At a Bakken oil pad in North Dakota, gas once burned off in a flare feeds 86 mobile data centers, and the stack that used to light up the dark stands unlit
At night on a Bakken oil pad, the sky used to glow orange from a flare burning gas nobody wanted.
Now the flame is gone.
In its place sits a steel shipping container, humming with rows of computers.
The firm deployed more than 110 mobile generation systems and abated over 39 billion cubic feet of gas, equivalent to removing over 300,000 cars from the road.
The gas that once lit up the prairie dark now runs a small engine, making electricity on the spot.
That electricity never touches a power line.
It goes straight into machines that were never meant to share a fence line with a pumpjack, and the question is how that works at all.
Why gas that nobody wanted became fuel nobody has to move
Oil wells do not produce oil alone. They bring up natural gas as a byproduct, a mix that separates out of the crude the moment pressure drops at the surface. In booming basins like the Bakken, drillers often hit gas faster than anyone built pipe to carry it away.
When there is no pipeline nearby, the economics do not work to build one for a single remote well. So operators have historically just burned the surplus off in a flare, because it is cheaper to waste it than to capture it.
That changed once someone realized a computer rack does not need a pipeline. It needs electricity, and a generator can be built small enough to sit on a trailer, bolted straight onto the wellhead.
The gas never has to travel anywhere. It is burned a few yards from where it leaves the ground, turned into power, and consumed on the spot by machines running cryptocurrency mining or cloud computing workloads. So the fix was never about making the gas more valuable, it was about cutting out the transport step entirely.
What a pad full of wellhead computers actually looks like
Picture a standard oil pad: a pumpjack nodding up and down, a tank battery, a flare stack standing off to the side. Now add a shipping container parked a short walk from the wellhead, vented and insulated, with a generator feeding power into banks of processors stacked floor to ceiling inside.
A thick cable runs from the generator into the container. That is the entire footprint.
One of the companies behind this approach started with exactly that setup in North Dakota’s Bakken shale. Its cofounder has said roughly 1.4 billion cubic feet of natural gas is flared daily in the United States, with about two thirds of that in Texas and another 500 million cubic feet in North Dakota.
That company built its business on the remainder, deploying modular units directly at the flare. The scale has grown fast: it now operates 86 mobile data centers across 30 sites in major oil fields plus Argentina, processing over 10 million cubic feet of gas daily.
How the numbers stack up once someone actually measured it
By the time researchers and the company itself tallied the results, the volumes were no longer trivial. The firm deployed more than 110 mobile generation systems and abated over 39 billion cubic feet of gas, equivalent to removing over 300,000 cars from the road.
That is not a pilot anymore. That is a large scale diversion of a byproduct that used to simply go up in smoke. The company’s own accounting backs that trajectory.
Between 2018 and 2025, its flare mitigation systems repurposed over 21 billion cubic feet of stranded natural gas, enough to generate 2.5 TWh of electricity and avoid 2.7 million metric tons of CO2 emissions, with over 10 billion cubic feet captured in one recent year alone.
A Pennsylvania operator running a similar pilot called it a way to be “good stewards of our assets, focused on innovative solutions to best use resources while minimizing environmental impact and improving performance.” Regulators have put numbers on the combustion difference too: a Texas legislative bill analysis found that redirecting gas from a flare to a generator produces a 63 percent decrease in carbon emissions and a 99 percent drop in methane emissions, since an open flare does not always burn completely.
Where the arrangement runs into its own limits
None of this turns flaring into a clean process, and the company is careful not to claim that. A spokesman for one of its investors wrote that the wells it works with are already flaring and would continue to do so without its involvement, and that it has turned down projects where it would become a net adder of demand and emissions.
That is a narrow promise. It mitigates existing waste rather than eliminating the underlying need to flare at all. Environmental researchers have also pushed back on the broader framing.
One energy researcher interviewed by public radio does not begrudge entrepreneurs for solving the problem of stranded gas, though critics argue using it for computing can incentivize more drilling rather than less. If a wellhead rack makes flared gas worth something, it could, in theory, make a marginal well worth drilling that otherwise would not be.
As this model scales, it touches wider questions about energy security and price swings tied to foreign oil investment that compete for the same capital and attention. In the Bakken itself, one operator’s president described the approach to CNBC as a way to solve two problems with one fix, pairing stranded gas with an appetite for power that did not exist on these pads before.
What happens next to the gas nobody used to want
The pattern keeps spreading into new basins. Regulatory pressure is pushing in the same direction: one industry research note found that New Mexico has set a target of 98 percent gas capture by 2026, while North Dakota has rolled out tiered flaring targets tied to production volumes.
Wellhead computing offers operators a way to hit those targets without laying a mile of new pipe. The honest caveat is that this does not make flaring disappear, it only makes a fraction of it productive.
But on a pad where a flare used to burn through the night for no return at all, a container full of steadily working machines is a strange, concrete sign that even industry’s most wasted byproduct can find a buyer standing right next to it.
Hugo is an engineer with strong technical expertise. Multilingual from an early age, his writing combines technical clarity with a strong interest in science and energy.
