It turns out rocket science is still rocket science.
At 11 p.m. on June 18, 2025, SpaceX engineers initiated what should have been a routine six-engine static fire test — a ground test for an upcoming launch — at Starbase’s Massey test site. Instead, Ship 36 experienced a catastrophic failure during propellant loading, which Elon Musk attributed to a nitrogen COPV failure “below its proof pressure.”
That triggered an explosive chain reaction that lit up the South Texas night with the fireball consuming the vehicle. The damage extended to the test stand and sent flames burning into the early morning hours of June 19. No one was injured, but the explosion, dubbed a “major anomaly,” marked SpaceX’s fourth major Starship failure of 2025.
While SpaceX emphasized that “there are no reported injuries, and all personnel are safe,” the explosion is part of a pattern visible in their own mission updates: Flight 7 in January (which SpaceX attributed to “harmonic response” causing propellant leaks), Flight 8 in March (where the company noted “inadvertent propellant mixing and ignition”) and Flight 9 in May (resulting in what SpaceX called “loss of attitude control” from propellant leaks). Each failure is different, each documented in carefully minimized language, each adding to a pattern that R&D professionals recognize as something more significant than isolated “anomalies.”
The financial costs of the failures, writ large, are significant. At $90-100 million per vehicle, SpaceX has destroyed over half a billion dollars in Starship hardware since 2023 alone. Yet the true program cost is far higher: Musk stated SpaceX would spend $2 billion on Starship in 2023 alone, with total program investment exceeding $5-6 billion since development began. For R&D professionals managing innovation portfolios, the June 18 explosion of Ship 36 prompts the question: Is the current trajectory a sustainable way to conduct R&D?
Failing fast only works when you learn faster than you introduce new problems
The pattern of Starship failures underscores a key point: When iteration cycles don’t build on each other, the results aren’t pretty. Each of the four recent failures stemmed from a different root cause, suggesting that SpaceX’s rapid testing cadence may be outpacing its ability to fully understand and address systemic issues.
Consider the progression: Flight 7’s harmonic vibrations weren’t present in earlier tests, indicating new problems emerging at scale. Flight 8’s propellant mixing pointed to a different subsystem. Flight 9’s attitude control failure (with both Ship 35 and Booster 14-2 destroyed) suggested yet another vulnerability. And Ship 36’s composite overwrapped pressure vessel (COPV) failure, reminiscent of SpaceX’s 2016 Falcon 9 explosion, demonstrates that even known failure modes can resurface in new contexts.
This divergent failure pattern with Starship contrasts with iterative development, where each test narrows the problem space. NASA’s Saturn V program, expensive though it was, achieved 13 consecutive successful launches after methodical ground testing. The Soviet N1, on the other hand, never reached orbit after four attempts, each failing for different reasons.
Despite these failures, Musk has maintained an optimistic perspective: “With each launch, especially in the early days…each launch is about learning more and more about what’s needed to make life multiplanetary and to improve Starship to the point where it can be taking, ultimately, hundreds of thousands if not millions of people to Mars.”
