What the Lavasa failure actually teaches Dholera

Whenever I argue in public that Dholera might work, somebody reaches for Lavasa. It is the right instinct and usually the wrong argument, because the person reaching for it rarely knows what happened there beyond the photographs of an empty promenade with Italian styling. So I went and scored it properly instead. In the comparative study behind this site, an original index that rates eighteen built-from-scratch cities on eight weighted viability dimensions, Lavasa lands at 0.95 out of 5. Only NEOM's The Line, at 0.90, scores lower. Dholera lands at 3.10, eighth on the board. That is a wide gap, and no reader should accept it from me on trust. This piece is the audit behind it: the four mechanisms that actually killed Lavasa, then Dholera tested against every one of them, including the places where the test does not go Dholera's way.

One thing before the diagnosis, because it sets the tone. Lavasa deserves better than the smirk it usually gets. Announced in 2006, it was a serious attempt to build a hill city in India with private money and real design ambition, and a great many people, farmers and contractors and lenders and buyers, are still living with how it ended. I am not going to make jokes about it, and I am going to stay neutral about every company that has passed through its story, including the one whose rescue bid is still in litigation. A postmortem is only worth reading if it is honest about mechanism. Mockery is what people do instead of diagnosis.

What the index actually says about Lavasa

Scores first, because they discipline the argument. On the eight dimensions, Lavasa scores 1 on demand anchor realism, 1 on anchor delivery, 1 on connectivity integration, 3 on proximate metro gravity, 0 on financing durability, 1 on land assembly durability, 1 on governance continuity, and 0 on population traction. Weighted, that is 0.95. The single most striking number in that row is the zero on financing durability, because it is the only zero on that dimension anywhere in the sample of eighteen. Not the lowest score. The only one.

The outcome numbers are worse than the score suggests. Against a target of 200,000 to 250,000 residents, Lavasa had under 2,000 people as of 2026, an attainment ratio of 0.9 percent. For scale, the same ratio reads 78.9 percent at Sejong, 75.3 percent at Songdo, 35.7 percent at Putrajaya, 11.5 percent at Naya Raipur, and 0.5 percent at Saudi Arabia's KAEC. Lavasa sits in that last company. The project went into insolvency with Rs 6,642 crore of admitted claims, the rescue bid that emerged from that process remains in litigation, roughly 600 hectares of its land purchases were contested, and the environment ministry issued a stop-work order in 2010 that froze construction at exactly the moment momentum mattered most.

Even the size of the thing is disputed. The concession is reported anywhere from 32 to 100 sq km depending on the document you open, and the study I built this on carries that conflict openly rather than picking the convenient number. I mention it because it is a small tell about the whole category: when a project's basic dimensions cannot be pinned down years after the fact, the paperwork was never as settled as the renderings implied. The full row, the rubric and every justification are on the interactive index board, and the paper itself is free on the research shelf, so anyone who thinks a score is wrong can go and argue with it directly.

Mechanism one: the anchor was a preference, not a payroll

The first finding in the study is blunt and it does more work than everything else combined: the top seven cases on the board are every case with a substantially delivered anchor, and the bottom five are every case whose anchor failed or was never real. Lavasa's anchor was tourism and second homes. That is a preference, not a payroll. Nobody's livelihood required the city to exist. When sentiment cooled and the regulator intervened, there was no weekday reason for anyone to be there, which is why demand anchor realism scores 1 and anchor delivery scores 1.

This is where Dholera is genuinely, structurally different, and I want to be precise about how much of that difference is banked. Dholera scores 5 on demand anchor realism, the highest score on that dimension, because its reason to exist is industrial employment backed by statute and contract rather than by taste. The Rs 91,000 crore Tata Electronics and PSMC fabrication plant was approved by the Union Cabinet on 29 February 2024 under the India Semiconductor Mission, with a fiscal support agreement signed on 5 March 2025 citing Rs 91,526 crore and the Mission covering half the eligible cost. Around 300 MW of the sanctioned 1,000 MW solar phase is commissioned and generating. The NICDC monitoring unit's report to DPIIT dated 30 June 2026 records 545 acres across 14 plots already allotted, 476 of them industrial, with Tata Chemicals named as the anchor industrial allottee. None of that is a brochure. All of it is a document with a date.

Now the part that is not banked. Dholera scores only 2 on anchor delivery, and that number is the honest one. No chip has been produced. Civil work at the fab was reported past the halfway mark by mid-2026 with cleanroom fit-out underway, first silicon is targeted around December 2026, and commercial production is reported for mid-2028. Targets are not events. A reader who takes the 5 on demand realism and forgets the 2 on delivery has learned nothing from Lavasa at all, because Lavasa's fatal gap was also the gap between an intended anchor and an operating one. I have argued at length that the fab is the whole ballgame, and the index is simply a more disciplined way of saying the same thing.

Mechanism two: the buyers were the bank

The second finding is the one with the sharpest teeth. Four of the five weakest cases on the board were financed by selling the city to buyers before an economy existed. Lavasa reached insolvency. Forest City in Malaysia is roughly 1 to 1.5 percent occupied, with a little over 10,000 residents reported by the company in 2024 against a target of 700,000 by 2035, before it pivoted to a finance-zone story in that same year. NEOM's The Line is halted past 2030, with over 50 billion US dollars reported spent against a 500 billion dollar envelope and its population target cut in stages from 9 million toward roughly 100,000 by 2030. Yachay's delivery company was liquidated in 2021, leaving about 2,000 students on 4,462 hectares expropriated from over 100 owners.

Presale financing has a specific failure geometry. The developer needs sales to fund construction, construction to justify sales, and confidence to hold both together. Any interruption, a court order, a credit squeeze, a bad news cycle, breaks the loop from the middle, and the people holding the risk turn out to be the buyers who paid first. Lavasa's stop-work order in 2010 is a textbook example of exactly that break.

Dholera's financing is a different animal, and this is the difference I would defend hardest. It scores 4 on financing durability because the money is statutory and sovereign rather than dependent on the next sale. The SIR sits under the Gujarat Special Investment Region Act of 2009. DSIRDA is the statutory planning authority and DICDL the delivery company, incorporated on 28 January 2016 with Gujarat holding 51 percent through DSIRDA and the Centre 49 percent through the NICDC Trust. The same monitoring report records Government of India approved activation packages worth Rs 2,784.83 crore across five packages, with matching equity of Rs 2,784.83 crore released, and 48.31 sq km of land transferred to DICDL. The semi-high-speed rail line was approved by CCEA on 13 May 2026 at Rs 20,667 crore. Whatever else is uncertain here, the city is not waiting on plot sales to build its trunk.

I want to block the lazy version of that comfort, though, because the dataset blocks it too. Yachay was built with over 1 billion US dollars of public money and still scores 1 on financing durability, and its developer was liquidated in 2021 anyway. Public money is not automatically durable money. What earns a 4 is money that is legislated, budgeted in tranches, matched by equity and audited by a monitoring unit that files reports on dates you can check. Sovereign funding removes one specific failure mode, the one that killed Lavasa. It does not remove the others.

Mechanism three: the land, and who the regulator was working for

Lavasa scores 1 on land assembly durability, and the reason is not complicated. Roughly 600 hectares of its purchases were contested, and in 2010 the environment ministry ordered work to stop. Read that as a structural fact rather than a scandal: the project's land and clearances were not settled enough to survive scrutiny, and the state, when it finally arrived, arrived as an obstacle. A private city that needs the government to stay uninterested is running an unhedged risk, and Lavasa ran it into the ground.

Dholera scores 3 here, which is a mediocre number and I am not going to dress it up. The land record has real scars. In 2015 the Gujarat High Court stayed acquisition in the SIR after farmer petitions. In 2017 Business Standard reported that only around 290 of the 900 plus square kilometres had then been secured, and the same year the wider corridor programme was described in the business press as "a tale of abandonments and delays". Roughly a third of the developable land sits inside the Coastal Regulation Zone, and the Bhal is flat, low-lying and documented in the planners' own material as a drainage constraint.

What earns the 3 rather than a 1 is that the machinery kept working through all of it. Environmental clearance was granted on 19 September 2014 and has held. The delivery company was incorporated in 2016 with the litigation still live. The monitoring report of 30 June 2026 records 48.31 sq km transferred to DICDL, with 1,043 acres of industrial land and 1,031 acres of other land ready for allotment. Land here moves through sanctioned town planning schemes rather than through a private developer buying farms one at a time, which is a slower and more contestable process at the front end and a far more defensible one at the back end. The blunt difference is this: at Lavasa the regulator was the thing that stopped the city, and at Dholera the regulator is the thing building it. That is a genuine structural advantage for delivery. It is not automatically a comfort to a farmer or a buyer, and the 2015 stay is proof that courts here still bite.

Mechanism four: when the promoter broke, nobody was left to carry it

Lavasa scores 1 on governance continuity, and this is the mechanism people least appreciate. The city's delivery institution was a company. When the company went into insolvency, the plan had no successor. There was no statutory authority holding the master plan, no budget line that renewed itself, no monitoring unit obliged to report on it. Everything that survives at Lavasa today survives inside a court process. A city whose only institution can go bankrupt is a city with a single point of failure, and the index punishes that harshly for good reason.

Dholera scores 4. The stack is DSIRDA as statutory planning authority under a 2009 Act, DICDL as the delivery company jointly owned by the state and the Centre, NICDC as the central nodal agency, and a Delivery Monitoring Unit filing to DPIIT. That structure has now absorbed a High Court stay in 2015, the quiet lapse of the activation area's original target of roughly 120,000 residents and 80,000 jobs by 2020, and the collapse of the Vedanta and Foxconn semiconductor venture when Foxconn withdrew on 10 July 2023. It kept going through all three. Seventeen years of uninterrupted institutional attention is the least glamorous asset Dholera owns and probably the most valuable.

And here is the correction I would make to the usual Indian comparison. If you want the case that should genuinely worry a Dholera buyer, it is not Lavasa, whose failure mode Dholera's structure largely closes. It is Amaravati, which scores 1 on governance continuity, sat frozen for five years when the politics changed, and stands at 2.9 percent attainment against a target of 3.5 million by 2050. Amaravati proves that a statutory anchor and public money do not save you if the mandate itself becomes reversible. I have set the three Indian greenfields side by side in the greenfield report card, and that comparison is the one I would run before this one.

Laid out dimension by dimension, the two rows look like this.

Where Dholera is not different at all

Look at the last two rows of that table again, because they are the honest ones. On proximate metro gravity, Lavasa and Dholera score exactly the same 3. Being within reach of a working labour market did not save Lavasa, and it will not save Dholera by itself. On population traction they also score exactly the same 0, and Dholera is one of nine cases out of eighteen sitting at zero on that dimension, in the company of KAEC, New Clark City, Konza, Forest City, Yachay and NEOM's Line. That is not a flattering neighbourhood.

The two zeros do mean different things, and the difference is stage rather than reassurance. Lavasa's zero is measured: under 2,000 people against a 200,000 to 250,000 target, an outcome that has already happened. Dholera's zero is unmeasurable: there is no credible count of new residents, the 22 pre-existing villages are excluded by the study's own rule so that inherited population cannot be counted as traction, and Census 2011 put Dholera village itself at 2,779 people. Dholera is also excluded from the study's comparative ratio medians for a specific reason: its target year of 2042 has not arrived. A zero at year seventeen of a thirty-three year plan is not the same claim as a zero after insolvency. It is still a zero.

And there is one measured miss on the Dholera record that belongs in this section rather than in the flattering one. The activation area's original goal of roughly 120,000 residents and 80,000 jobs by 2020 lapsed unmet. That is Dholera's own small Lavasa-shaped datapoint, and anyone quoting the 2042 population promise to me has to explain the 2020 one first. The airport, similarly, has been about to open since around 2010; 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. Those are real milestones and they are also, once again, targets.

The final similarity is the uncomfortable one, and it is about you rather than about the city. Dholera's financing is sovereign, but an individual buying a plot today on the expectation of a future economy is running a small private version of the presale bet that destroyed the bottom of this index. The city is not depending on that person's money. That person is depending on the city. If it slips a decade, the state absorbs delay and the buyer absorbs cost. I keep the full ledger in the risks nobody lists, and the practical version of the same warning lives in the safety essay. Whatever the index says about the city, the personal protections do not change: verify the scheme's GUJRERA registration, insist on clear title inside the notified SIR boundary, and treat any promise of assured returns as marketing rather than fact. No reliable public per-unit price series exists here, so anyone quoting one is quoting themselves.

Am I being too generous to Dholera?

Fair question, and the study anticipates it. 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. Even at the pessimistic end of that band, the distance to Lavasa's 0.95 is not close, which is the honest headline of this whole comparison.

The flattering adjustment is available too, and I decline to lean on it. Excluding population traction entirely and renormalising, Dholera scores 3.65, which is exactly Songdo's total. But removing the dimension that measures whether anyone actually lives there is precisely the adjustment a promoter would make, so I quote it and leave it alone. The study's own acknowledged limitation is that the scoring is single-rater, mine, which is why every score and justification is published and why cases like Naypyidaw, Rawabi and Duqm were considered and excluded with reasons stated. If you think Dholera's 4 on financing or its 3 on land is wrong, the rubric is public and the argument is winnable. That openness is the difference between a study and a sales deck.

What would flip each verdict

Diagnosis is only useful if it produces things to watch, so here is my list. Dholera starts genuinely resembling Lavasa if first silicon slips repeatedly past December 2026 without a stated engineering reason, then drags past the mid-2028 commercial target; if allotted acreage in the activation area stays frozen near the 545 acres recorded on 30 June 2026 while land continues to be marketed; if the funding conversation shifts from statutory packages to a dependence on land sales; or if a change of government lets the mandate go soft the way Amaravati's did. Any one of those would knock the anchor delivery, financing or governance scores down, and those three carry 0.35 of the total weight between them.

Dholera leaves this comparison behind on three events, in order of how much they would tell me: a wafer actually produced and shipped, an airport actually carrying scheduled passengers, and a published, credible count of new residents living there. The third is the one nobody in the sales chain ever mentions, and it is the one the index says decides everything. I have written the longer version of that argument in the piece on whether Dholera can succeed, and nothing in Lavasa's record changes the conclusion, it only sharpens the test.

The lesson, stated plainly

Lavasa does not teach that new cities fail. Shenzhen, Sejong and Putrajaya are on the same board with delivered anchors and real populations. Lavasa teaches something narrower and more useful: a city needs a reason to exist that survives a bad year, money that does not depend on the next buyer, land and clearances that can survive a regulator taking an interest, and an institution that outlives whoever started it. On those four tests, Dholera passes three today on the documentary record and has explicitly not been asked the fourth question, the one about people, because its own clock has not run out. That is a real distinction and it is not a guarantee. Treat it as the reason to keep watching rather than the reason to stop checking, and keep every date you are shown written in pencil.

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