On March 13, 1968, a drilling crew working for Atlantic Richfield and Humble Oil broke through the final layer of rock on Alaska’s North Slope, near a bend in the Sagavanirktok River, 250 miles north of the Arctic Circle. The well came in with such pressure that it bent the steel casing. When the engineers ran their calculations, they found themselves looking at somewhere between nine and ten billion barrels of recoverable oil in a single place—the largest oil field ever discovered in North America. They named it Prudhoe Bay.

The oil was there. Reaching it on a good day took two hours by plane from Fairbanks, when the weather allowed flying at all. Getting the oil out was a different problem entirely, and that problem had a shape you could draw on a map. Eight hundred miles stretching south from the Arctic coast, across three mountain ranges, one of the longest rivers in the country, four hundred miles of permanently frozen ground, and thirty-four major river crossings, down to a port on Alaska’s southern coast where ships could load it and carry it to refineries in the lower forty-eight.

No one had ever built a pipeline across land like this. No one had ever moved that much equipment that far into a country with almost no roads. And the men who would eventually build it could not begin construction until seven years after the oil was discovered. The reason for that delay, and the reason they built it as fast as they did once they started, was the same reason.

The land itself rejected nearly every assumption the engineers brought to it. The solution they arrived at changed what a pipeline could be. Before the embargo, before the lawsuits, before the engineers solved the problems of permafrost, earthquakes, or river crossings, the pipeline needed a route. And the route was not obvious.

The oil sat at Prudhoe Bay on the Arctic Ocean. The only Alaskan port deep and warm enough to load supertankers year-round was Valdez, on Prince William Sound on the southern coast. The straight-line distance between the two was almost exactly eight hundred miles. That line ran through everything.

Through the Brooks Range, the northernmost major mountain chain in North America, through a pass called Atigun that rises 4,739 feet, where Arctic winds blow hard enough to drive snow horizontally, and where winter temperatures drop to sixty below zero. South of the mountains, the route crossed the Yukon River, nearly a mile wide at spring flood. After the Yukon, the route climbed over the Alaska Range, the chain that holds Denali, through another pass, then across the Chugach Mountains before the final descent to Valdez. Valdez sits at the head of a fjord in one of the most seismically active regions on Earth.

Twelve years before the pipeline was proposed, a magnitude 9. 2 earthquake struck directly east of that area, dropping parts of Valdez’s waterfront into the ocean. The engineers at Alyeska Pipeline Service Company, the consortium of oil companies formed in 1970 to build and operate the line, knew what the route was. What they did not yet know in 1969 was how the ground along that route would react when a hot pipe was placed inside it.

In Fairbanks, at the University of Alaska, and at a field station run by the U. S. Geological Survey, soil scientists had been studying permafrost for decades. Permafrost is ground that stays frozen year-round, sometimes for tens of feet, sometimes for hundreds.

It covers about eighty percent of Alaska, and it is not uniform. Some of it is frozen sand or gravel, dry and relatively stable. Some of it is mostly ice. And the critical thing about that kind, the kind that would determine how nearly half the pipeline was built, is what happens when it warms up.

When frozen, permafrost is solid. Buildings and roads can sit on it for decades if the surface temperature never rises. Heat it, and the ice inside melts, and the ground turns into something between mud and cold water. Engineers called the result thermokarst—irregular mounds, collapse pits, and soft patches where solid ground used to be.

Roads sink into thermokarst. Buildings tilt. And a buried pipeline carrying oil at 145 degrees Fahrenheit, which is roughly the temperature of crude coming out of the ground at Prudhoe Bay, would warm the surrounding permafrost from the inside out in a radius of thirty feet or more. The pipe would settle unevenly and bow, and eventually, under a mountain range, it would rupture.

Alyeska’s first plan was to bury the entire pipeline underground. That was how pipelines were built everywhere else in the country, and that was what the engineers proposed when they filed their federal construction permit applications in 1969. Arthur Lachenbruch, a geophysicist who had been studying permafrost in Alaska since the 1950s, reviewed that proposal. His analysis was not encouraging.

He concluded that along at least four hundred miles of the eight-hundred-mile route, burying a hot oil pipeline would melt the surrounding permafrost enough that the pipe would sink into the ground at different rates in different places. And that uneven settling of a buried pipeline under three mountain ranges meant one thing: it would break. That finding did not stay in the lab. Environmental groups—the Wilderness Society, Friends of the Earth, the Environmental Defense Fund—cited it in federal court in 1970, along with a separate legal argument that the right-of-way Alyeska had requested was wider than the Mineral Leasing Act of 1920 allowed.

A federal judge agreed and issued an injunction. Construction stopped before it started. What followed was not merely a legal battle. It was four years of overlapping arguments about soil science, wildlife biology, federal land law, and what the National Environmental Policy Act of 1969, which had just taken effect, actually required the government to study before granting a right-of-way across federal land in Alaska.

The environmental impact statement the Interior Department eventually produced ran to nine volumes. It studied permafrost, salmon runs, caribou migrations, seismic risk at Valdez, and the risk of a tanker spill in Prince William Sound. The tanker question was one the oil companies would rather not have raised, and the answer the study gave—that a spill was possible and that its consequences would be severe—proved accurate sixteen years later, when the Exxon Valdez ran aground in 1989 and poured eleven million gallons of crude into that same sound. But in 1973, the study was complete.

The right-of-way issue was settled when Congress amended the Mineral Leasing Act while the pipeline was still stuck in court. The Canadian route—running east across Alaska, down the Mackenzie River valley in Canada, and connecting to American pipelines in the Midwest—had its own supporters and had been studied for years. What settled the matter in the end was not an engineering argument. It was an oil tanker from Saudi Arabia that never arrived.

October 17, 1973. The Organization of Arab Petroleum Exporting Countries announced it would stop selling oil to any nation that had supported Israel in the October War. The United States was at the top of that list. Within weeks, gasoline prices doubled.

Lines formed at gas stations across the country. The federal government imposed a national speed limit of fifty-five miles per hour, cut fuel allocations to the states, and dimmed the lights in federal buildings. President Nixon appeared on television and asked Americans to set their thermostats to sixty-eight degrees. The country was consuming about seventeen million barrels of imported oil a day, and the tap had been shut off.

The Senate had already voted on a bill authorizing a trans-Alaska pipeline in July. The House passed it in November. The final Senate vote was fifty to forty-nine, broken by Vice President Gerald Ford. The bill contained a provision that had never appeared in an American pipeline law before.

It stripped federal courts of jurisdiction over any further environmental challenges specific to the Alaska pipeline. The lawsuits that had held the project under injunction for four years could not be refiled. Nixon signed the bill on November 16, 1973. The injunctions were gone by evening.

But the embargo had also done something legislation could not undo. It had told every contractor, engineer, and project manager in the country how fast Washington needed this line built. Permits were issued in January 1974. The first crews were moving equipment by February.

Work began on all major sections simultaneously by spring 1975. The goal was oil reaching Valdez within three years. Three years to build eight hundred miles of pipeline meant building about 270 miles a year—about a mile a day on average—across three mountain ranges, the Yukon River, and four hundred miles of Alaskan permafrost in winter. Alyeska divided the route into five sections, each managed by a different prime contractor.

Fluor Engineering and Construction took the northern section. A joint venture of J. Ray McDermott and Perini took the sections in the middle. Peter Kiewit and Sons took parts of the southern route.

Each section was further divided into work groups—self-contained units with their own equipment fleets, their own crews, and their own daily goals. At the peak of construction in the summer of 1975, twenty-eight separate work groups were active along the route at once. The total workforce was 28,000 workers directly on the pipeline, with about 40,000 more working on access roads, pump station construction, the Valdez terminal, and logistics support. That number—70,000 workers in total at peak—is what surprises most people.

Alaska’s entire population in 1974 was about 350,000. The pipeline project employed the equivalent of one out of every five Alaskans, except that most of the workforce was not from Alaska. They came from every state in the country. Moving them and their equipment there required building a road to the job site.

There were no roads across the Brooks Range. There was one road of any significance in the interior, the Richardson Highway, running from Fairbanks to Valdez, and it passed near only part of the southern section. To build the northern half of the pipeline, Alyeska first had to build a road to reach it. The haul road ran 414 miles from the Yukon River north to Prudhoe Bay.

Construction crews built it in under five months in 1974, laying the roadbed on frozen permafrost in winter so the frozen ground would hold heavy equipment. The gravel for the road surface was mined from along the route. The finished road, now called the Dalton Highway, was built as a supply road, not a public road. Trucks ran in convoys day and night in temperatures that froze diesel in the tanks.

The road still exists today. It remains, by a wide margin, the most remote public road in the country. The pipe itself—48 inches in diameter, made of steel with walls up to half an inch thick—had been ordered from Japanese and American mills years before construction began. It arrived by ship at Valdez, Seward, and Whittier, then went north by truck in convoys.

A single 40-foot section of 48-inch pipe weighed about 7,000 pounds. Moving it along a road that climbed mountain passes and was sometimes only one lane wide, in convoys running around the clock, required a separate traffic coordination system to keep convoys from blocking each other. The permafrost problem did not end with the injunction. It was solved, and the solution is what most people see if they fly low over the Alaska pipeline.

Along 420 miles of permafrost-sensitive ground, Alyeska raised the pipe. Instead of burying it, they set it on supports—pairs of steel posts driven into the frozen ground, each pair carrying a horizontal crossbeam, with the pipe resting in a saddle above the beam, off the tundra. Engineers called these supports VSMs, vertical support members. The pipeline used 78,000 of them.

Laid end to end, those supports would stretch from Prudhoe Bay to Seattle. Each pair of VSMs sat about 60 feet from the next. The pipe passed over the ground at a height that allowed caribou to walk freely beneath it. That was not accidental.

The design had been reviewed by wildlife biologists during the environmental impact process, and the height was set specifically so the four major caribou herds crossing the work area would not have their migration paths cut. In the years after the pipeline opened, biologists tracked the Central Arctic herd through the pipeline corridor. The herd grew. But the VSMs had a permafrost problem of their own.

Steel posts driven into frozen ground conduct heat downward. In summer, the posts could warm the permafrost around their bases and begin to sink, slowly and unevenly. The solution was a heat pipe—a sealed tube running through the center of each vertical support, filled with anhydrous ammonia. In winter, ammonia at the base of the tube absorbs heat from the ground, evaporates, rises to the top of the post, releases the heat to the cold air above the snow line, then condenses and returns to liquid and falls.

It is a cooling system with no moving parts, no power source, no fuel, and no maintenance except leaving it alone. There are 122,000 such heat pipes on the pipeline today. They have been running continuously since 1975, keeping the permafrost frozen beneath a pipe full of hot oil. The elevated design also solved a problem no one had tackled at this scale before: seismic movement.

The 1964 Good Friday earthquake, the second-largest ever recorded, struck 75 miles east of Valdez at magnitude 9. 2. It dropped parts of Valdez’s original waterfront into the ocean and triggered a tsunami that killed 131 people. The pipeline route crossed two specific active fault systems.

If the ground moved under a rigid buried pipeline, the pipe would break cleanly at the fault line. The elevated design allowed something entirely different. Each section of pipe on the elevated portions rested in its cradle saddle without bolts holding it down. The engineering specification required the pipe to be able to slide twenty feet horizontally in any direction and five feet vertically without breaking a weld or a joint.

To give it room to move, the elevated sections zigzag in a shallow repeating S-curve—a deliberate geometric deviation designed so the pipe has somewhere to go when the ground moves. On November 3, 2002, a magnitude 7. 9 earthquake ruptured the Denali Fault beneath the pipeline, south of the Alaska Range. The ground moved fourteen feet laterally in about three seconds.

The pipeline slid in its cradles. The welds held. The oil kept flowing. In the mountains south of Prudhoe Bay, the route climbed into the Brooks Range through Atigun Pass, the highest point on the pipeline at 4,739 feet.

The Atigun section was buried rather than elevated. The ground there was solid rock, not permafrost, and the grade of the pass was so steep that anchoring an elevated pipeline there was nearly impossible. But burying pipe through a mountain pass in the Brooks Range in winter meant drilling blast holes in rock that reached forty below zero, threading pipe through a notch blasted into the mountain, and getting the work done on a schedule that left almost no margin for weather delays. The weather did not care about that schedule.

In the winter of 1975 to 1976, storms on the North Slope shut down helicopter supply flights for days at a time. Week-long blizzards isolated work crews at the Atigun Pass camps. The construction camps—prefabricated housing units, kitchen trailers, generator shelters, repair shops—had been designed for this, and they held up. But welding work stopped when winds reached sixty miles per hour, and they reached that speed far more often than the project budget had planned for.

South of the Brooks Range, the pipe descended into the Dietrich River valley and headed toward the Yukon. The ground here shifted between permafrost and non-permafrost within as little as a quarter mile. The soil classifications made from survey samples years earlier did not always match what crews found when they dug. Field engineers reclassified each section as the work crews advanced, deciding on the spot whether to bury the pipe or elevate it.

There are places along the Alaska pipeline today where the pipe dives underground, runs for hundreds of yards, then rises again onto vertical supports, then goes underground again—not because anyone planned it in Washington, but because the permafrost map on the ground did not match the map on the desk. The Yukon River crossing was a separate problem. The Yukon at the pipeline crossing point is 2,295 feet wide at normal flow. At spring breakup, the river carried ice chunks the size of building lots moving downstream at several miles per hour.

An underwater crossing required reaching a depth sufficient that ice chunks scraping the riverbed would not reach the pipe. That meant blasting a trench in the riverbed, which meant working inside the river, which meant doing it in the short window between ice-out and freeze-up, a window measured in weeks. It also meant shutting down river navigation for the whole period. Instead, Alyeska built a bridge.

The Yukon River Bridge is 2,295 feet long, carrying both the haul road and the pipeline on the same structure. It was built on piers sunk into the riverbed, and the available time for constructing those piers was no more than a few weeks each construction season. The bridge opened in October 1975. At completion, it was the longest pipeline bridge in the world, and it remains among the longest in North America.

The river crossings did not stop at the Yukon. Thirty-four major waterways crossed the pipeline route. Most of the smaller crossings were handled by directional drilling—a technique in which a drill enters the bank at an angle, passes under the riverbed, and emerges on the other side, pulling the pipe behind it. In 1975, directional drilling at this scale was not standard practice.

Alyeska used it anyway, because open-trench digging in an active salmon stream would not pass the permitting review. The directional drill at the Tazlina River ran under 1,800 feet of riverbed. At the time, it was one of the longest horizontal directional drills ever completed on an oil pipeline anywhere in the world. The workers who built the pipeline lived in construction camps strung along the haul road and at intervals along the southern sections.

At the project’s peak, thirty-one camps were operating at once, some housing more than 3,000 workers each. The largest camps were cities in every sense of the word. They had their own water systems, power plants, sewage treatment, medical facilities staffed with doctors and nurses, and recreation buildings where workers could watch television and play pool after a twelve-hour shift. Work ran on a rotation schedule.

Workers flew in from Anchorage, Seattle, or Houston on chartered aircraft, worked seven days a week for two weeks straight, then flew home for a week of vacation before coming back. The pay was exceptional by 1975 standards. Pipefitters earned between twelve and fifteen dollars an hour at a time when the average manufacturing wage in the lower forty-eight was under five dollars. Overtime was paid at time-and-a-half, and overtime was constant.

Workers with the proper certifications—especially certified pipe welders, who were in acute shortage throughout the project—were taking home $1,500 to $2,000 a week, month after month, in 1975 dollars. That attracted people from everywhere. Job applications came in from every state and many foreign countries. Alyeska and its five prime contractors hired welders, pipefitters, electricians, heavy equipment operators, truck drivers, laborers, carpenters, cooks, and medics.

The unions—the Teamsters, the International Union of Operating Engineers, the United Association of Plumbers and Pipefitters, and Pipeliners Local 798 of Tulsa—were central to the hiring process. Hiring halls in Fairbanks and Anchorage processed thousands of applicants weekly during peak hiring. Local 798, which supplied the pipe welders who joined the pipe sections, sent members to Alaska from jobs in Texas, Louisiana, and Oklahoma. A pipeline welder who had been working Gulf Coast oil fields suddenly found himself working eighty degrees colder, on a schedule that did not stop for the cold.

The welding itself was the critical skill. Every joint, occurring every 40 feet, had to be X-rayed for integrity. Any joint that failed the X-ray had to be cut out and re-welded, costing hours on a schedule measured in miles per day. The weld rejection rate at the start of the project was higher than the specification allowed.

Alyeska and the contractors brought in additional inspectors, applied new pre-heating procedures for steel in extreme cold, and retrained welders whose rejection rates were high. As the project went on, the rejection rate fell. The joints that went into the pipe were good joints. In nearly fifty years of operation, the pipeline’s major leak incidents have come from pump station failures and a shooting sabotage, not from weld joints.

In 1969, when the oil companies first proposed the pipeline, their engineers estimated the cost at $900 million. By the time construction began in 1974, that figure had already risen to $3. 5 billion. By the time the final weld on the mainline was completed in 1977, the actual cost was about $8 billion.

In 1977 dollars, the cost overran the original estimate by about eight hundred percent. The reasons were familiar to anyone who has studied megaprojects in difficult terrain. The permafrost was worse than the initial surveys predicted in several sections. Seismic design requirements, heavily upgraded after the engineering review that followed the environmental impact process, added thousands of sliding support sets to sections originally designed to be fixed.

Environmental monitoring requirements added inspection procedures and stop-points at every major water crossing, slowing the pace of work. The scale of the camps, the haul road, the supply chain, and the medical system across eight hundred miles of nearly roadless wilderness cost more than anyone expected when the $900 million estimate was written. And there was the fact that the embargo had made clear that speed mattered more than cost. Every month the pipeline went unfinished was another month the country paid for foreign oil while oil sat under American ground.

When cost overruns happened, the pressure from the oil companies, who were making record profits in the post-embargo market, was not to slow down and re-design. It was to fix the specific problem and keep working. Eight billion dollars over three years is a different number if the pipeline will operate for fifty years than if it will operate for five. And that is the number worth pausing on, because the original $900 million estimate was not the result of inefficiency.

It was an estimate made before anyone fully understood the permafrost requirements, the seismic design requirements, and the cost of building a city every thirty miles across the Arctic for three winters. The final weld on the mainline was made on May 31, 1977, at Pump Station 3 on the south slope of the Brooks Range. From there to Pump Station 1 at Prudhoe Bay, and south for eight hundred miles to the Valdez marine terminal, the pipeline was complete and pressure-tested. Twelve pump stations pushed the crude along the way, maintaining the pressure and temperature needed to keep it moving at three to five miles per hour through a pipe crossing three mountain ranges and facing winters that dropped to sixty below.

On June 20, 1977, oil began moving south from Prudhoe Bay. It took thirty-eight days to reach Valdez, traveling through the 48-inch pipe in the same direction the construction crews had worked. South through the passes, over the Yukon, across the interior, and down through the Chugach Mountains to the terminal. The Valdez terminal had its own construction story.

Eighteen storage tanks, each 250 feet in diameter and 62 feet high, built on ground that had been unstable since the 1964 earthquake. Terminal engineers drove 3,000 steel piles into the ground beneath the tank farm and built a system of berths for supertankers in the fjord. The terminal alone cost more than the original estimate for the entire pipeline. The first tanker loaded at the terminal was the Arco Juneau.

It took on its first cargo on August 1, 1977. Governor Jay Hammond was there. So were the executives of every oil company in the consortium, and a crowd of construction workers who had come up from the job site. One ironworker who had spent two years building the Valdez terminal later said the thing that most struck him, when the oil finally began loading into the tanker, was how quiet the crowd was.

Three years of noise—helicopters, generators, drilling, blasting—and then the pumps ran, and nobody said much for a while. The Trans-Alaska Pipeline has operated continuously since 1977, with short shutdowns for inspection and repair. In its peak years in the early 1980s, it carried more than two million barrels of crude oil a day. That was about twenty-five percent of total American domestic oil production at the time.

By the early 2020s, with Prudhoe Bay production in decline, that flow had fallen to about 500,000 barrels a day. Discussions about new Arctic fields and pipeline expansion have gone on for years. The pipeline changed Alaska in ways still being assessed. State oil revenue, beginning in 1977, funded the Alaska Permanent Fund, established by constitutional amendment in 1976.

By the mid-2020s, the fund had paid out more than $47 billion in annual dividends to Alaska residents. It also changed the state’s employment base, political balance, and physical character in ways that were not all voluntary and not all welcome. The Dalton Highway, the 414-mile haul road built in five months in 1974, is now a public road open to any vehicle prepared to take it seriously. Truckers drive it year-round to supply the Prudhoe Bay oil fields.

Tourists drive it in summer to see the Brooks Range and the Arctic coast. It is the only road in the United States that crosses the Arctic Circle for any significant distance and continues north. The ground conditions that required building it in winter on frozen tundra are the same conditions that make it difficult to maintain in summer, when the surface layer thaws. The heat pipes in the 78,000 vertical supports still run their continuous cycle.

In winter, ammonia vapor rises from the frozen ground, releases its heat above the snow line, then falls and does it again. No moving parts, no fuel, no one supervising. They have been running since 1975, and if the structure stays intact, they will keep running as long as the temperature difference between the ground and the air keeps driving them. Again, there is a section of elevated pipeline north of the Yukon, near the Five Mile Camp complex, where the route cuts across open tundra in a straight line, and the Brooks Range stands clear to the north on a sunny day.

The pipe runs over the tundra on its vertical supports, heat pipes rising from every post, and an S-curve visible in the line if you walk alongside it. In every direction except the mountains, the land is flat. That stretch cost someone something to put there. The steel was shipped to Valdez, trucked four hundred miles north on a road that had not existed five months before the trucks needed it, hauled to the job site on temporary ice roads that have long since reverted to tundra, driven into the frozen ground by crews working against a deadline measured in barrels of oil, and fitted with cooling systems no one had ever used at this scale on anything.

The permafrost scientists said it could not be buried. The lawyers held it in court for four years. The embargo imposed a three-year deadline. The seismic engineers designed it to move with the ground rather than resist it.

It is all still there. Every piece. That is how it was built. Eight hundred miles, twelve pump stations, three mountain ranges, thirty-four river crossings, 78,000 steel supports, 122,000 heat pipes, a bridge nearly half a mile long over the Yukon River.

Five prime contractors, twenty-eight active work sites at peak, 70,000 workers in thirty-one camps across three Alaskan winters. The most expensive private construction project in American history, completed in three years on land that rejected every initial assumption. That is the story of how eight hundred miles of pipe crossed the top of the world.