Build it and they will come: how China builds so fast

In short: China builds so fast because it builds supply-push: the state lays rail, stations and metros first, on land it owns, pays for them with the land value they create, and keeps the plan going for decades. That gave it 50,000 km of high-speed rail by 2025 and urban rail in 58 cities. It works where a place has a reason to exist, is connected from day one and outlives its leaders. Indonesia copied the building, not the connecting or the continuing.

• What did I see in China that made me ask this?
• Demand-pull or supply-push: what are the two ways to build a city?
• How does “build it and they will come” work in China?
  • Where did “they” actually come?
  • Where didn’t they come, and who pays?
• Is Australia turning towards the Chinese model?
• Why hasn’t it worked as well in Indonesia (yet)?
  • Why does Indonesia wait for the jam, and then still not build?
  • What did Xi tell Jokowi?
  • Did “they” come to the Trans-Java Toll Road?
  • Why are Kertajati, Whoosh and IKN struggling?
• Do you need a one-party state to build like this?
• So, how does China build so fast?
• References

What did I see in China that made me ask this?

How do you build 2,862 km of high-speed rail in a single year? That is what China opened in 2025: about twenty times the whole length of Whoosh, Indonesia’s only high-speed line (142.3 km).¹² I have been asking myself versions of that question on every trip to China, and I have made at least one a year since 2023.

So far I have been to 14 Chinese jurisdictions: the four municipalities of Beijing, Tianjin, Shanghai and Chongqing; Jiangsu, Heilongjiang, Hunan, Shaanxi, Sichuan, Guangdong, Fujian and Taiwan; and the two special administrative regions (SARs), Hong Kong and Macau.

The impression repeats wherever I go. A high-speed rail network that reaches cities I had never heard of. Train stations the size of airports (the first few times, I caught myself looking for the check-in counter 😅). Metro maps as tangled as Singapore’s, in city after city. And skyscraper skylines not only in Shanghai and Beijing, but in second- and third-tier cities whose names I had to practise before I could say them to a taxi driver.

The numbers back the impression up. At the end of 2025, China had 165,000 km of railway, 50,000 km of it high-speed rail (HSR).¹ Fifty-eight mainland cities had urban rail, 13,068 km of it in total.³ For scale, Singapore, the city most of Southeast Asia treats as the gold standard for public transport, has 271.3 km of MRT and LRT.⁴ Shanghai’s metro alone runs 906 km.⁵ Chengdu, a city many people outside China couldn’t place on a map, has 870 km of urban rail.³ That is more than three Singapores in a single city.

My stake: I spent my first 25-plus years in Indonesia, and I have lived in Australia for almost two. Both countries build. Both argue about it endlessly. Neither builds like this.

So this post asks two things. How does China build so fast? And why can’t Indonesia, which has tried hard to copy it, simply do the same?

Demand-pull or supply-push: what are the two ways to build a city?

Most countries build demand-pull: they build in response to proof. They wait for real, measured signals (a jammed road, a full train, a lost opportunity) and then build to fix that bottleneck. China, in its big projects, builds supply-push: planners build ahead of demand, from a long-term forecast, a master plan or a vision, and expect the people and the businesses to follow. Build it, and they will come.

Demand-pull has real strengths. Little capital sits idle, because every rupiah goes where the need is already proven. The commercial risk is lower too, because the users already exist: you know the train will be full, since the road beside it is full today.

Its weakness is the price of waiting. By the time demand is undeniable, the city has grown around the problem, so you build inside the congestion: you tunnel under it, you buy land in a built-up city, and you disrupt everyone while you work. Jakarta is the example I know best. Its first MRT line, Phase 1 from Lebak Bulus to Bundaran HI, opened on 24 March 2019: 15.7 km for about Rp16 trn.⁶ That is about Rp1 trn per kilometre (my own calculation). Jakarta built it when the traffic had become unbearable, which is exactly the point at which building is hardest.

Supply-push has the opposite profile. Its strength is scale: coordinated, multi-decade projects that shape where a region grows, which industries settle and how much capacity it will have. Building in bulk, on a long plan, also lowers the cost per unit and prevents the chronic under-capacity that demand-pull cities live with. Its weakness is the bet. If the plan is wrong, you have built idle capacity: stations and towers that cost money to run and maintain every year, plus interest if you borrowed to build them.

The table below puts the two side by side.

	Demand-pull	Supply-push
Trigger	A proven need: congestion, full trains, lost business	A forecast, a master plan or a vision
Who decides, and how	Reacts to real-time signals and actual bottlenecks, one project at a time	Planners and the state decide ahead, for decades, often many projects together
Strengths	Little idle capital; low commercial risk, because users already exist	Scale and long-term capacity; lower cost per unit; no chronic under-capacity
Risks	Builds late and expensively, inside congestion; the city outgrows its network	Idle capacity, running losses and debt if nobody comes
Examples	Jakarta MRT, Sydney Metro, most of the world	Shenzhen, Shanghai Pudong, Perth’s Yanchep line

Neither is “the right way”. Demand-pull is careful and late. Supply-push is early and risky. The question is what makes the risk pay off.

How does “build it and they will come” work in China?

China’s version of supply-push runs on land. In Chinese cities the land belongs to the state, so a local government can do something an Indonesian or Australian city cannot: build a railway, a road or a metro line to empty land it already owns, and keep the value the new line creates.

The loop goes like this. A local government borrows, usually through a local government financing vehicle (LGFV), a company it owns that can raise money off its books. It builds the infrastructure. The land near the new line becomes valuable. The government sells the right to use that land (land-use rights) to developers, and the revenue repays the debt and pays for the next project. At its peak in 2021, land-use-right sales brought China’s governments RMB 8,705.1 bn in a single year.⁷

The flowchart below draws that loop, including the branch where it breaks.

flowchart TD
  A["State owns the land"] --> B["Government builds rail, roads and metro,<br/>borrowing through local financing vehicles"]
  B --> C["Land near the new line rises in value"]
  C --> D["Government sells land-use rights to developers"]
  D --> E["Revenue repays the debt<br/>and funds the next project"]
  E --> B
  C -. "nobody comes: no jobs, no reason" .-> F["Land does not sell, the debt stays<br/>(Yujiapu)"]

Two more things keep the loop turning. The first is the plan: China works in Five-Year Plans, and its cities draw master plans that run for decades, so a line drawn on a map in one decade is usually still being built in the next. The second is the incentive: local officials have long been judged on the growth their area produces, and nothing produces growth on paper faster than a construction site.

Does it work? Sometimes spectacularly. Sometimes not at all.

Where did “they” actually come?

Shenzhen is the most famous case. In 1979 it had 314,100 residents,⁸ and a year later it became a special economic zone (SEZ).⁹ By the end of 2025, Shenzhen had 18,248,500 residents and a GDP of RMB 3,873.18 bn.¹⁰ That is 58 times the population of 1979. A fair concession: Shenzhen wasn’t built by roads alone. It got market reforms that the rest of China didn’t have yet, and it sits right next to Hong Kong, which supplied capital, know-how and a door to the world. The infrastructure made the growth possible; the reforms and the neighbour gave it a reason.

Shanghai’s Pudong is the case I have seen with my own eyes. The opening up of Pudong was announced on 18 April 1990, when the east bank of the Huangpu was mostly farmland and warehouses.¹¹ In 2025 Pudong’s GDP was RMB 1,876.91 bn: about a third of Shanghai’s, which was RMB 5,670.87 bn (33.1%, my own calculation).¹²

Shanghai Hongqiao is the hub done right. On the city’s western edge, Shanghai put an airport, a high-speed rail station and the metro in one place. The rail station opened in 2010.¹³ The airport sits about 13 km from the city centre in a straight line, and Metro Line 2 links it to the centre.¹³¹⁴ A business district then grew around the hub: the offices came to where the trains and planes already were.

Beijing’s Chaoyang CBD was planned and launched in 1993, in that year’s Beijing master plan.¹⁵ In 2025 Chaoyang counted more than 140 buildings that each paid over RMB 100m in tax, and 125 regional headquarters of multinationals.¹⁵

What the four have in common: each had an economic reason to exist (a border with Hong Kong, a port and a financial centre, a national transport hub, a capital’s business district), each was connected from day one, and each plan kept going across several sets of leaders.

Where didn’t they come, and who pays?

Yujiapu, in Tianjin, is the cautionary tale. It was built by a local government financing vehicle (LGFV) and billed as the world’s largest financial district, “China’s Manhattan”, with at least RMB 200 bn (US$32 bn) invested.¹⁶ In 2014 the South China Morning Post (SCMP) was already calling its towers “ghost buildings”.¹⁷ Businesses trickled in by 2016,¹⁸ but around 2019 about 80% of its office space was still empty and several towers had stalled.¹⁶

Yujiapu had the towers, the roads and even the financing. What it didn’t have was a reason for a bank to leave Beijing or Shanghai and move there. A ghost town is a master plan that forgot to give anyone a reason.

The bill for the whole land-finance loop is now arriving. Land-use-right sales fell from RMB 8,705.1 bn in 2021 to RMB 4,151.8 bn in 2025, the fourth fall in a row: roughly half gone (−52.3%, my own calculation).⁷ Meanwhile the International Monetary Fund (IMF) estimates LGFV debt at RMB 71.4 trn in 2025, 50.9% of China’s GDP. Counting that debt, China’s “augmented” government debt reaches 126.6% of GDP, and the IMF projects 153.7% by 2030.¹⁹

This is the concession that matters. Supply-push is borrowing against the future, and borrowing is only as good as what it builds. I wrote about that difference in Is debt bad?: debt that builds a productive asset can pay for itself, debt that builds an empty tower cannot. China’s loop worked brilliantly while land prices kept rising, and it is now paying for every project where they didn’t.

Is Australia turning towards the Chinese model?

Partly, yes, and it is a sign that supply-push isn’t only for one-party states. Australia’s tradition is firmly demand-pull, though. Sydney Metro North West opened on 26 May 2019, into a city that had been congested for years.²⁰ Sydney’s suburban trains are double-deckers, which tells you how long the city has been squeezing more people into the network it has.

Melbourne’s airport line shows the cost of waiting. In 2018 the Melbourne Airport Rail link was estimated at A$8 to 13 bn. The 2024 Victorian budget delayed it by at least four years, to 2033 at the earliest.²¹

Housing is the most visible demand-pull failure. The national Housing Accord aims for 1.2 million new homes in the five years from mid-2024 to June 2029. In April 2026 the National Housing Supply and Affordability Council (NHSAC) estimated that only 980,000 would be built in that period, about 220,000 short (my own subtraction).²²

Now the change. Perth’s METRONET extended the northern line to Yanchep, with stations at Alkimos, Eglinton and Yanchep. It opened on 14 July 2024.²³ And crucially, those stations were built before the suburbs around them. The satellite images below compare each station in 2024 and 2026.

They came. In two years, new street grids and house lots have grown around Eglinton and Yanchep, and new estates are rising near Alkimos.

Sydney’s new airport is the bigger bet, and it comes with a warning. Western Sydney International Airport opened to passengers on 25 October 2026.²⁴ Around it, the plan is a whole new city, Bradfield. But its metro line opens only in 2027, so for now passengers ride a free interim bus to the train network.²⁴ An airport that opens before its rail link is the classic way a supply-push project starts life looking empty. Australia’s saving grace is that the line is already being built and is due next year, not merely promised.

Why hasn’t it worked as well in Indonesia (yet)?

Because Indonesia copied the “build” and skipped the rest. For most of its history it built in an extreme form of demand-pull, and when it switched to supply-push under Jokowi, it built the big things without the connections, without a way to capture the land value, and without a guarantee that the next government would keep going.

Why does Indonesia wait for the jam, and then still not build?

Indonesia’s demand-pull is extreme: even where the demand is undeniable, the proper infrastructure often still doesn’t arrive. Proper rapid transit exists in only two Indonesian cities. Jakarta has its MRT (Phase 1 opened on 24 March 2019, 15.7 km),⁶ its LRT and its KRL commuter trains. Palembang has an LRT.

Then look at Indonesia’s next three big cities. Surabaya, Bandung and Medan have no metro and no LRT, and until recently not even a proper bus rapid transit (BRT) system. Construction of the Bandung Basin BRT finally started in July 2026: about Rp1.3 trn for 18 routes. Medan’s Mebidang BRT is due to start service in 2027, and Surabaya is still at the detailed-design stage.²⁵

Three of Indonesia’s biggest cities, millions of people each, and the first proper mass transit for any of them is a bus, starting this year. That is not demand-pull. That is demand, ignored.

What did Xi tell Jokowi?

On 18 November 2014, President Jokowi told participants of Lemhannas, the national resilience institute, that China’s President Xi Jinping had given him three keys to building a developed country.²⁶²⁷ In the official transcript, they were:

1. A united party.
2. A big, long-term plan.
3. Connectivity now: “Kejar infrastrukturnya”, chase the infrastructure: connectivity between cities and between islands, and “jangan ditunda-tunda”, don’t keep putting it off.²⁶

I’d call the three keys “planned and strategic development”, and the third is the one the Jokowi government took most to heart: toll roads, airports, ports, dams and a high-speed railway.

Did “they” come to the Trans-Java Toll Road?

Mostly yes, slowly and unevenly, and it is the best Indonesian example of supply-push I know. The Trans-Java Toll Road runs about 1,167 km from Merak to Banyuwangi. Merak to Surabaya, about 930 km, was connected in December 2018.²⁸

The evidence of “they came” is real but mixed. A panel study from IPB, the Bogor agricultural university, finds a positive effect on growth in the Central Java districts the road passes.²⁹ A study from Universitas Indonesia (UI) finds no stable effect yet on gross regional domestic product (GRDP) per capita, unemployment or poverty in the medium term.³⁰ And there are losers: in towns along the old north-coast road, Pantura, a Pekalongan study reports that batik sellers lost the passing trade that used to stop on their street.³¹

So yes, they came, mostly to the interchanges and the new industrial estates. The road had what most supply-push projects need: an economic reason (Java’s cities already trade with each other) and a connection at both ends.

Why are Kertajati, Whoosh and IKN struggling?

Because each took the “build” from China and left something else behind. Five things, in my view.

1. No continuity between administrations. Each ministry and agency guards its own turf (ego sektoral, a.k.a. “my budget, my project, my ribbon”), and a five-year term rarely lines up with a fifty-year plan. China can plan long and actually enforce the plan, a benefit it gets from its one-party system. Indonesia can plan long, but a plan that the next government doesn’t own tends to stall.

2. The main infrastructure was built without its connections. Kertajati, West Java’s international airport (BIJB, Bandara Internasional Jawa Barat), was inaugurated on 24 May 2018 with a phase-1 capacity of about 5.6 million passengers a year.³² Since the Cisumdawu toll road opened in 2023, it is about 60 km and about an hour by road from Bandung.³³ Distance is not the real problem. The connection is: there is no rail link at all. When Bandung’s in-city Husein Sastranegara airport moved its jet flights to Kertajati from 29 October 2023, passengers didn’t follow. In 2024 Kertajati handled 413,240 passengers,³⁴ about 7% of its capacity (my own calculation). And from 14 August 2026, the government let jets fly from Husein again: a quiet admission.³⁵

Shanghai shows that distance alone was never the problem. The table below sets Kertajati beside Shanghai’s two airports.

Airport	Distance from the city	Rail link	Fastest way in
Shanghai Hongqiao	About 13 km from the centre13	Metro Line 2, and a high-speed rail station next door14	Metro Line 2 into the centre
Shanghai Pudong (opened 1999)	About 30 km from the centre36	Metro Line 2, and the 30 km Maglev since 200437	Maglev: about 7 minutes 20 seconds to Longyang Road, then the metro
Kertajati (opened 2018)	About 60 km from Bandung33	None	About 1 hour by road on the Cisumdawu toll road

Pudong airport is more than twice as far from the centre as Hongqiao, and it still took over most of Shanghai’s international flights, because the rail was there.³⁶

Whoosh has the same problem at its ends. The Jakarta-Bandung high-speed railway was inaugurated on 2 October 2023, with stations at Halim, Karawang, Padalarang and Tegalluar.² Tegalluar, the station built for Bandung, is more than 10 km from central Bandung. Halim is in East Jakarta, not the centre. So most Bandung riders get off at Padalarang and take a 15-to-22-minute feeder train to Bandung station.³⁸ In 2025 Whoosh carried 6.2 million passengers: 3.1 million at Halim, 2.2 million at Padalarang and only 750,000 at Tegalluar.³⁹ The station built for Bandung is the least used.

3. No land-value capture. China pays for a line by selling the land the line made valuable. Whoosh carries its cost without that land. It cost about US$7.2 bn, 75% of it borrowed from the China Development Bank (CDB), and the debt is about Rp116 trn (US$7.3 bn).² PSBI, the consortium that holds the Indonesian stake and is 58.53% owned by KAI, the state railway, lost Rp4.99 trn in 2025 and Rp5.13 trn in the first half of 2026 alone.⁴⁰

4. Slow, expensive land acquisition, and projects that start before they are settled. Land acquisition (pembebasan lahan) in Indonesia is a negotiation with every owner along the route, and projects often break ground before the feasibility study, the financing and the connections are agreed. Each of those, settled late, comes back as a delay, an overrun or a missing link.

5. Building where there isn’t yet a reason to come. IKN, the new capital in East Kalimantan, is the purest supply-push project Indonesia has. Under Perpres 79/2025, IKN is to become the “political capital” in 2028, with Rp48.8 trn budgeted for 2025 to 2028.⁴¹ For 2026, the IKN Authority (OIKN) got Rp6.3 trn in the draft budget, and the DPR rejected a Rp14.92 trn top-up.⁴¹ The open question is the one that left Tianjin’s Yujiapu, billed as China’s Manhattan, about 80% empty around 2019,¹⁶ beyond the civil servants who are told to move, who comes, and when?

Do you need a one-party state to build like this?

No, but it would be dishonest to pretend the system doesn’t help. Let me concede that first. A one-party state can keep one plan for thirty years, can acquire land without years of court cases, and can enforce what the plan says. Shenzhen and Pudong needed all three. Indonesia’s ministries guarding their own turf, and Australia’s airport line slipping to 2033, are what happens without them.

The costs are just as real. The same system built Yujiapu, with at least RMB 200 bn invested and about 80% of its offices empty around 2019.¹⁶ It left local financing vehicles with debt the IMF estimates at RMB 71.4 trn, 50.9% of GDP, in 2025.¹⁹ It moves people out of their homes with weak ways to say no, and nobody in the system has much reason to tell the planners they are wrong until the towers are already standing empty.

Democracies can keep long plans too; they just have to write them into law and institutions instead of a party. Perth’s METRONET has run as a state programme across more than one term of government, and its Yanchep extension opened in 2024. Australia has an independent body, Infrastructure Australia, to vet the big projects. And Indonesia already has the legal tool: Law No. 59 of 2024 on the long-term development plan (RPJPN, Rencana Pembangunan Jangka Panjang Nasional) for 2025 to 2045.⁴² A twenty-year plan in law is exactly the “fifty years ahead” that Jokowi was told about. The hard part is making each new government build on it, not around it.

So the lesson worth copying isn’t the political system. It is the discipline: plan long, connect everything on day one, pay for it with the value it creates, and keep going after the ribbon is cut.

So, how does China build so fast?

It builds first, on land it owns, pays with the value the building creates, and doesn’t stop when the leaders change. That is the whole trick, and it explains both Pudong and Yujiapu.

My own principle, after more trips to China than my passport strictly needed: building ahead of demand is a bet, and building without a connection is a bet you have already lost. Build ahead, by all means. But connect it, give people a reason to come, and keep going.

So here is the small action. Next time you see a ribbon-cutting, on the news, on your feed or in your own city, ask two questions. What connects to it on day one? And who pays for, and keeps building, the next phase after this government is gone? If nobody can answer either one, you are not looking at the next Pudong. You might be looking at the next Yujiapu.

And an honest question back to you: which project in your city would you build first, before anyone asks for it?

References