Antares engine failure at Wallops Island

On October 28, 2014, the sky over Wallops Island, Virginia, turned into a fireball of twisted metal and unspent propellant. The Antares rocket, built by Orbital Sciences Corporation, lifted off from NASA’s Wallops Flight Facility at 6:22 p.m. ET. Six seconds later, it died. The vehicle shuddered, lost thrust, and fell back onto the pad in a catastrophic explosion that lit up the coast for miles. No one was hurt, but the payload—a Cygnus spacecraft packed with 5,000 pounds of supplies for the International Space Station—was obliterated. The failure was a hard lesson for American spaceflight, and it came from a place nobody expected: a 40-year-old Soviet rocket engine.

Antares was designed to be a workhorse. Orbital Sciences won a $1.9 billion contract under NASA’s Commercial Resupply Services program, the same initiative that gave SpaceX its Dragon missions. But while SpaceX built its Falcon 9 from scratch, Orbital took a shortcut. For the first stage, they bought AJ26 engines—refurbished NK-33 engines originally built in the 1960s for the Soviet Union’s N-1 moon rocket program. These engines were powerful, efficient, and cheap. They were also sitting in a warehouse for half a century, and that’s where the trouble started.

The AJ26 was a marvel of Soviet engineering. Designed by Kuznetsov, the NK-33 used a highly efficient staged combustion cycle, blowing the competition away in specific impulse. But those engines were never meant to fly in a modern American rocket. They were stored for decades in humid conditions, their internal parts subject to corrosion and fatigue. Orbital Sciences did what they could—they tested each engine at NASA’s Stennis Space Center, ran diagnostics, and declared them flight-ready. But the analysis that followed the Wallops explosion revealed a fatal flaw. A turbopump bearing in one of the two AJ26 engines failed. The bearing broke apart, sending metal shrapnel into the engine’s oxygen-rich preburner. The engine lost pressure, then thrust, and the rocket’s guidance system couldn’t compensate. The whole thing came down in less than ten seconds.

Here’s the raw truth: the failure wasn’t a manufacturing mistake. It was a fundamental material problem. The bearing had microscopic cracks that inspection missed. Orbital had tested the engine and cleared it, but the physical limits of a 40-year-old machine were already baked in. You can polish a Soviet relic, but you can’t make it new. The lesson for the industry was brutal: heritage hardware doesn’t forgive you. You can run every test in the book, but if the metal is tired, it will fail when you least expect it.

The aftermath was a mess of blame and accountability. NASA and Orbital Sciences spent months picking through the debris, analyzing telemetry, and running simulations. In the end, Orbital admitted the engines were a risk they didn’t fully understand. They switched to a new first stage powered by RD-181 engines—modern, reliable, and fresh off the production line in Russia. The new Antares flew successfully in 2016 and continues to launch Cygnus missions today. But the switch cost time and money. Orbital had to redesign the entire first stage, delay their launch cadence, and absorb the reputational damage of a highly public failure.

For the casual space fan, the Antares failure is a textbook example of why innovation matters, and why shortcuts backfire. SpaceX was already proving that new engines, built from scratch with modern manufacturing, could fly reliably and cheaply. Orbital wanted to cut corners and save cash, and they paid the price with a smoldering crater on the Virginia coast. The failure also rammed home a point that engineers already knew: test data is only as good as the hardware you test. You can simulate all day, but when a bearing cracks at 200,000 RPM, your math doesn’t matter.

The broader takeaway for American spaceflight is that the old way isn’t always the smart way. For decades, NASA and its contractors relied on proven hardware—Space Shuttle parts, Russian engines, payload fairings from the Apollo era. The logic was simple: if it flew before, it’ll fly again. But the Antares failure proved that “proven” doesn’t mean “safe.” Hardware fatigues. Supply chains rot. And when you’re launching a multi-million-dollar payload, you don’t want your engine to be a 50-year-old gamble.

Since 2014, the industry has moved more aggressively toward new designs. Rocket Lab, Relativity Space, and Blue Origin all emphasize fresh manufacturing and modern testing. Even Orbital’s replacement engine, the RD-181, is a newer design. The lesson from Wallops Island is burned into every launch director’s mind: if you want to fly, don’t trust a grandfather’s engine.