Vietnam’s 19.3 GW, Indonesia’s 853 MW of rooftop and a $13bn-to-$30bn grid bill: what ASEAN’s 38 GW looks like divided by 680 million people
By: James Vance – SeaPRwire – Thirty-eight gigawatts. That is the whole of Southeast Asia’s installed solar capacity at the end of 2025, according to the IEA and IRENA’s country tables, against 1,202 GW in China alone. The regional figure is fifteen times what it was a decade ago, which is the line every press release leads with, and it is true.
Now divide it. Six hundred and eighty million people in the ten ASEAN member states, 38 GW of panels, and the answer is 56 watts per person. China’s equivalent number is roughly 852. Germany, which sees less sun than Vietnam, is somewhere near 1,000. One of these numbers is a market. The other is a pilot programme with good publicity.
Everything that follows is an attempt to take that 56 apart, because the ratio is not an accident of policy ambition. It is the output of a system where cells are cheap, land is available, demand is growing at 5.4% a year, and the thing that turns sunlight into delivered electricity, which is the grid plus storage plus a contract that pays for output rather than capacity, is being funded at about a third of the rate required.
Fifty-six watts per person
Country-level numbers make the regional aggregate look less like a trend and more like one country plus four laggards. Vietnam and Malaysia are in a different category from Indonesia, and Indonesia is in a different category from almost everyone.
| Country | Solar capacity, end-2025 | Added in 2025 | Nameplate watts per person |
| Vietnam | 19.3 GW | 0.59 GW | 191 |
| Thailand | 8.7 GW | 1.20 GW | 124 |
| Malaysia | 5.8 GW | 1.45 GW | 170 |
| Philippines | 3.9 GW | 1.10 GW | 34 |
| Indonesia | 1.5 GW | 0.55 GW | 5 |
| Rest of ASEAN | ~1 GW | ~0.2 GW | ~5 |
The four largest markets account for about 78% of the regional total. Indonesia, with 284 million people and the region’s biggest electricity demand growth, has less solar than Belgium. That single comparison is worth more than any projection, because it locates the constraint precisely: it is not irradiance, and it is not the price of modules.
Vietnam added 586 MW last year, after adding 11.7 GW in five years
The Vietnamese number is the most misread in the region. Cumulative capacity reached 19,252 MW at the end of 2025, up 586 MW on the year, according to IRENA’s renewable capacity statistics. In 2024 the addition was 79 MW. In 2023 it was about 1.6 GW. In the two years to 2020, it was 11.7 GW.
Read those four numbers in sequence and the story is not growth. It is a system that ran a very successful feed-in tariff experiment between 2017 and 2020, discovered it had built more generation than its 500 kV backbone could evacuate, and then spent five years arguing about what the earlier contracts were worth.
The argument is not abstract. A 2023 audit found that a large share of FiT-era projects lacked construction completion certificates at the point of commissioning, and the utility began a process of retroactive tariff adjustment covering 173 solar and wind projects with roughly 12 GW of combined capacity and an estimated USD 13 billion of investment. Some producers saw receivables more than twelve months past due before payments resumed.
The revised Power Development Plan VIII, adopted in April 2025, sets a solar target of 46 GW to 73 GW by 2030, up from about 12.8 GW in the previous plan. BloombergNEF’s forecast lands at 39 GW to 43 GW, close to the base case and far from the top of the range. The gap between the plan and the forecast is the cost of the last five years of uncertainty, expressed in gigawatts.
Capacity factors are where the headline dies
Nameplate capacity is a count of panels, not a measure of electricity. A solar farm in Ninh Thuan runs at roughly 18% to 20% of nameplate across a year, because the sun sets, clouds pass and the inverter clips. Run the arithmetic: 38 GW × 8,760 hours × 0.18 is about 60 TWh.
ASEAN’s total electricity generation is somewhere around 1,150 TWh. Sixty divided by 1,150 is 5.2%, which is exactly the share the IEA reports for solar and wind combined, and which the agency expects to reach 11% by 2030 as both technologies grow at more than 20% a year. The arithmetic and the official number agree, which is reassuring and also the point: the region’s solar story is a 5% story being told as if it were a 30% story.
Capacity factors in central Vietnam deteriorated between the first quarter of 2025 and the first quarter of 2026, in the region where generation has most severely outrun evacuation capacity. That is a polite way of describing curtailment. When a system operator tells a plant to stop generating, the plant does not produce, the meter does not turn, and the capacity factor for that region falls.
Curtailment has no line in a PPA contract
Here is the part of project finance that rarely makes it into a policy announcement. A standard Vietnamese power purchase agreement pays for energy delivered, not energy available. There is no compensation mechanism for curtailment. If the grid cannot take the output at 11 a.m. on a sunny Sunday, the developer absorbs the loss, and so does the lender who sized the debt against a production forecast that assumed the electrons would flow.
That single contract term explains the shape of the market better than any subsidy does. It pushes developers toward rooftop and behind-the-meter projects, where the electricity is consumed on site and never touches the transmission queue. It explains why Ember’s analysts found rooftop performing best in Vietnam, with interest concentrated among commercial and industrial users who have a factory roof, a daytime load and a Scope 2 commitment to a foreign buyer.
It also explains the storage mandate. The revised PDP VIII contains a planning requirement that new utility-scale solar include storage equal to at least 10% of installed capacity with two hours of discharge, and sets a battery target of 10 GW to 16.3 GW by 2030. Whether the requirement survives individual tenders depends on the province and the tender design, and not all of them have written it in.
Seven hundred million a year against three point six billion
The transmission number is the one that makes the rest legible. National Power Transmission Corporation spent an average of about USD 700 million a year on transmission over the past five years. The revised plan targets USD 3.6 billion a year between 2026 and 2030.
That is a fivefold increase in a single budget line, in a country whose state utility has been through a period of financial strain and a tariff dispute. Multiply the gap across the region and you get the IEA’s estimate that annual investment in grids and storage across ASEAN needs to rise from roughly USD 13 billion today to USD 30 billion by 2035.
None of that money produces a photo opportunity. A transmission line is a right-of-way negotiation, an environmental impact assessment, a tower foundation and a compensation schedule for eleven households. It takes five to eight years and it is politically thankless, which is why the region has spent a decade building the cheap, fast, visible thing and under-building the expensive, slow, invisible one.
What a solar farm earns: 6.1% and 7.3%
Ember’s Vietnam work produced two numbers that ought to be printed on the front of every bankable project. Using the current tariff structure, the projected internal rate of return is 6.1% for utility-scale solar and 7.3% for solar with a battery. The analysts’ view is that revenue for projects considered bankable would need to be closer to 10% to 12%.
| Asset type | Projected IRR | What the number implies |
| Utility solar, no storage | 6.1% | Below most infrastructure hurdle rates |
| Solar plus battery | 7.3% | Storage improves the case by 1.2 points |
| Bankable range, per Ember | 10–12% | Requires contracts that price firm output |
A 6.1% return is not a disaster. It is a number that a development bank can live with and a commercial lender cannot, which is exactly the split you see in who is actually financing Vietnamese solar. It also tells you why the market shifted to rooftop: a factory that consumes its own output is avoiding a retail tariff, not earning a wholesale one, and the spread between those two numbers is where the return lives.
Indonesia’s 853 MW and the quota that caps it
Indonesia has 284 million people, a GDP per capita around USD 4,925, some of the best and most consistent irradiance in the region, and 853 MW of rooftop solar as of 2025. Vietnam has about 6.9 GW of rooftop. Thailand has roughly 3.6 GW. Malaysia has about 1.8 GW. Indonesia, the largest country in the region by a factor of two, has less rooftop than a single Vietnamese province.
The reasons are specific and they are all policy. Under MEMR Regulation No. 2 of 2024, net metering was restricted, so a household can no longer fully offset its bill with exported surplus. Connections are subject to a quota administered by the utility, and the 2025–2034 supply business plan allocates 3,037 MW of new rooftop over nine years, which is about 337 MW a year for a country adding several gigawatts of demand annually. A one-kilowatt hybrid system costs IDR 20 million to 30 million up front, against a GDP per capita of about IDR 80 million. Payback runs 7 to 12 years, because the retail tariff it displaces is held below cost by subsidy.
Put those four facts together and you have a market that is technically ready, financially unattractive and administratively throttled. Installers will tell you the demand exists. The quota is the binding constraint, and quotas are a choice about who gets to connect first.
Laos exports 40.8 TWh and imports 0.6 TWh
The “battery of Southeast Asia” is the cleanest case study in the region of what happens when you build generation for export and under-build everything else. Laos has about 12.4 GW of installed capacity, of which large hydropower is roughly 73%. In 2025 it exported around 40.8 TWh and imported about 0.6 TWh, mostly in the dry season from Thailand at a higher price than it sells for.
The asymmetry is the story. In 2024 the country earned more than USD 2.6 billion from electricity exports and spent more than USD 177 million importing power because its own transmission and distribution could not move electrons to where they were needed. On the southern grid alone, Laos imported 782 GWh from Thailand’s EGAT in 2025 while exporting 1,065 GWh in the other direction and sending 2,988 GWh to Cambodia. The wires work; the domestic network does not.
Hydropower at 73% of capacity also means the system’s output is a function of rainfall, and 80% of generation depends on it. Dry-season shortfalls have pushed the government toward coal, with capacity projected to rise from 1.9 GW to 3.7 GW by 2035, against sovereign debt reported at USD 12.6 billion. A country that sells hydro to its neighbours and buys coal capacity to keep the lights on is not a paradox. It is what the arithmetic produces when storage and transmission are the missing terms.
There is a small, instructive correction underway. In April, the state utility began buying residential rooftop output at LAK 992 per kWh, about four US cents, under contracts of up to ten years, capped at 50 MWp through the end of 2026. Fifty megawatts is a rounding error against 12.4 GW. It is also the first time the export-first model has been pointed inward.
Batteries: 4.2 GW against 144 GW
Southeast Asia had about 4.2 GW of battery storage at the end of 2025, against 144 GW in China and 24 GW across Japan, South Korea, Australia, New Zealand and Singapore combined, according to Ember’s September assessment. Against 38 GW of solar, 4.2 GW of storage is roughly eleven per cent, and most of that is not co-located with solar in a way that shifts midday output to the evening peak.
This matters because of the shape of the demand curve, not the size of the annual total. A region adding 5.4% of demand a year, driven by industry, cooling, data centres and electrification, is adding exactly the kind of load that peaks after the sun goes down. Every gigawatt of solar without storage increases the midday surplus and the evening deficit at the same time. Grid operators respond the way grid operators always do, with curtailment and with gas.
The IEA’s projection is that most ASEAN systems stay in the early phases of variable renewable integration through 2030, where flexibility needs can be met by better forecasting, shorter dispatch intervals and more efficient use of existing assets. Vietnam is the exception, already encountering the higher-phase problems of large supply-demand swings. That sentence is the most useful thing in the agency’s review, because it says the region has about four years to build the flexibility it will need in 2031, and it is currently building about 11% of the storage it would take.
The Philippines went from 510% growth to 2.4% of supply
The Philippine case is the region’s best argument for auctions and its best warning about percentages. Solar capacity reached about 3.9 GW in 2025 after 1.1 GW of additions, following a 2024 in which the market grew more than 500%. Generation from solar was about 3 TWh, or 2.4% of supply, up 44% year on year.
Both numbers are true and they describe different things. A 500% growth rate off a small base is a statistic about the base. Two point four per cent of supply is a statistic about the system. The fourth green energy auction allocated more than 10 GW of solar, including 2.3 GW of floating capacity and 1.2 GW of hybrid solar-plus-storage, and the MTerra project, completed in February 2026, pairs 3.5 GW of solar with 4.5 GWh of battery. That is the template the region needs: auctions that clear, plus storage attached at the point of build rather than ten years later.
Singapore imports what it cannot build
The region’s richest market has resolved the land problem by importing the electricity instead. Singapore’s Energy Market Authority is working toward importing up to 4 GW of low-carbon power by 2035, roughly 30% of its electricity demand, through projects in Sarawak, Indonesia, Cambodia and Vietnam, carried on new subsea cables and on the Lao PDR–Thailand–Malaysia–Singapore pipeline that has been running in test quantities for years.
This is the most rational version of regional integration and also the most fragile. A 1 GW hydro import from Sarawak depends on a cable, a treaty, a price formula and a counterparty that will still be there in 2045. Every one of those is a political variable rather than an engineering one. The engineering, in the case of submarine HVDC, is the easy part.
For the exporting countries the trade is straightforward and slightly awkward. Laos sells hydro and buys coal capacity. Indonesia would sell solar from Sumatra and keep building coal for Java. Malaysia sells from Sarawak while its peninsula builds gas. The region is not short of electrons. It is short of the wires that would let a surplus in one province reach a deficit in another, and of the settlement rules that would make the trade worth doing.
Demand is not the weak link
None of this is a story about weak demand. ASEAN electricity demand has grown about 6% a year since 2015 and is projected to grow 5.4% a year through 2030, with Indonesia accounting for roughly 40% of the increase and Vietnam about 26%. That is the fastest sustained growth of any large region outside Africa, and it is being driven by exactly the loads that are hardest to serve with solar alone: cooling, industry, electric vehicles and data centres.
A data centre campus is a flat load that runs all night. It is a wonderful customer for a nuclear plant, a hydro dam or a gas turbine, and a poor one for a solar farm without storage. The region’s incoming compute build-out, concentrated in Johor, Batam and around Jakarta, is therefore not automatically a solar opportunity. It is a firm-capacity problem wearing a sustainability commitment.
Three numbers that decide 2027
Strip out the announcements and three figures will tell you whether the region’s solar build-out turns into delivered electricity. All three are published quarterly, and none of them is a capacity target.
- Curtailment, in GWh, on the Vietnamese 500 kV system. If it rises while capacity rises, the build-out is decorative.
- Rooftop connections approved in Indonesia, against the 3,037 MW nine-year allocation. The quota is the policy, so the queue is the outcome.
- Storage commissioned as a share of new solar, region-wide. Eleven per cent is the current number; anything below 20% by 2028 means evening demand is still being met with gas.
If the grid bill gets paid, the next number is 90 GW
A counterfactual is worth stating plainly, because it clarifies what is actually being bet on. If the region spends the USD 30 billion a year the IEA says is needed by 2035, if the storage share goes from 11% to 30%, and if Vietnam’s DPPA framework under Decree 57/2025 does what its supporters expect and lets generators sell directly to factories at negotiated prices, then 90 GW of solar by 2030 is not ambitious. It is the base case.
None of those three conditions is about panel prices, which have already collapsed. They are about contracts, queues and wires. The cheap part of this transition is finished. What remains is the part where a developer has to know, five years before financial close, whether the grid will accept the output at 11 a.m. on a Sunday in April.
Divide by hours, then by the evening peak
The 56 watts per person is the number to keep, because it survives every revision of every target. But the sharper test is what happens after division by hours and by the shape of the day. Thirty-eight gigawatts of panels produce nothing at seven in the evening, when air conditioners in Jakarta, Manila and Ho Chi Minh City are all still running, and four point two gigawatts of batteries cannot cover the difference.
That is the entire investment case in one sentence, and it does not require a forecast. It requires a spreadsheet with 8,760 rows, a capacity factor, an evening load curve, and an honest line for curtailment. Most of the region’s policy documents have the first two. Almost none of them have the last one.