Oct 5, 2026 · 40 min
Astro Teller explains how X turns moonshots into businesses
Google X's Astro Teller: The $1B Bet No CEO Will Back, Moonshots 3x Cheaper in 16 Yrs, and Clean Water at 1¢/L| EP #300
The episode examines whether organizations can make radical innovation repeatable without sacrificing judgment, speed, or ambition.
- 1X kills weak ideas early by testing their hardest economic and technical assumptions first.
- 2Technology and AI can lower moonshot development costs, but human judgment still determines which problems deserve solving.
- 3Radical innovation depends on protected teams and a culture that other organizations struggle to reproduce.
Don't miss
Astro Teller explains why teams should test the hardest assumption first, using failure as learning rather than waiting for a project to become expensive.
The brief
Astro Teller defines a moonshot as an enormous problem paired with a science-fiction-like solution, then explains how X tests whether an idea is ambitious, beneficial, and economically viable.
X’s funnel is designed to kill projects early: teams test the hardest assumption first, because an exciting result is worthless if its underlying premise fails.
The discussion traces how Google Brain, deep learning, TPUs, and transformer architecture emerged from an environment willing to fund uncertain, high-upside work.
Teller argues that technology and AI can make moonshots cheaper, while meaningful problem selection, social acceptance, and deployment still require human judgment.
The standout tension is cultural: X can protect tiny teams from corporate incentives, but that carefully engineered innovation culture remains difficult to copy.
Clean water, materials science, hunger, education, Mars, and longevity all test the same standard: audacious projects must plausibly become enduring businesses within roughly two decades.
What was said on this episode
33 statements · 21 positive · 6 negative · 6 neutral
A moonshot requires a named, huge problem that the team wants to solve.
“there has to be a huge problem with the world that you can name and you want to solve”
Listen at 1:22
A moonshot requires breakthrough technology offering at least a small chance of success.
“there has to be some kind of breakthrough technology that gives us a prayer, at least a tiny chance”
Listen at 1:46
Moonshot teams need very high audacity.
“you have to have very high audacity”
Listen at 2:24
Moonshot teams need humility equal to their audacity.
“you have to have in equal measure is humility”
Listen at 2:40
First-principles techno-economics can eliminate moonshots unlikely to become enduring businesses.
“You can often kill moonshots in the very early days because it's just not going to turn into an enduring business.”
Listen at 4:34
Projects expected to lose money probably will not change the world.
“if you're doing something that's going to lose money, it's probably not going to change the world.”
Listen at 4:44
AI will shorten the time needed to de-risk moonshot ideas.
“I think AI is going to somewhat shrink that time from a crazy idea to us having de-risked it”
Listen at 6:20
Google Brain produced the technology that ultimately became TPUs.
“TPUs ultimately came from Google Brain.”
Listen at 8:16
The transformer architecture underlying ChatGPT came from Google Brain.
“The transformer, which sort of underpins everything, it's literally the T in ChatGPT. Yes, that came from Google Brain.”
Listen at 8:20
Major innovations typically require 15 to 20 years of development.
“Every single innovation is like an overnight success that was 15 to 20 years in the making.”
Listen at 8:49
Clean-water technologies must reach about one cent per liter to transform access.
“You have to be able to get to like a penny a liter all-in cost for it to really change the world.”
Listen at 9:38
A clean-water approach costing ten cents per liter is not viable for X.
“it's going to be $0.10 a liter. Nope. Well, we'll stop doing that.”
Listen at 10:11
Energy storage and transmission that shift time and location will transform the world.
“Being able to time shift and location shift energy is absolutely going to change the world.”
Listen at 10:35
X will continue pursuing education innovation until it finds a solution.
“We're going to keep coming back to that one until we find something”
Listen at 10:53
Humanity had lost some of its explorer spirit during the stagnation period.
“I think functionally we had lost our way a little bit”
Listen at 13:12
Funding highly energetic optimists will produce some moonshots.
“you will absolutely get some moonshots”
Listen at 14:01
More organizations will establish moonshot factories in the future.
“I expect you will see a lot more of it in the future, not just from us.”
Listen at 15:05
High-risk innovation should be isolated where its uncertainty and disruption are tolerated.
“you have to sequester the choice Bs to a place where they're actually wanted and the mess that comes with them is tolerated.”
Listen at 17:16
X deliberately keeps moonshot teams small.
“We keep our teams tiny”
Listen at 18:43
X's cost per graduated project is declining approximately 10–20% annually.
“that is getting cheaper at, I don't know, 10, 20% year over year”
Listen at 21:27
Developing moonshots costs less now than previously.
“it's cheaper to get to moonshots now than it was in the past.”
Listen at 21:42
Systematizing radical innovation is X's meta-moonshot.
“the meta moonshot we often talk about as the meta moonshot is how to systematize radical innovation.”
Listen at 24:23
X prioritizes concrete global problems such as water, energy storage, and education.
“What we're obsessed with is the problem, whether it's clean water or it's a way to make grid-scale energy storage or it's education”
Listen at 26:49
Moonshot development is unlikely to become fully AI-driven within the next decade.
“I'm not sure any time in the next decade it's going to be 100%”
Listen at 28:20
Humans will remain necessary for socially acceptable deployment for at least one or two decades.
“you need humans as part of that process for at least the next decade or two.”
Listen at 28:42
X should pursue audacious ideas that can be de-risked within a decade.
“I want you to find the most audacious things you can work on that we can move from crazy idea to not crazy idea in less than a decade”
Listen at 29:35
X has produced approximately 35 to 50 graduates over 16 years.
“sort of 35 to 50 graduates”
Listen at 30:50
Teams should validate the hardest capability before building supporting infrastructure.
“If you can train the monkey, we can always build a pedestal afterwards.”
Listen at 31:51
Failure provides learning that success does not.
“You only learn things when you're wrong.”
Listen at 32:11
False-positive moonshot decisions can cost tens of millions of dollars.
“The cost for a false positive where we believe it's a moonshot, we run it for many years, and then it turns out not to be is very high.”
Listen at 34:52
Rejecting a genuine moonshot has zero cost if other opportunities remain available.
“The cost of a false negative, where it actually is a moonshot but I rejected it, is zero.”
Listen at 35:05
New materials require industrial-scale production to become businesses.
“How do you make, you know, thousands of tons a day of that material?”
Listen at 36:24
Replicating X requires leadership focused on engineering an innovation-supporting culture.
“you have to have a leadership team that is maniacally focused on engineering a culture in which people can show up in the ways that tend to drive radical innovation most efficiently.”
Listen at 37:17
Statements are attributed to the speaker as said on the episode and reflect their view at the time, not PodLume's. They are not advice.
Books & mentions
Some links are affiliate links — PodLume may earn a commission if you buy.