Astro Teller: The $1B Bet No CEO Will Back, Moonshots 3x Cheaper in 16 yrs, and Clean Water at 1¢/L

What happened in this episode?

Peter Diamandis interviews Astro Teller about how Alphabet’s X selects, tests, and kills “moonshots”, and what has changed as AI accelerates experimentation. Teller defines a moonshot as a testable hypothesis combining a huge real-world problem, a science-fiction-sounding solution, and a plausible breakthrough technology pathway, arguing teams need equal measures of audacity and humility to learn quickly and abandon dead ends.

He gives examples of areas X has repeatedly revisited—clean water, energy “time-shifting and location-shifting” (storage and transmission), and education—saying clean-water projects in particular fail unless they can reach roughly one cent per litre at all-in cost. On the economics of innovation, Teller says X measures efficiency and claims its cost to reach “graduates” has fallen about threefold over 16 years, partly due to improved process discipline and the use of AI.

Teller describes X’s project funnel as 1–200 ideas per year reaching code-name status, with roughly 2% later graduating, and argues radical innovation is easy if efficiency is ignored but hard to systematise inside normal corporate incentives. He says senior leadership must protect high-variance bets from organisational “immune systems”, and reports being approached weekly by companies and countries seeking to build their own “moonshot factories”. In a brief but striking aside, he says X is currently probing an internal lab result he believes is “almost certainly wrong” but that would be “a problem for physics” if validated, declining to share details unless it pans out.

What are the key points?

• Teller defines a moonshot as: huge problem + sci-fi solution + breakthrough technology pathway.
• He says X starts 1–200 code-named ideas a year and has produced roughly 35–50 “graduates” over 16 years (about a 2% rate).
• He argues moonshot teams must pair audacity with humility to test quickly and abandon failing ideas early.
• On clean water, he says costs need to reach about 1¢ per litre “all-in” to be world-changing; he says X has a team on it now.
• He claims X’s cost to reach “graduates” is about three times cheaper than 16 years ago, citing AI and process improvements.
• He says X is investigating an internal lab result he believes is likely wrong but would challenge physics if true, and he declines to disclose details.
• He says outside organisations, including countries, approach X frequently about setting up “moonshot factories”, but culture replication is difficult.

Who is speaking?

Named speakers identified in Egleze's source-linked moments from this episode.

• Astro Teller

Which topics are covered?

How is this page sourced?

Programme

Peter Diamandis

Published

5 October 2026

Duration

39m

Original episode

Open the original source ↗

Egleze verifies the speaker, source and timestamp for each linked moment. Unless a moment is explicitly marked independently verified, statements remain attributed claims by the named speaker rather than established facts.