 2,779 people in Dholera village. The plan targets roughly 800,000 jobs at full development, a GREEN-sourced target on the official site with a sector breakdown behind it, but a target with three decades of runway is a direction, not a forecast. Nobody has yet demonstrated the thing cities are actually made of: thousands of unrelated households choosing to live there. The fab's reported worker housing, in the hundreds of apartments, is a seed, not proof.

Third, physical geography is a real constraint, stated plainly in the planning record: flat, low-lying coastal Bhal land with about a third of the developable area inside the Coastal Regulation Zone and drainage engineered around a canal system. This is manageable, it is being managed, and it is also permanently true, which narrows where value can concentrate and raises the cost of getting corners wrong.

Fourth, concentration risk. Strip the fab out of the story and the remaining committed private capital thins fast: solar at 300 MW commissioned against a 1,000 MW Phase-I, a data-centre agreement at memorandum stage, and a pipeline of announcements whose collective habit of becoming steel is exactly what is untested. The bull case is honest only if it admits it is substantially one company's balance sheet plus a sovereign's patience.

The three deciders

Here is what actually settles the argument, and none of it is a matter of opinion. Decider one: silicon. First wafers are targeted around December 2026 with commercial production reported for mid-2028 at mature nodes first. If commercial silicon ships on anything resembling that schedule, Dholera permanently exits the category of announced places. Decider two: aircraft. Operations are targeted for September or October 2026 on a finished 3,200-metre runway that has already taken its trial landing. An operating airport converts the site from a drive to a destination, and a second consecutive missed season would say something equally loud in the other direction. Decider three: households. Watch for the first thousand-resident wave that is not construction labour: fab operators, their families, the first school run. That is the moment a project becomes a town, and it has no press release.

All three are trackable, and I track them in public: the quarterly scorecard grades every public deadline, and the State of Dholera report rolls the whole picture up twice a year.

The base rate, with its actual numbers attached

The bear case above leans on a base rate, and a base rate asserted without numbers is just pessimism with good manners. So I went and built the numbers rather than gesturing at them: a comparative study of cities started from scratch since 1980, eighteen cases in all, seventeen comparison cities plus Dholera, each scored on eight viability dimensions with the weights and justifications published so any reader can disagree in detail. Here is the part of that board that bears on this question, ordered by the index score.

Read that column of outcomes and the honest bear case changes shape. The pattern in the data is not that new cities die. It is that they undershoot, frequently by an order of magnitude, and that the undershoot is comfortable in some cases and fatal in others. Sejong at 78.9 percent is a working city that simply grew slower than promised. Putrajaya at 35.7 percent is also a working city, just a third of the size on the drawing. Naya Raipur at 11.5 percent has real buildings and thin life around them. KAEC at 0.5 percent has a world-class port and essentially no city attached to it, and Lavasa at 0.9 percent went to insolvency.

So the disciplined bear case for Dholera is not that the plains stay empty. It is that Dholera will very probably come in far below the brochure, and the only interesting question is which of those five outcomes it ends up resembling. That is a much harder question than the one sellers and skeptics keep arguing about, and it is the one worth your attention. Note also what the study does not claim: Dholera is deliberately excluded from the comparative attainment medians, because its target year of 2042 has not arrived and scoring a city against a deadline it has not reached would be dishonest arithmetic. You can work through the full board yourself, including the cases I have not quoted here.

The one rule that sorted every case

When the scoring was finished, one split explained the ordering better than anything else I tested. The top seven cases in the index are every case with a substantially delivered anchor. The bottom five are every case whose anchor failed or was never real. Master-plan quality did not sort them. Budget size did not sort them. Neither did architecture, sustainability credentials, or the production values of the promotional film. What sorted them was whether the reason to be there actually turned up.

That single rule explains why Dholera's scorecard reads the way it does. It takes 5 out of 5 on demand anchor realism, because the fab is contracted, capitalised and physically rising rather than announced. It takes 2 out of 5 on anchor delivery, because no wafer has shipped. Dholera holds the joint highest anchor-realism score in the entire sample alongside one of its widest gaps between realism and delivery, and every serious argument in this essay lives inside that gap. It also carries a 0 on population traction, since no credible count of new residents exists and the twenty-two pre-existing villages are excluded by the study's own rule, which asks who arrived because of the city rather than who was already living on the land. The same distinction is recorded elsewhere in the dataset, where Amaravati's roughly 100,000 and Nusantara's 147,430 are both logged as largely pre-existing villagers.

The rule is also a filter for what to ignore, which is the more practical gift. Renderings, model apartments, drone footage over empty plots and phase maps tell you nothing about anchor delivery. Tool move-in at the fab tells you everything. That is why the three deciders above are stated in physical terms rather than financial ones, and why the 2030 scenarios essay is assembled only from items that already have a schedule attached rather than from ambitions that have a press release attached.

The GIFT City warning, which is closer to home than Amaravati

The comparison everyone reaches for is Amaravati, because a stalled capital makes a dramatic story and a satisfying one. The more instructive Indian case sits in the same state, just beyond Ahmedabad at Gandhinagar, and it is a success rather than a failure, which is exactly what makes it useful.

GIFT City scores 3.75 in the index, fourth of eighteen, comfortably ahead of Dholera. Announced in 2007, established on 10 April 2015, it is India's first operational greenfield smart city and its first International Financial Services Centre, with its own regulator, two international exchanges and the country's first international bullion exchange. The anchor was delivered on any reasonable reading: the study records over 1,000 entities and about USD 100 billion in banking assets, and the official-side count reported in June 2025 was around 939 registered entities, the sort of discrepancy that comes from different dates and definitions rather than from anybody exaggerating.

Now the number that should hold a Dholera investor's attention. GIFT City's attainment against its own jobs ambition is 2.8 percent: roughly 28,000 workers against a target of a million jobs by 2025, in a district that is expanding from 886 acres to over 3,300. The study's status line for it is blunt and I stand by it: a commuter district rather than a city. GIFT City is not a failure. It is the specific shape of partial success that Dholera should expect to be measured against, where the anchor works, an economy genuinely forms, and people drive in rather than move in.

That is the quiet cost buried inside Dholera's own connectivity story. Ahmedabad is now reported at roughly forty minutes to an hour away on the open expressway, which is precisely the distance at which commuting substitutes for settling rather than complementing it. Every connectivity upgrade that makes the fab easier to staff also makes it easier to staff without anybody moving house. It is not a reason to want worse roads. It is a reason to watch households rather than headcount, and to treat the jobs target as a weaker signal than it looks. I have set the two cities side by side in the GIFT City comparison, because the honest benchmark for Dholera in 2035 is probably GIFT City rather than Shenzhen.

Where I actually land

Asked to bet, I put it this way. The probability that Dholera becomes a functioning industrial node, fab running, cargo moving, tens of thousands working, is now high, because most of the money that outcome requires is spent or contractually committed. The probability that it becomes the million-person brochure city on the brochure schedule is much lower, because that outcome depends on the one force no government can schedule, which is other people's life decisions, compounding over decades. Success will arrive, if it arrives, the way dawn does on flat land: nothing, nothing, nothing, then unmistakable. The three deciders are the horizon to watch. Two of them report within eighteen months.

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