Google's moonshot factory just laid bare its brutal math. X CEO Astro Teller revealed at TechCrunch Disrupt 2025 that only 2% of the lab's wild experiments ever graduate - and he's perfectly fine with those odds. The confession offers rare insight into how the world's most ambitious research lab actually operates behind closed doors.
Google's most secretive division just opened its playbook. Astro Teller, the man behind Alphabet's X moonshot factory, used TechCrunch Disrupt 2025 as his confessional - admitting that 98% of everything his team touches crashes and burns. But here's the twist: he couldn't be happier about it.
"We have a 2% hit rate," Teller told the packed conference room Monday, his matter-of-fact delivery making the statistic hit harder. "Most of the things we try don't work out, and that's okay." It's a stunning admission from the lab that birthed Waymo and Wing, but Teller's math tells a different story about failure.
The numbers reveal X's counterintuitive strategy. While only 2% of projects graduate after five to six years, those rare successes consume 44% of X's entire budget. "We kill all the bad ideas pretty early in the process," Teller explained, describing a system that's more surgical than most realize. The lab launches over 100 experiments annually, treating each as a hypothesis to disprove rather than prove.
Teller's moonshot definition cuts through Silicon Valley's buzzword fog with three non-negotiable requirements. First, tackle a massive global problem. Second, propose a product or service that could theoretically solve it, however unlikely. Third, identify breakthrough technology that offers "a glimmer of hope" at making the impossible possible. Miss any component, and X isn't interested.
"If you propose something and it sounds reasonable, we're not interested," Teller said, drawing laughs from an audience accustomed to startup pitches promising incremental improvements. "That wouldn't be a moonshot by definition." The admission reveals how X deliberately seeks projects that sound insane to conventional wisdom.
The teleporter example crystallized X's philosophy. "If you came up with a teleporter, I'd say awesome, here's a tiny amount of money, go see if you can prove it's wrong," Teller explained. "I don't want you to make it work. I want information about whether this is a once-in-a-generation opportunity." It's venture capital turned inside out - funding failure rather than success.
Teller's leadership philosophy centers on what he calls the "audacity and humility" balance. "If you don't have really high audacity, you won't start these unlikely journeys," he noted. "But if you have anything less than really high humility, you'll go luminously far down that unlikely journey." The phrase captures X's core tension between ambitious vision and ruthless pragmatism.
The broader tech industry watches X's approach with fascination and skepticism. While companies like Meta and Microsoft pour billions into AI research, X's model deliberately embraces spectacular failure as a feature, not a bug. The strategy contrasts sharply with quarterly earnings pressure that dominates most of Alphabet's other divisions.
Teller's conference remarks come as Google faces intensifying competition in AI and autonomous vehicles. OpenAI and Tesla have captured public attention with splashy product launches, while X's methodical approach keeps most projects under wraps until they're ready to graduate. The timing of Teller's transparency suggests confidence in X's pipeline.
The human element emerged in Teller's closing thoughts about innovation being learnable. "Every person was creative as children, but we unlearn things useful for radical innovation," he observed. "You can find them again by creating environments where you don't feel stupid." It's a philosophy that explains why X encourages wild ideas while maintaining rigorous kill criteria.
Teller's candid breakdown of X's 2% success rate offers a masterclass in strategic failure management. While other tech giants chase quick wins and quarterly growth, X's willingness to kill 98% of its projects early creates space for the breakthroughs that reshape entire industries. The approach isn't scalable for most companies, but for Alphabet's moonshot factory, embracing failure as a feature rather than bug might just be the secret to betting on tomorrow's impossible becoming today's inevitable.
