Where to set up a factory in Gujarat: Dholera or a proven node

Almost everything I publish here is written for someone deciding whether to buy a plot. This piece is not. Over the past year the messages that have made me think hardest came from a completely different reader: a plant head, a promoter, a project consultant, somebody carrying a board mandate to put a manufacturing unit somewhere in Gujarat inside a defined window, with Dholera sitting on a shortlist next to places that already have working factories on them. That reader is not asking whether land will appreciate. That reader is asking when the first product can leave the gate and what it costs to keep it leaving.

It is a better question than the investment one, and it is answered with different evidence. An investor can afford to be vague about commissioning dates because time is the asset. An occupier cannot, because for an occupier time is the liability. So this is a site-selection brief rather than a comparison of stories, and I am going to be explicit about the one thing that makes it lopsided: I have primary documents for Dholera's side of the table, and I do not have sourced figures for the established alternatives. No vendor counts, no land rates, no power tariffs, no absorption data for Sanand or Becharaji or Halol or anywhere else on your list. I will therefore argue their advantages structurally and print no numbers for them at all. Anyone who prints you those numbers without a source is doing you a disservice you will discover eighteen months later.

The occupier's question is not the investor's question

An investor buying land in a greenfield region is buying an option. The downside is opportunity cost and illiquidity, the upside is whatever the region becomes, and the whole thing can be carried quietly for a decade. An occupier is doing the opposite trade. You are committing capital equipment, a workforce, a supply contract and usually a customer's qualification schedule to a specific patch of ground, and every month of delay has a name and an amount attached to it. A site that is wonderful in 2032 and unusable in 2027 is not a good site for you. It is a good site for the person selling it.

That reframing changes which facts matter. Appreciation drivers, resale liquidity and the price per square yard debate all fall away. What replaces them is a shorter and much harder list: can I get a parcel of the size and tenure I need, can I get firm power at a knowable tariff, can I get water in and effluent out legally, can I move raw material and finished goods, can I staff and house the plant, who is my counterparty when something goes wrong, and can I expand without moving. Everything else is decoration. I wrote the plot-market version of this comparison in the piece on Dholera against Gujarat's proven nodes, and it reaches a different conclusion for a different reader, which is exactly as it should be.

What Dholera can answer with a document

Start with land, because parcel size is where the greenfield advantage is most concrete. The sanctioned plan provides for around 11,000 hectares of industrial area in total, with roughly 3,000 hectares of it in Phase I, and it explicitly contemplates parcels from 0.5 hectare up to 150 hectares. That upper bound is the interesting one. A 150 hectare contiguous parcel with services at the boundary is not something you assemble quickly inside an established cluster, because in an established cluster the land next to the good land already has a factory on it. These are plan figures published by the authority, so read them as design intent rather than as inventory.

Actual inventory is where the record gets unusually specific. The NICDC Delivery Monitoring Unit report to DPIIT dated 30 June 2026, which is the single most useful document about this place, records 48.31 sq km of land transferred to the delivery company, 14 plots totalling 545 acres already allotted of which 476 acres are industrial, and 1,043 acres of industrial land plus 1,031 acres of other land ready for allotment. Tata Chemicals is named in that report as the anchor industrial allottee. A published, dated land ledger with acreage available is a governance quality most industrial estates in this country simply do not offer, and for a site-selection team it means the first question, is there anything left, has a citable answer rather than a broker's answer.

Power next, because for most manufacturing it decides the shortlist. Inside the activation area the built stack recorded in the sanctioned plan and the monitoring report includes three 66 kV substations and roughly 115 km of underground power ducting. Underground ducting sounds cosmetic and is not: it is what removes the overhead-line failure mode that costs a process plant its batch. Alongside that, about 300 MW of the 1,000 MW sanctioned Phase I of the Dholera solar park is commissioned and generating, with the remaining 700 MW under development and targeted for March 2027, a target that has already been slowed by tariff disputes. I would treat the solar as a procurement opportunity and a corporate reporting benefit, not as your firm power answer. Green electrons and firm electrons are different products, and a fab or a foundry buys the second one first.

Water and effluent are where Dholera's design genuinely diverges from the norm. The recorded utilities include roughly 100 MLD of potable water available, a 50 MLD water treatment plant, a 10 ML reservoir, 82 km of water pipeline with smart metering and non-revenue water reported under 5 percent, a 10 MLD sewage treatment plant, a 20 MLD common effluent treatment plant, 81 km of recycled water pipeline and stormwater handled through a 6.5 km canal. The line I would underline for an occupier is the 20 MLD common effluent plant, because in most Indian industrial geographies the common effluent facility arrives years after the tenants do, and the gap between those two dates is where units get shut down. Here the plant is recorded as built before the tenants. Solid and hazardous waste has a similar shape: 25 TPD of segregation capacity, 30 TPD of bio-methanation, two 25 TPD incinerators and a 28 hectare integrated landfill. I have set out the constraint side of the water story, the semi-arid coast and the drainage question, in the water and climate essay, and an occupier should read the constraint and the engineered answer together rather than one at a time.

Movement is the fourth pillar and it is the one that changed most recently. The Ahmedabad to Dholera expressway, around 109 km of greenfield access-controlled road with a 120 km/h design speed, was reported inaugurated on 31 March 2026 and operational, cutting the run from roughly two hours to a reported 40 to 60 minutes. That is a delivered fact with a date, and I have argued its consequences in the expressway piece. Two more links are approved and unbuilt: the Ahmedabad to Dholera semi-high-speed rail was cleared by CCEA on 13 May 2026 at Rs 20,667 crore for about 134 km of double line, with completion targeted somewhere up to 2030-31, and the Dholera international airport has a 3,200 m Code 4E runway, was reported around 80 percent complete in July 2026 after a trial and calibration landing on 4 June 2026, and has operations targeted for September or October 2026. It is not open. Do not put a passenger or air-cargo assumption into a 2027 schedule on the strength of a target, and read how to read the airport dates before you do any capacity planning around it.

Finally the counterparty, which occupiers underrate and later care about more than anything else. Dholera is governed by a statutory stack rather than by a developer: DSIRDA is the planning authority under the Gujarat Special Investment Region Act of 2009, DICDL is the delivery company incorporated on 28 January 2016 with the state holding 51 percent through DSIRDA and the Centre 49 percent through the NICDC Trust, NICDC is the central nodal agency, and the monitoring unit files to DPIIT on dates you can check. Environmental clearance for the region is recorded as granted on 19 September 2014. The official sector list the region is planned around runs to ten, Defence, Aviation, Electronics and Semiconductors, High-Tech, Pharma and Biotech, Heavy Engineering, Auto and Ancillary, General Manufacturing, Agro and Food, and Metals, with renewables alongside, and the jobs model behind the plan is weighted towards electronics, which alone carries a potential of 87,300 jobs in the official breakdown. If your unit is nowhere on that list, you are not the buyer this plan was drawn for, and you will feel it in every approval conversation.

What a proven cluster answers that Dholera cannot

Now the other side, argued without a single figure, because I do not have sourced figures for it and I am not going to borrow somebody's marketing to fill the gap. An established industrial node in Gujarat gives an occupier things that no amount of new concrete replaces quickly. There is a vendor ecosystem: the machine shop, the fabricator, the plating unit, the packaging supplier, the man who can rewind your motor on a Sunday. There is a contractor market that has already built your kind of building nearby and priced it correctly. There is a labour market that already travels to that location every morning, and a town with housing, schools and hospitals for the staff you transfer. There is regulatory precedent, which matters more than people expect, because an authority that has approved twenty units in your category knows what your file looks like. There is often a ready shed you can rent while your own building goes up. And there is a resale market, which means a mistake is recoverable.

Those advantages compound in exactly the dimension an occupier is short of, which is time. Against them, Dholera offers scale, services engineered ahead of demand, a corridor position, an institutional landlord and the beginnings of an anchor cluster. The fab is the centre of that cluster: 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, designed for up to 50,000 wafers a month, with first silicon targeted around December 2026 and commercial production reported for mid-2028. A 66.16 hectare semiconductor special economic zone was notified around April 2026 with roughly 21,000 jobs projected. If you supply electronics, adjacency to that is a commercial reason rather than a sentiment. If you do not, it is somebody else's anchor, and you should say so in your board note. No chip has been produced there yet, which is the sentence I attach to every optimistic paragraph I write about this place, and the longer argument sits in the piece on why the fab is the whole ballgame.

The labour question is the one I would put in front of the board first, because it is the least fixable in the short run. Census 2011 recorded 2,779 people in Dholera village. There is no resident workforce of any scale, the activation area's original goal of roughly 120,000 residents and 80,000 jobs by 2020 lapsed unmet, and the residential city is a plan rather than a place. One anchor occupier has reportedly planned around 530 worker apartments on its own land, which tells you what the practical answer currently is: you house or you bus your people, and you budget for it. The expressway makes an Ahmedabad catchment reachable in a reported 40 to 60 minutes, which is a genuine change to that arithmetic, but a bus fleet is still a line item and a hostel is still a project.

The occupier's checklist, side by side

Here is the whole comparison compressed into the form a site-selection committee actually uses. The middle column is what I can support with a dated document. The right column is what remains yours to establish, at Dholera and equally at every alternative.

What the record does not tell you, and you should not let anyone pretend otherwise

The honest gaps in this dossier are as decision-relevant as the strengths. There is no reliable public price for land here, whether by allotment or in the secondary market, so any rate quoted to you is an assertion until the authority puts it on letterhead. I cannot source occupier tariffs for power, water or effluent, which means the operating cost comparison against an established node is a blank on my side of the table and must be filled by written quotes rather than by assumption. Toll rates on the expressway are not reliably sourced. The as-built lane count of that expressway is genuinely contested, described as six lanes in the monitoring report and as four expandable to eight in earlier documents, and the airport's passenger capacity appears variously as two, 2.8 and 3.5 million, so I use around two million initially and label the number disputed. The plan's own end year reads 2040 in some documents and 2042 in others.

Two gaps matter more than the rest for a factory. First, roughly a third of the developable area falls inside the Coastal Regulation Zone, which is a constraint on where you can build, not a footnote, and it has to be settled for your specific parcel with the authority rather than inferred from a map you were shown. Second, per-zone land use percentages are not sourced anywhere I can verify, so the question of what will eventually be built next to you, which is a real industrial risk in both directions, is answerable only from the sanctioned town planning scheme for that pocket. Ask for it by scheme number and final plot number. If the person selling cannot produce it, you have learned something useful for free.

The schedule question, and the discount I would apply

Dholera's delivery record splits cleanly, and an occupier should split it the same way. Delivered, with dates: trunk infrastructure recorded complete in the 22.5 sq km activation area in the monitoring report of 30 June 2026, the expressway reported open since 31 March 2026, around 300 MW of solar generating, environmental clearance dated 19 September 2014, and a land ledger that has actually moved with 545 acres allotted across 14 plots. Targeted, not delivered: the airport, which has been close to opening since roughly 2010 and missed a December 2025 date before the current September or October 2026 target; the remaining 700 MW of solar, now aimed at March 2027; the rail line, targeted up to 2030-31; and the fab's own commercial production, reported for mid-2028.

The practical rule I would give a project committee is to build the schedule using only the delivered column and to treat everything in the targeted column as upside that arrives when it arrives. If your business case needs the airport to work, your business case is not ready. If your business case works on an open expressway, a serviced parcel and grid power, the pending items simply improve it later. That is not pessimism, it is how you keep a commissioning date credible in front of a board. The starter zone itself, the 22.5 sq km activation area, is what all of this actually refers to, and I would not evaluate a parcel outside it on the same terms as one inside it.

Which occupier this genuinely suits

Dholera is a strong answer if you need a large contiguous parcel, if your process carries heavy power, water or effluent loads that an older estate cannot absorb, if you are building greenfield anyway so the absence of ready sheds costs you nothing, if you want an institutional counterparty and a published record rather than a private landlord, if you are in one of the ten planned sectors, and if your horizon comfortably absorbs a slipped year. It is a weak answer if you need an existing vendor within a short drive next quarter, if you want to start in a rented shed and grow, if your unit is labour-intensive and depends on workers who already live nearby, or if a six-month delay breaks a customer contract. Neither profile is smarter than the other. They are different constraints, and the failure I see most often is an occupier choosing on narrative when the constraint was sitting right there in the process sheet.

The sequence I would actually run

Write your non-negotiables as numbers before you look at any location: connected load, water in KLD, effluent volume and character, parcel in hectares, truck movements a day, headcount at full ramp. Send that identical sheet to the authority at every candidate location and ask for written answers with timelines, because the quality of the reply is itself a datapoint. For Dholera specifically, ask DICDL and DSIRDA for allotment terms and milestones, the connection timeline and tariff basis for power and water, effluent acceptance for your category, the zone and CRZ status of the exact parcel, and whether your sector is permitted there. Visit twice, and make one of those visits in the monsoon, because this is flat, low-lying coastal Bhal land and drainage is a documented planning constraint you should see rather than read about.

Then cost the workforce honestly, since housing and transport are real capital and real operating expense here in a way they are not in a node with a town attached. If, instead of the allotment route, someone offers you land in the secondary market, treat it as a different and riskier transaction: verify the scheme's GUJRERA registration where a project is being marketed, confirm clear title inside the notified SIR boundary with a thirty year encumbrance certificate and the chain of deeds, and check the N.A. and town planning scheme status of that specific final plot. The mechanics of the institutional allotment route, which I think is the better one for most occupiers, are set out in the industrial allotment guide linked below. Finally, price your schedule risk instead of assuming it away, and run the identical questionnaire at the established alternatives so you are comparing answers rather than brochures.

My verdict for an occupier

If I were choosing a site in Gujarat today with a manufacturing mandate, I would put Dholera on the shortlist and I would not put it there for the story. I would put it there because a monitoring unit publishes what has been built and what land is left, because the effluent plant exists before the tenants, because the parcel sizes go up to a scale established estates cannot offer, and because the counterparty is statutory. I would then be ruthless about the two things the documents do not cover, operating tariffs and people, and I would insist on written answers before the board note goes in. A proven node sells you certainty and charges for it. Dholera sells you scale and headroom and asks you to carry the schedule risk yourself. Both are rational purchases. The irrational one is buying either without asking the same ten questions of both.

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