Investigation PIA-INV-024HydropowerCost: Audited final settlement 207.873 billion yuan (2013; ~US$34 billion then); approved static budget 90.1 billion yuan (1993 prices); true economic cost contested8 min readSuccess Investigation

Three Gorges Dam: Delivered at Design Capacity — and an Invoice That Never Closes

The World's Largest Power Station Met Its Engineering Targets on Its 17-Year Plan. What It Cannot Settle Is the Argument About Everything Else.

Filed under: Project Successes · China Three Gorges Corporation · China · Energy & Utilities

Written and edited by Ramesh Dixit·Published 2026-08-24·Last updated 2026-08-24·Last fact-checked 24 August 2026·Editorial Standards · Editorial Policy · Corrections Policy · Methodology · Source Standards · AI Disclosure
Download Executive Brief (PDF)

The Three Gorges Dam is the largest power station ever built: 22.5 gigawatts, delivered broadly to its seventeen-year programme and, by China's National Audit Office, within its final budget — an outcome so rare in megaproject history that it deserves serious study. It is also a project that relocated well over a million people, drowned thirteen cities, and carries unresolved debates over siltation, seismicity and flood control. This investigation does both jobs at once: it explains the delivery machinery that worked — staged construction, a dedicated national funding mechanism, early revenue — and it refuses to call the result an uncomplicated success.

The Three Gorges Dam spanning the Yangtze River in Hubei, China, the world's largest hydroelectric station at 22.5 GW, completed in stages between 1994 and 2012
Watch the Documentary

Watch: Three Gorges Dam: Delivered at Design Capacity — and an Invoice That Never Closes

Documentary in productionThe documentary episode for this investigation has been announced and is in production. The video will be published here when it premieres on YouTube — no placeholder is shown.
Key Facts
Cost
Audited final settlement 207.873 billion yuan (2013; ~US$34 billion then); approved static budget 90.1 billion yuan (1993 prices); true economic cost contested
Date
Approved April 1992; construction December 1994; full generating capacity July 2012; ship lift 2016
Category
Hydropower
Executive Dashboard
Industry
Hydropower / water infrastructure
Country
China
Organisation
China Three Gorges Corporation (developer-operator); Government of China (sponsor)
Programme Value
Approved 1992 static budget about 90.1 billion yuan (May 1993 prices); dynamic estimate up to 203.9 billion yuan (~US$25 billion at contemporary rates)
Actual Cost
Audited final settlement 207.873 billion yuan (National Audit Office, 2013; roughly US$34 billion then). Independent estimates of true economic cost vary far more widely; contested
Delay
Broadly to the 17-year (1993-2009) plan for main works; full generating capacity July 2012; ship lift operational 2016, later than first envisaged
Status
Delivered — operating at design capacity since 2012, with enduring caveats
Human Impact
Officially about 1.13 million people relocated (independent counts run to roughly 1.3 million); 13 cities and over 1,300 villages submerged; documented resettlement hardship and fund mismanagement
Success Score
82
PIA assessment
Governance Score
75
PIA assessment
Risk Rating
Medium
Complexity Rating
Extreme
By the Numbers
Audited final settlement 207.873 billion yuan (2013; ~US$34 billion then); approved static budget 90.1 billion yuan (1993 prices); true economic cost contested
Cost — key facts, Three Gorges Dam: Delivered at Design Capacity — and an Invoice That Never Closes investigation
Contents
  1. Executive summary
  2. Watch the documentary
  3. Key facts
  4. Executive dashboard
  5. What happened
  6. Why it matters
  7. Timeline
  8. Root cause analysis
  9. Frameworks applied
  10. Executive lessons
  11. Executive recommendations
  12. PMOS intelligence
  13. Insider take
  14. Evidence
  15. Everything from this investigation
  16. Sources
  17. Author & review
  18. FAQs

What Happened

The Three Gorges Project was approved by the National People's Congress in April 1992 by a margin unprecedented in that body's history — not for its size, but for its dissent: roughly a third of delegates abstained or voted against, the largest recorded opposition to any proposal before the Congress. Construction formally began in December 1994 at Sandouping, Hubei, on a plan first drawn by Sun Yat-sen in 1919: a 2,335-metre concrete gravity dam, 185 metres high, holding back a 600-kilometre reservoir, with twin five-stage ship locks, a ship lift, and what would become a 22.5-gigawatt power station. The stated purposes were flood control, power and navigation — in that official order. The delivery machinery is the part most often skipped in Western accounts. Financing was secured upfront through a dedicated national mechanism: the Three Gorges Construction Fund, raised by a surcharge on electricity bills across most of China, supplemented by revenue from the downstream Gezhouba plant, state bank loans and corporate bonds. Construction was staged across three phases with a river diversion in 1997; crucially, the first generating units entered service in July 2003 — nine years into a seventeen-year programme — so the project began paying for itself while still under construction. The main dam wall was completed in May 2006; the reservoir reached its full 175-metre operating level in October 2010; and in July 2012 the last of 32 main units came online, taking the station to its full 22.5 GW. In 2020 the plant generated 111.8 terawatt-hours, a world record for a single station, against a design annual output of 88.2 TWh. By its thirtieth anniversary in December 2024 it had produced over 1.7 trillion kWh and intercepted floods on nearly seventy occasions, according to the Ministry of Water Resources. The National Audit Office's 2013 report fixed the final settlement at 207.873 billion yuan — within the dynamic budget. Against that record stands a ledger the engineering cannot close. The reservoir displaced officially about 1.13 million people (the figure rises towards 1.3 million in independent and later counts), submerging 13 cities, 140 towns and over 1,300 villages. Resettlement outcomes were uneven at best: academic and World Bank-reviewed evidence on earlier Chinese dam resettlement showed persistent impoverishment patterns, and China's own auditors found hundreds of millions of yuan of Three Gorges resettlement funds misappropriated in the programme's early years. Environmental concerns catalogued since the 1990s — reservoir-induced seismicity, landslides on destabilised banks, siltation, blocked fish migration, estuary salinity change — have been partly managed, partly confirmed, and never fully resolved; the state itself acknowledged 'urgent' ecological problems in 2011. The flood-control claim also deserves precision. The dam's designers and operators point to 2020 — when the reservoir absorbed record inflows and the station simultaneously set its generation record — as vindication. Critics note that the 2020 floods still displaced millions downstream and argue the event demonstrated the limits of any single structure against basin-scale flooding. Both readings are defensible; what is not defensible is the claim, heard from boosters and detractors alike, that the case is simple. PIA's framing: as a delivery programme, Three Gorges is a documented success; as a development decision, it remains a live argument — and the honest investigation holds both.

Why It Matters

Three Gorges is the central exhibit in any honest debate about state-delivered megaprojects. It demonstrates that a committed state with upfront financing, staged construction and revenue-generating phasing can deliver the largest infrastructure object on earth close to plan — a finding that embarrasses both the 'megaprojects always fail' school and the 'only markets deliver' school. It equally demonstrates the limits of delivery success as a verdict. Cost-benefit accounting that excludes displaced populations, ecosystems and downstream risk will certify almost anything. As countries from Ethiopia to Laos build mega-dams with Chinese finance and contractors, the question Three Gorges poses is not 'can it be built?' — it answered that — but 'what does the delivery record omit?'. That question is now the standard lens for every comparable project.

Timeline
  1. approval 1992-04

    The National People's Congress approves the Three Gorges Project — with roughly a third of delegates abstaining or voting against, the largest dissent in NPC history.

  2. construction 1994-12-14

    Construction formally begins at Sandouping under a seventeen-year programme plan.

  3. milestone 1997-11

    The Yangtze is closed and diverted at the dam site, enabling main-channel works — the programme's first great staging milestone.

  4. milestone 2003

    The reservoir begins filling to 135 m and the first five-stage ship lock enters service; in July the first 700 MW generating unit produces power — revenue starts nine years into the programme.

  5. completion 2006-05-20

    The main dam wall is completed at 2,309 m long and 185 m high; state media hail the end of the core concrete works.

  6. milestone 2008

    Trial impoundment toward the 175 m full pool begins; relocation officially totals about 1.13 million people.

  7. completion 2009

    The original seventeen-year programme concludes with all 26 originally planned main units in service; the underground six-unit extension continues.

  8. milestone 2010-10-26

    The reservoir first reaches its full 175 m operating level, enabling design-capacity generation and flood storage.

  9. review 2011-05

    China's State Council publicly acknowledges urgent ecological and geological problems around the reservoir and commits to remediation.

  10. completion 2012-07

    The final (32nd) main generating unit enters service; the station reaches its full 22.5 GW design capacity.

  11. audit 2013-06

    The National Audit Office reports the final settlement at 207.873 billion yuan — within budget — while noting past misappropriation of resettlement funds.

  12. opening 2016-09

    The ship lift — the world's largest — enters trial operation, completing the navigation works later than originally envisaged.

  13. stress-test 2020

    Record monsoon floods test the reservoir (inflow peaks of 75,000 m³/s); the plant simultaneously sets a world record of 111.8 TWh annual generation.

  14. legacy 2024-12

    Thirtieth anniversary: cumulative generation passes 1.7 trillion kWh; the Ministry of Water Resources credits the project with nearly 70 flood interceptions; a new US$11 billion ship-lock waterway is later launched to relieve chronic lock congestion.

Root Cause Analysis

Root cause through the Project Failure Pyramid™ lens

Symptoms

The visible indicators: delays, cost overruns, quality defects, team attrition

Management Failure

Operational breakdowns: poor risk management, scope creep, inadequate resources

Governance Failure

On time and on budget did not settle siltation, seismicity, resettlement hardship or flood-control limits. Governance frameworks that score only cost and schedule will certify projects that history judges more harshly.

Root Cause

River diversion, left-bank plant, right-bank plant, underground plant, ship locks, ship lift: each tranche was commissioned and proven before the next peaked. Scale stopped being a risk because it was never attempted all at once.

The Weekly Brief — one investigation, one executive lesson and one project warning sign each week.

Subscribe free
Executive Lessons

Lessons for Leaders

Hypothecated Funding Beats Annual Appropriations

The Three Gorges Construction Fund — a national electricity surcharge — gave the programme a continuous, dedicated revenue stream insulated from annual budget politics. Megaprojects die of funding dribble; this one never went thirsty.

Design for Revenue Before Completion

First power flowed in July 2003, nine years into a seventeen-year programme. Phased commissioning turned the project into a self-financing asset mid-build and gave engineers an operating feedback loop while construction continued.

Stage the Impossible Into Deliverable Tranches

River diversion, left-bank plant, right-bank plant, underground plant, ship locks, ship lift: each tranche was commissioned and proven before the next peaked. Scale stopped being a risk because it was never attempted all at once.

Delivery Success Is Not Project Success

On time and on budget did not settle siltation, seismicity, resettlement hardship or flood-control limits. Governance frameworks that score only cost and schedule will certify projects that history judges more harshly.

Resettlement Deserves Engineering-Grade Governance

The concrete programme had world-class controls; the relocation of over a million people had demonstrably weaker ones — auditors found resettlement funds misappropriated in the early years. Social programmes need the same QA, audit and accountability as works contracts.

Publish the Caveats Yourself

China's own 2011 acknowledgement of ecological problems and the National Audit Office's public findings did more for the project's credibility than any promotional record. Institutions that document their own caveats retain control of the narrative.

A Single Structure Cannot Own a Basin-Scale Problem

The 2020 floods showed the dam absorbing record inflows and downstream districts still flooding. Flood control is a portfolio — reservoirs, forecasting, land use and evacuation — and overselling one asset invites backlash at the first extreme event.

Executive Recommendations

Executive Recommendations

For mega-infrastructure, establish a dedicated, hypothecated funding mechanism before mobilisation; do not rely on annual budget cycles for multi-decade programmes.

Design programmes so that partial completion generates revenue or service early — first units, first spans, first segments — converting cost centres into self-financing assets.

Apply the same audit, QA and accountability regime to resettlement and environmental programmes as to engineering works, with independent monitors and published results.

Commission independent, continuing review of long-tail risks (siltation, induced seismicity, downstream ecology) with a mandate to publish, for the asset's whole operating life.

Communicate flood- or risk-control benefits as probabilistic and portfolio-based; never market a single structure as eliminating a basin-scale hazard.

PMOS Intelligence

PMOS Intelligence

Preview — illustrative assessment; PMOS is in development
Governance WeaknessEngineering governance was disciplined and audited; the weakness sits in the resettlement and environmental programme, which was never subjected to the same independent audit, QA and accountability regime as the engineering works.
Escalation FailureResettlement hardship and ecological concerns were documented domestically and internationally for years; escalation into programme-level re-decisions was limited by the project's status as a national commitment ratified by the NPC.
Decision DelayStaged construction with early revenue was the model's strength — first units online July 2003 converted a cost centre into a self-financing asset; the ship lift (2016) shows long-tail components still slipped beyond the main-works plan.
Leadership Blind SpotThe blind spot was the human ledger: 1.13-1.3 million relocated people and downstream ecological effects were treated as manageable externalities of an engineering success, a framing independent monitors continue to contest.
Risk VisibilityFinancing risk was engineered out upfront — a hypothecated electricity-surcharge construction fund, Gezhouba revenue, state loans and bonds committed from the outset; long-tail risks (siltation, induced seismicity, downstream ecology) remain open and monitored but contested.
Evidence QualityThe National Audit Office's 2013 settlement of 207.873 billion yuan gives the programme an audited cost basis few megaprojects have; independent estimates of true economic and social cost vary far more widely and remain unresolved.
Assurance MaturityAssurance over engineering and finance was institutionalised (state audit, staged acceptance); assurance over resettlement and environment was not independent and its results were not published to the same standard.
Suggested InterventionAn independent PMO would have extended the engineering audit regime to resettlement and environment with independent monitors and published results, and commissioned continuing independent review of long-tail risks with a mandate to publish for the asset's whole operating life.
“Strip away the politics and three delivery decisions did the work. First, money before concrete: a hypothecated national electricity surcharge meant the project never waited on annual appropriations — the single most common megaproject killer. Second, cashflow from partial completion: first power in 2003, nine years in, converted the project from pure cost centre into a revenue machine mid-build; nothing disciplines a programme like its own income. Third, staging: diversion, left bank, right bank, underground plant — each phase commissioned and learned from before the next peaked. The counterweight insiders acknowledge quietly: the resettlement programme was run with far less rigour than the concrete programme, and it shows. Three Gorges is what happens when engineering governance is world-class and social governance is an afterthought — you get both a record-breaking power station and a permanent caveat.”Ramesh's Insider Take — opinion
Evidence

Documentary Evidence

government-news

Three Gorges project marks 30th anniversary of construction

Xinhua via english.www.gov.cn (Ministry of Water Resources data) · 2024-12-15

Official operational record: over 1.7 trillion kWh generated, first unit July 2003, nearly 70 flood interceptions, 2.1 billion tonnes cumulative lock cargo — the state-side case for delivery success.

View document →

official-corporate

Three Gorges Dam sets world record for annual power generation in 2020

China Three Gorges Corporation · 2020-12-31

Operator record: 111.8 billion kWh in 2020 (a single-station world record), 22.5 million kW installed capacity and 88.2 billion kWh design annual output — evidence the plant performs at and above design.

View document →

government-news

Last Leg of Work on Main Three Gorges Dam Wall

Xinhua via china.org.cn · 2006-05-19

Contemporaneous construction record: main wall completion May 2006, 2,309 m length and 185 m height, estimated investment 203.9 billion yuan, and 1.13 million people relocated — the official figures PIA hedges against independent counts.

View document →

government-science

Three Gorges Dam, China (Image of the Day)

NASA Earth Observatory · 2009-06-08

Independent scientific catalogue of the caveats: relocation of 1.2 million-plus people, 13 cities and 1,300-plus villages flooded, and environmental concerns including reservoir-induced seismicity, landslides, ecosystem change, pollution and estuary salinity.

View document →

academic-review

The Three Gorges: the unexamined toll of development-induced displacement

Forced Migration Review (Refugee Studies Centre, University of Oxford) · 2000

Critical assessment drawing on World Bank resettlement evaluations: patterns of impoverishment in earlier Chinese dam resettlements, weak legal guarantees for the displaced and the absence of independent monitoring — the basis for this investigation's resettlement caveats.

View document →

Asset Hub

Everything from this investigation

One investigation. Many outputs. Everything derived from this case, connected in one place.

Watch & listen

  • DocumentaryIn production
  • Episode pagePlanned
  • Related documentariesNone yet
  • ShortsNot published
  • Podcast episodePlanned

Share & follow

Sources
  1. Three Gorges project marks 30th anniversary of construction Xinhua / english.www.gov.cn · 2024-12-15 · government-news
  2. Three Gorges Dam sets world record for annual power generation in 2020 China Three Gorges Corporation · 2020-12-31 · official-corporate
  3. China country profile International Hydropower Association · 2025 · industry-body
  4. Last Leg of Work on Main Three Gorges Dam Wall Xinhua via china.org.cn · 2006-05-19 · government-news
  5. No Major Embezzlement of Funds Involving Three Gorges Dam (financing and audit oversight) People's Daily (English) · 2000-09-08 · state-media
  6. Three Gorges Dam, China NASA Earth Observatory · 2009-06-08 · government-science
  7. Yangtze Dams Spill Water (2020 monsoon floods) NASA Earth Observatory · 2020-07-24 · government-science
  8. The Three Gorges: the unexamined toll of development-induced displacement Forced Migration Review (University of Oxford) · 2000 · academic-review
Author & reviewer

Written and edited by Ramesh Dixit

Published 2026-08-24Reviewed 2026-08-24

Last fact-reviewed: 24 August 2026 — see our corrections policy and log.

FAQ

Frequently Asked Questions

The Weekly Brief

Get the Next Investigation First

Receive forensic analyses of billion-dollar failures every Monday. No fluff. Just lessons.

Subscribe to The Weekly Brief