How Nepal Built a $140M Dam That Took 10 Years to Build—and 7 Months to Break transcript

The mountain [music] doesn't want you here. Nepal spent 10 years drilling 4 km through solid Himalaya to build this. >> [music] >> Three turbines, 111 MW, enough to light a good slice of a country that used to ration its own electricity. It took 9 years longer than planned, cost 36% [music] more than budgeted, buried two of the men who built it under a landslide before it was even half finished. Then, >> [music] >> on 2024, it finally worked. >> Here's the crazy part.

Seven months later, the same river that made this place perfect for power came back and took it apart in minutes. This is the story of a plant built inside a loaded gun and the engineers who knew it the entire time. >> The river that could power a nation could also erase [music] it in an afternoon. High in the Himalayas, on the border between Nepal and Tibet, the Bhote Koshi river drops through one of the narrowest [music] gorges in the range.

Steep, fast, loaded with glacial silt. The kind of water that makes engineers reach for a turbine and makes everyone else reach for higher ground. The river didn't [music] want anyone here. I'm standing where the valley narrows to almost nothing. No flat ground, no room to build anything without touching the water. This is exactly [music] why they came. A river dropping this fast through mountains this steep is pure energy.

Catch it and you can light [music] up half a valley. Here's the crazy part. The same steepness that makes this place perfect for power also means there's nowhere for a flood to go. When the water rises here, it doesn't spread, it accelerates. On 2015, [music] that risk stopped being theoretical. A 7.8 magnitude earthquake tore through central Nepal. Landslides [music] came down on the construction site. 250 workers were trapped.

Two never made it out. They flew a child out by helicopter. >> [music] >> 24 people off a mountainside because the roads were gone. The project had barely [music] started. >> 10 years of construction still lay ahead through that earthquake, through a fuel blockade that would starve the site [music] of cement for months, >> through rock so unstable it had never been tunneled through before. But how do you [music] drill 4 km through a mountain that's still moving?

Feel this. >> This isn't old rock. This is young [music] rock, the kind that doesn't sit still. The Himalayas are still rising, a few millimeters a year, but rising. And the rock underneath this [music] valley is migmatitic nice, quartzite schist, layers pressed together by tectonic force, not settled by time, cut through with fault zones that can shift without warning. Here's the crazy part. This rock doesn't just sit there and be [music] hard.

It fights back. Engineers had to answer one question before they could even start [music] drilling. What happens when you push water through cracked rock at high pressure? Not a small leak, a mountain splitting itself open from the inside, a phenomenon called pH equals [music] hydraulic fracturing. Miss that calculation and the tunnel doesn't [music] fail slowly. It fails all at once underground with no way to see it coming.

So they tested. 10 sections along the proposed tunnel route modeled against Norwegian engineering standards for exactly this failure mode. The number they needed, a safety factor above two. Double [music] the margin required to call it safe. Every single section passed. That number sounds abstract [music] until you remember what it protects. This tunnel runs 4 km through rock that's still deciding what shape it wants to be.

But solving the geology on paper was the easy part. The valley [music] itself was the next problem. Barely a road, a single track cut into the cliff side, 150 km from Kathmandu, closed by the monsoon every single year. >> Every ton of equipment, every bag [music] of cement had to survive that road before it could touch the mountain. >> Now watch what they had to do just to reach the river itself. You don't stop a river [music] like the Botekoshi.

You steer a piece of it. 400 m downstream of the exact point where [music] two rivers and two countries meet, engineers laid out a diversion weir. Not a dam holding back a lake, a low barrier just tall enough to lift the water and push some of it sideways into an intake while the rest of the river [music] carries on exactly like it always has. But lifting the water was only step one. The Botekoshi's Anglo [music] fit So, before the water ever touches the tunnel, it passes through three [music] underground desanding ponds.

Wide, slow chambers built for one job. Let the sediment [music] drop out before the current can do any damage. Building any of it meant working inside a channel that was still running. Standard practice for headworks like this. Wall off half the river at a time, push the water to one side, work in the dry [music] gap behind the barrier, then move the river again and finish the other half. Here's the problem nobody mentions [music] in the brochures.

The Botekoshi doesn't wait for construction [music] schedules. I've seen footage from a season when the river simply took a new course through the site overnight. Flow measured near 5,000 [music] cubic meters a second. Work stopped, they rebuilt and kept going. Then came 2015. The earthquake that had [music] already trapped 250 workers on this site shook loose the slopes directly above the headworks. [music] Landslides buried access roads that had just been cleared.

For months afterward, the only supplies reaching the valley came up that [music] same single track, when it was open at all. The project's own leadership admitted the obvious. [music] The completion date they'd promised in >> wasn't happening. It would take, [music] uh, they said, at least another year and a half. It would actually take nine more. >> The headworks held. Now, they had to get that water 4 km through solid mountain. 4 km 185 m.

That's the exact length of tunnel standing between the river and the power it could generate. [music] 6 m wide the whole way, bored through rock that had already proven it doesn't cooperate. There was only one [music] way to do this safely. Meter by meter. Drill a ring of holes into the rock face, pack them with explosive, clear the debris, lock the fresh walls in place with steel arches and sprayed shotcrete before moving 1 m >> [music] >> further.

Some days the crews advanced 10 m. Some days the rock shifted and they lost [music] a week rebuilding what had already collapsed. Stand at the tunnel face and you understand something fast. The mountain sets the schedule, not the contractor. Watch why that mattered. Every [music] meter of standing water in that finished tunnel, under pressure moving fast, becomes a hammer [music] if the turbines shut down suddenly. Engineers built a vertical shaft near the tunnel's end specifically to absorb that shock, [music] giving the surge somewhere to rise instead of somewhere to explode.

By 2018, the tunnel was finished. The hardest single component of the entire project, done. But the water still needed [music] somewhere to become electricity. Underground, carved directly out of the mountain, engineers excavated a cavern 76 m long, 15 wide, 35 tall. Big enough to hold three turbines and the machinery to lift them into place with the entire weight of the mountain sitting on the ceiling above. Into that cavern went three Francis turbines, [music] vertical axis, 38.5 megawatts each.

Three synchronous generators rated at 43.75 megavolt amperes. Water dropping 168 [music] m from tunnel to turbine. That fall is the entire product. Everything else [music] is just the machine that catches it. A tailrace tunnel 659 [music] m long carries the spent water back to the river downstream. >> A transmission line 10 km of double circuit 132 kilovolt cable carries the power [music] out. On paper, it was finished. >> In practice, the project still owed more money than it had ever earned.

The earthquake didn't kill this project. The blockade didn't kill it, either. What nearly did was simpler and slower. Money. >> By 2020, 6 years into construction, the company building Rasuwagadhi [music] was posting a net loss every single quarter. >> Zero revenue because there was no electricity [music] yet to sell against financing costs that never stopped accruing. Retained earnings had gone negative and kept sinking.

The completion date promised for [music] 2016 had already been pushed to 2020. Then it slipped again to 2023. Here's the number that mattered [music] most. The currency itself worked against them. Over 10 years of construction, [music] the Nepali rupee lost value against the US dollar the project's loans were measured in. The original estimate, 13.68 billion rupees, wasn't wrong [music] when it was written in 2011. It was just measured in a currency that wouldn't hold still for a decade.

The company's answer was to keep raising money [music] the only ways it had left, loans from Nepal's employees provident fund at 9.75% interest, and in 2019, [music] a public share offering, asking ordinary citizens and even the project's own workers to buy into [music] a plant that still wasn't generating a single unit of power. >> Final cost at completion, [music] 18.69 billion rupees, 36% over the original estimate.

The debts absorbed, the shares sold, they kept going. >> December 23rd, 2024. 10 years after ground [music] was broken, engineers opened the intake gates for a real test, not a drawing, not a projection. The river poured into the tunnel that had taken 4 years [music] just to drill. 4 km through the dark, 168 m [music] down the shaft. It hit the first turbine, 52 MW on the very first [music] attempt. Lower than full capacity only because winter had thinned the river's flow.

Full output would have to wait for the rains. >> Weeks later, the transmission line connected to the national grid through Chilime A Trishuli. >> For the first time, the mountain wasn't just holding a plant. It was powering one. Then it happened. 7 months. That's how long the plant ran before the mountain answered back. On July 8th, 2025, a glacial lake far up the Langde Khola, one of thousands swelling as the ice above it melts, gave way.

The water came down the same gorge the tunnel had been drilled into, carrying rock and debris, and it didn't stop at the border. It tore out the Nepal-China Friendship Bridge. It kept going. >> [music] >> It hit the headworks directly. This is the exact structure from part three, the weir, the intake, the three desanding ponds. The flood buried the entire headworks under more than 2 m of mud and boulders. Downstream, the tailrace flooded, too.

The turbine floor itself went underwater. The intake choked. Generation stopped completely. Here's the crazy part. To fix it, engineers reached for the only tool that [music] had ever worked on this river. They diverted it again. The exact [music] same method used to build these headworks back in 2014. Wall off the channel, work [music] in the dry gap. Now used to dig the plant back out of its own wreckage. >> By December 2015, one of three turbines was running again. 37 MW through a temporary river diversion while the real headworks were rebuilt underneath it. >> The other two [music] stayed dark, expected back within a month.

Full reconstruction, not the patch, the real fix, was targeted for [music] completion before the next monsoon. Not a triumph, not a failure. A plant half alive [music] in a gorge that had already proven it doesn't forgive mistakes or good [music] engineering, either. The river that could power a nation could also erase it in an afternoon. It has now done both. Two workers who [music] never made it out in 2015. 24 people flown off this mountain by helicopter.

A plant that [music] took 10 years and 36% more money than planned just to reach the moment [music] it could finally generate power for seven straight months. The mountain didn't lose this fight. It never does. It just let them build for a while. As of this video, the [music] story isn't finished. In 2026, the same valley was struck by flooding again, larger than the [music] first, and Rasuwagadhi has been offline since.

Whether Nepal rebuilds a third time isn't [music] an engineering question. It's a question about how badly a country needs the light this river can give it. >> If you want to understand why engineers keep coming back to rivers like this one, subscribe. The next build is already underway.