Dholera vs Konza: the funding lesson from the Silicon Savannah

Kenya announced Konza Technopolis in 2008 and gave it the best nickname this genre of project has ever produced: the Silicon Savannah. The pitch was a technology city on open grassland outside Nairobi, a flagship of the country's Vision 2030 programme, with software and outsourcing tenants landing first and a city of more than 200,000 people forming around them by 2030. One year later, Gujarat passed the Special Investment Region Act, and Dholera acquired its legal skeleton. Two governments, twelve months apart, made the same species of promise: take a very large piece of nearly empty land, lay the services before the people arrive, and trust an industry anchor to conjure the town.

Eighteen years into Konza's life and seventeen into Dholera's, my Greenfield City Viability Index scores them 1.75 and 3.10 out of 5, fourteenth and eighth among the eighteen cases I studied. That gap deserves a precise explanation, because the lazy explanations do not survive contact with the record. It is not vision: Konza's plan is perfectly competent. It is not land assembly: both carry middling scores on that dimension. The difference that actually shows up in the numbers is the plumbing that moves money, and the Konza file is the cleanest demonstration I know of a rule that ought to hang framed above every greenfield programme office: these projects do not usually die of bad ideas. They die of interrupted cashflow.

What Kenya promised, and what stands on the grass

Start with what Konza actually is, because the internet's versions run to extremes and neither extreme survives the dataset. The index row reads like this. Announced in 2008, type: technology city. Population target: over 200,000 people by 2030. Residents counted against that target: none published, and the attainment column therefore records zero. The anchor verdict is blunt: a national data centre is live, a university took its first students in 2025, and a city has not formed around either of them. The status line reads: horizontal infrastructure, framed in decades.

None of that is a sneer, and Kenya's own reporting broadly matches it. The Vision 2030 delivery secretariat's project-status page for Konza, which I read in August 2026, records the National Data Centre and the Konza complex as complete and reports considerable progress on the horizontal works, the roads and the utility corridor. Kenyan press coverage this year describes the same place from the outside: basic infrastructure visible, new industries and institutions still limited, contracts signed without much clarity on why implementation crawls, more than a decade after construction began. Put the two accounts together and the finding writes itself, because the state's best case is finished trunk works and the independent case is finished trunk works with very few tenants on top of them, which means that neither account, official or critical, is describing a city.

So Konza is neither fraud nor ruin. It has a delivery authority, gazetted land, tarmac, ducts, a working data centre, a genuine university, and a real labour market within reach in Nairobi, which is why proximate metro gravity is one of the few dimensions where it manages a middling 3. What it does not have, eighteen years in, is the thing all of that was procured to produce: employers at scale, and people. On my rubric it scores 2 for demand anchor realism, because the anchor was always a hope about tenants rather than a contract with one, and 1 for anchor delivery, because what has landed is real but small against the promise. Population traction scores zero for the simplest reason a dimension can have: the target was a city, and there is no published count of anyone living in one.

The budget-line city

Now the part of the file that earns this essay: how Konza is paid for. The technopolis is a public project inside a national development programme, and its works are funded the way public works are funded almost everywhere: an allocation is proposed each budget year, argued against every other national need, trimmed or topped up, and released in tranches. On top of the exchequer line sit development-partner agreements, negotiated deal by deal, each with its own timetable and conditions. Nairobi's business press has tracked that budget line for well over a decade, and the coverage reads like a cardiogram: a generous voted allocation one year, a cut the next, a partner agreement announced, a supplementary item slotted in. I am deliberately not reprinting the shilling figures, because I cannot tie each one back to a primary budget document I have verified, and this site does not print numbers on those terms. The shape, though, is not in dispute, and the shape is the point.

What annual money does to a construction programme is mechanical, not moral. Contractors mobilise when a tranche lands and demobilise when it runs out, and remobilisation is slow and expensive. Procurement stalls between votes. Management spends its energy defending next year's line instead of spending this year's. Private investors watch the rhythm and time their own commitments to it, which mostly means waiting. A road built this way is eventually still a road. A city built this way is a permanent construction site, because a city is a network good: the parts only pay off together, and a funding model that delivers the parts one budget cycle at a time keeps pushing the payoff over the horizon. That is what the status line means by framed in decades.

Here is the detail I find genuinely instructive: Konza's governance continuity scores a respectable 3, because the delivery institution survived changes of government, which is more than several cases on the board can say. Financing durability scores 2, and that pairing is most of the diagnosis. The institution endured; the cashflow never did. A masterplan is a promise to spend steadily for twenty years, and an annual budget line is a promise to reconsider every twelve months. The two promises are not the same species, and a greenfield city lives exactly as long as its funding cadence says it can.

Row against row

The full eighteen-city board, with every dimension score and the reasoning behind each, sits on the interactive index, and the survey essay on all seventeen foreign cases tells the whole sample's story in one sitting. Here I only need two of its rows.

Two readings fall out of that table. The rows agree perfectly on one dimension, and it is the humbling one: population traction is zero on both sides, because neither project can yet show a counted resident against its promise. And the rows disagree hardest on exactly two dimensions, demand anchor realism and financing durability, which is another way of saying that the entire measured difference between these cities is a contracted anchor and a redesigned cashflow. Everything else is nearly a tie.

The counter-design: money approved as a block

Dholera's answer to the budget-line problem was structural, and it predates most of the delivery now visible on the ground. In January 2016 the state and the centre incorporated a joint delivery company, DICDL, with Gujarat holding 51 percent and the centre 49 percent through the national industrial corridor programme. Then, instead of feeding the project an annual line, the centre approved the starter zone's works as sanctioned packages: five of them, worth Rs 2,784.83 crore, with matching equity of the same amount recorded as released. Those figures are not press-release numbers; they sit in the NICDC Delivery Monitoring Unit's report to DPIIT dated 30 June 2026, which also records the result. Trunk infrastructure in the 22.5 sq km activation area is complete, 48.31 sq km of land stands transferred to the company, and 14 plots covering 545 acres have been allotted, with Tata Chemicals named as the anchor industrial allottee.

Spell the design difference out, because it is the thesis of this whole comparison. A sanctioned package is money approved once, for a defined scope, at a defined size, and then drawn down against progress rather than re-argued every budget season. The activation area was drawn small enough to be finishable, 22.5 sq km out of a planned envelope of about 920, so the funded block could produce one complete, serviced, allottable unit instead of a thin layer of progress smeared across an enormous map. Konza must re-win its own existence every fiscal year. Dholera's core, once approved, mostly had to be built. I keep the full public-money map, packages, equity, budget lines and what each implies, in the essay on who actually funds Dholera.

The anchor is the other half of the counter-design, and it obeys the same logic. Konza's anchor was a category of tenant that the city hoped to attract. Dholera's is a named company spending mostly its own capital on its own schedule: the Tata fab, approved by the union cabinet on 29 February 2024 at Rs 91,000 crore, reported past 50 percent civil work by mid-2026, with first silicon targeted for December 2026 and commercial production reported for mid-2028. Half of the eligible cost flows back as fiscal support under the India Semiconductor Mission, but the build itself is a corporate project with corporate urgency, not a line item hoping to survive a vote. The 300 MW of commissioned solar and the expressway opened on 31 March 2026 belong to the same pattern: each was one entity's funded project with a completion incentive, which is why they exist and the outer phases do not yet. Why the fab dominates every serious Dholera forecast has its own essay.

Where Dholera still rhymes with Konza

Now the discipline, because this comparison flatters Dholera and I distrust comparisons that flatter. The zeros match. Dholera's population traction is exactly Konza's, and it is not obvious to a resident of either place that one lives in the stronger project. Dholera has also already lived through one full Konza decade of its own: the activation area's original target was roughly 120,000 residents and 80,000 jobs by 2020, and it lapsed unmet. The airport has been slipping since around 2010 and is currently targeted, never scheduled, for September or October 2026. The pattern of promised dates and their fates is documented in the timeline of slipped dates, and it looks more Kenyan than the marketing would like.

The sharper rhyme is spatial. The funded block covers 22.5 sq km, and the sanctioned plan runs to about 422 sq km of urban-developable land across six town planning schemes phased out to the early 2040s. Everything beyond the activation area currently lives on exactly the diet Konza lives on: plans, phases and future approvals that do not yet exist as sanctioned money. If the later tranches arrive as blocks on schedule, the model scales. If they arrive late, or arrive as annual dribbles, the outer phases will stretch precisely the way Konza's phases stretched, while the finished core waits. And the index's third finding refuses to take sides here: infrastructure is not traction. Completed trunk works have not yet moved a single verifiable new resident in either city. The difference is that Dholera's design gives traction a mechanism to arrive through, a fab with a payroll, and Konza is still waiting for its equivalent.

The verdict, and what would change it

Set side by side, the two files read like a controlled experiment on one variable. Konza at eighteen is what a sound idea looks like on annually argued money: alive, slow, real, and permanently one budget cycle from stalling. Dholera at seventeen is what the same idea looks like after somebody redesigned the cashflow: a finishable core funded as approved blocks, an anchor spending corporate capital against its own deadlines, and delivery visibly clustering in the 2024 to 2026 window. Konza never got that redesign. Dholera got it in 2016, seven years in, and the board scores the consequence at 1.35 points of daylight. I rank the design, not the render, and the design difference is genuine.

What would change my mind runs in both directions. For Konza, the flip would be an anchor with its own balance sheet committing at scale, and any published resident count at all moving the zero. For Dholera, the tests are dated: first silicon at the fab, targeted December 2026; airport operations, targeted September or October 2026 against a long slip history; the remaining 700 MW of the solar park's first phase, targeted March 2027; and, quietest but most Konza-shaped of all, whether money for the phases beyond the activation area arrives as sanctioned blocks or as annual dribbles. Watch that last one even though nobody markets it, because this entire file says it is the one that decides, and because it is the only item on the list that a press release will never announce.

None of this is a buy signal, and a scoreboard is not due diligence. If this comparison is feeding a plot decision anyway, the boring rule of this site still applies in full: only GUJRERA-registered schemes, only title a lawyer has checked end to end, only land that verifiably sits inside the SIR. Konza's lesson, if the word tragedy even applies to something still moving, is that nobody rebuilt its funding machine when the drift became obvious. Dholera's redesigned machine has produced trunk roads, an expressway, an anchor, and zero counted residents so far. Both of those facts fit in one head, and the reader who holds both at once is ahead of nearly everyone selling either version of this story.

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