Kicking off our Impact 2025 conference last month was Helen Riley of Google X, a/k/a “the moonshot factory.” She serves as both COO and CFO there, and in our on-stage interview, she discussed the organization’s mission; several current projects; how they deal with failure; and their efforts to systematize innovation.
Below is a lightly-edited transcript of the conversation I had with Riley at Impact.
• • •
How do you explain to people what Google X is, and why it exists?
X’s mission is to invent and launch moonshot technologies that can make the world a radically better place.
…We really sit at the intersection of purpose and profit. Ultimately, of course, we want to build profitable businesses that are self-sustaining, that are driving revenue and adding value to our partners. We call ourselves a moonshot factory because, of course, we are going after moonshots. That’s a very deliberate choice of words, with reference to Kennedy’s space race, and putting someone on the moon in a decade, which at the time seemed to be impossible. And then factory, because we’re really trying to systemize innovation in a sort of factory-like, repeatable way.
We always start with a huge problem in the world — something that if solved could positively impact the lives of millions or billions of people around the planet.
I think it’s worth explaining, when we talk about a moonshot, what do we actually mean? There’s three components to everything that we work on with X. First, we always start with a huge problem in the world — something that if solved could positively impact the lives of millions or billions of people around the planet. Then we come up with a radical solution to that problem. Often the ideas we come up with sound a little bit impossible and sci-fi sounding. Then, we endeavor to invent a breakthrough technology, something that can take that impossible sounding idea and turn it into a reality, in a 10- to 15-year timeline.
Were there any role models when Google X was being created, in terms of other R&D labs that existed — AT&T’s Bell Labs was famous for a long time. Lockheed Martin had the Skunk Works to develop new kinds of aircraft…
Absolutely… so many corporations that have created these corporate incubators, and skunkworks. But I think one of the things I am most proud of is, relative to a lot of those that ended up fizzling away for a number of different reasons — either they didn’t have the same power because there wasn’t that sponsorship and commitment to investment over the long-term, or there wasn’t focus on commercialization, so they didn’t really turn it into something. We’ve done whatever we can to learn those lessons. And I’m really proud of the fact that, 15 years on, Google is still excited to invest in moonshots that can have a 10- to 15-year timeframe.
Your role combines being COO and CFO. How do you play with others in the organization? I think there’s a CEO, who is Astro Teller. I don’t think two of the founders of X are still around, but tell us a little bit about how you sit in the management team of this organization?
I talked about the fact that this is really a moonshot factory, and so we’re trying to systematize innovation… The best way to think about my role is really trying to operationalize that. I oversee all of the centralized teams that help shepherd our moonshots through the factory — finance, HR, real estate, legal, policy, marcom… And then on top of that, I oversee a portfolio of projects. So I’m not just in a support role, but I’m also living day-to-day what it takes to build some of these moonshots. …Our model is to then graduate these businesses. They move back into Google, or we end up working with investors…[who] will come in and take a stake in our businesses, and then I serve on the boards of those companies. I really enjoy my role. I’m really lucky that I get to see things all the way from ideation at one end of the funnel, right the way through to capital investment and engaging commercial [partners.]
To what extent do people from the core Google business ever move over to X, or is X mostly hiring people from the outside world for projects?
It’s a real mix. Unsurprisingly…we have a number of people that move over from Google. They joined [the company] earlier, and it’s now scaled, and they’re excited to move back to something smaller and more nimble. But we also do attract a lot of talent from the outside, because it can be a very different mindset and approach if you’re working on something that is billions of dollars of scale — you want some rigor and discipline that looks very different than if you’re working on early-stage ideation, [and you need] people who have this growth mindset and are willing to face complete uncertainty… And so we attract a lot of serial entrepreneurs who’ve had experience building companies and are excited to do it again, but also benefit from what Google brings them.
…The first project that I wanted to chat about is Waymo. …Waymo started about 15 years ago at X, and the huge problem that we were tackling is, of course, road safety. There’s about a million people that are killed in road accidents every year, and the vast majority of those are as a result of human error. And so the big idea was, how can we create an autonomous vehicle and essentially take the decision-making in driving away from humans? …Waymo has now expanded to a lot of different cities around the US. I think they’ve driven 100 million miles on roads today, and they are serving 250,000 paid rides a week, and that is skyrocketing. So, really excited. I still get shivers every time I get in a Waymo car…
Another business that we created is a company called Wing. This is our drone delivery business. If you think about last-mile delivery, there is a lot of friction in the system. It can take a long time to get items delivered with large trucks on the road, creating traffic jams. The team at Wing thought, “Well, what if we got off the ground and up into the sky, and we can deliver really rapidly?” The Wing team has partnered with Walmart, and now in a number of cities, […you can literally go to an app and select “drone” as your delivery method. I was in Dallas-Fort Worth just a few months back, and within 10 minutes, I had a Starbucks bottle of coffee delivered to my doorstep, which was pretty amazing.
When Hurricane Helene hit, there were places that were essentially inaccessible, and so partnering with Walmart, they were able to deliver baby formula, and prescription medications, to people that otherwise couldn’t get access. [In] downtown London, the Wing had a partnership with the National Health Service, and so they were actually delivering blood samples across the City of London from hospital to hospital in a matter of just a few minutes, compared against as long as 45 minutes, if they’d taken the roads.
…This is a project that didn’t work out — a company called Loon. The problem that we tried to solve is connectivity. About a third of the population on the planet still doesn’t have access to internet connectivity, and we all know that outcomes with humans improve, educationally, healthcare-wise, financially, economically, if you have access to information, access to opportunity.
The team at Loon… would fly these balloons in the stratosphere and beam connectivity between the balloons and then down to the ground. …It was the right problem space to be going after — to provide connectivity to people, especially in underserved regions — and the technology worked, but the techno-economics just didn’t pan out.
But one of our principles at X is never to throw anything away, because there’s so much that we learn with every single attempt that we make to build a business… And even though the techno-economics didn’t work out, because the tech worked, we hoped that it would live on in some form. [We created a company] called Taara… They have these terminals…[about] the size of a traffic light. And essentially, instead of being internet from the stratosphere, they now have these terminals on the ground. At high speeds, they can provide abundant, affordable, fast internet between these terminals, sometimes 20 kilometers apart. And they’re providing service now to communities in Africa and India and South America.
We also pursued a moonshot in agriculture. The challenge here is, population growth is still increasing, and so there’s going to be 70 percent more people on the planet by 2050. That’s two more million mouths that we’re going to need to figure out how to feed. And so we wanted to see whether we could actually help farmers as they’re growing their crops. The intent here was to see whether we can use advanced perception capabilities and AI to really understand individual plant health. …Our philosophy is to rapidly prototype and test ideas and not spend too much money.
…Tapestry is a moonshot for the electric grid. The grid is one of the biggest and most complex machines in the world that everybody in this room relies on. …One of the big challenges that grids face right now is we’ve got new sources of [energy] supply that are bidirectional in nature, and you’ve also got these unpredictable spikes in demand, and extreme weather events. It’s a real challenge to be able to create stability and resilience in the grid. What the team at Tapestry has done is essentially created a digital twin of the electric grid, so the grid operators have a much better sense of how to visualize and operate and also plan, so they can mitigate downtimes in the grid, and really accelerate connection of renewables to the grid…
Is there a sector or a kind of technology, a business, where you guys wouldn’t do a moonshot project? Is there anything off-limits?
I don’t think so, so long as it aligns with our mission… impacting people’s lives in a positive way. If we can see a benefit to humanity in investing, we’ll go ahead…
Should we do one more? And I’m eager to hear what questions are have been suggested.
Our project for circularity [is] called Materra… Something like 91 percent of plastics end up in a landfill. …The team is leveraging AI and machine learning to be able to understand the molecular components of the plastics, and scan them at high speed running through our system to be able to break them down and recycle [them] at the molecular level. …Even for us, we can’t solve this alone. Working with partners is absolutely critical, because, when you have a bag of chips and you’re throwing it into recycling, if you want to solve this problem, then not only do we need to work with the chip manufacturer, but also the manufacturer the packaging for the chips, and then the chemical company that’s building materials that go into the chip…and then, of course, the companies that do the recycling. So a lot of what we focus on is convening partners so that together we can solve these problems.
Here’s a question from the audience: Moonshots occasionally fail, and land among the stars — or maybe even on the ground. What is your team’s tolerance and margin for failing experiments with purpose at the expense of profit?
The first thing I would say is that it is scary for everybody. Every single one of us can have this aversion to falling flat on their face, but the reality is this is we’re trying to build a culture where people are willing to try really bold things. There’s a lot of things that we do to create a culture and environment where we’re removing the stigma from failure. …We’re looking for people with a humble growth mindset and insatiable curiosity and a high tolerance for uncertainty…
At most companies, rewards will be tied to hitting specific milestones. But when we’re working on radical innovation, we sometimes don’t even know whether the milestone we’re setting is the right thing to be pursuing.
At most companies, rewards will be tied to hitting specific milestones. But when we’re working on radical innovation, we sometimes don’t even know whether the milestone we’re setting is the right thing to be pursuing. What we really want to incentivize is trying something really hard, even if it doesn’t work… What we really want to reward is the learning.
When we are assessing how individuals are doing, so long as they try something really, really hard, so long as they have sort of [absorbed] the learnings and have a better thesis around either proving or disproving their hypothesis, and they share the learnings more broadly with the organization, they will still be rewarded, even if the thing that they’re shooting for has failed. That can take some of the stigma out of the fact that it may not actually have worked.
We also have everybody share every quarter…[with] the entire company… If everyone in the organization can see that everybody else is trying really hard things, it actually encourages everybody to raise the bar and raise their ambition… The other thing I would say is that, by definition, everybody knows that we start with taking hundreds of ideas, and most of them are not going to work.
How do you identify and prioritize projects across so many different areas? You just suggested there are dozens or hundreds of different projects that people have suggested.
…We really try to make sure that [a project] big enough for us to work on. … It’s a radical solution that is non-obvious, that doesn’t really exist today, and everything that we work on [has] some sort of breakthrough technology at its core. …Unsurprisingly, with today’s tools, [like] artificial intelligence [and] machine learning, we’re applying those tools pretty much to every industry. The last thing I would say is that we’re pulling together interdisciplinary teams that may be non-obvious combinations. As an example, we have an ocean health project that we actually partner marine biologists with a roboticist and an expert in machine learning, and that can unlock incredible ideas…
Audience member Jan Beránek: …One of the things that you said early on was that you’ve systematized innovation so that it’s like a factory. And DeepMind [has been doing that] a lot of that in other scientific areas. So I was just curious if and how you’re trying to do that similarly to DeepMind?
… DeepMind is actually a really close partner of ours. We sit in different parts of the organization, but there’s a lot of research that they do, and they have a lot of expertise, and we want to make sure that we’re not working on overlapping things. We’ll frequently come together and find opportunities for collaboration. I’m not as familiar with a lot of their focus, but I think that the majority of what DeepMind is doing, a lot of that focuses on augmenting existing Google products — Gemini and others. From our standpoint…
Audience member David Matheson: You’ve talked about being a factory a couple times, and you’ve hinted at iteration and hypothesis testing. But what is your process at a high level? How do you structure that?
We really think about it like a funnel. We’ll have hundreds of different ideas that we’re working through, and those can be sourced from across Google. We also have teams that are really focused on ideation and brainstorming. We have partnerships with entrepreneurs, with universities, educational institutions… We have business, technical, and financial milestones that we’re setting. Essentially think of it like stagegates — they may be much more technically-oriented in the beginning. We’re much more focused on the technical economics and market opportunity as they move through this journey.
…We’re trying to break the idea, or see what are the ways in which it won’t work, and we’ll dispel the ones that won’t work. But where we have increasing levels of confidence in our hypothesis, then we have the opportunity to fund them a little bit more, and we’ll set another set of milestones and see whether they’re making progress towards that milestone. But I also want to say that it’s really important not to be too rigid with the milestones that we’re setting, because often we don’t know what’s the right milestone to set. It is as much art as it is science in making that assessment. Even if a team hasn’t hit a specific milestone, but they’ve learned something that helps us — maybe we pivot in a different direction, or maybe we double down, then we want to be fluid with that process… So we whittle down over time the number of projects that we’re working on are the ones that we get really excited about, where the technology works, we have great traction with partners who have a lot of confidence that the solutions we’re providing… and we have confidence that the techno-economics will work out…
It’s a myth that Google is spending a crazy amount of money on these earliest stage ideas. We are ruthlessly efficient. It’s built into the system. We’re trying to create a factory, and so it’s absolutely critical that we do not overspend.
Let’s take one more audience question. X is part of this division called Other Bets at Google… basically Google spends a billion dollars on this Other Bets category, not all on X, but a billion is a big sum. So if this person who asked this question wanted to hire you into their organization, how would you take some of these lessons and ideas and do them with fewer resources?
I think it’s a myth that Google is spending a crazy amount of money on these earliest stage ideas. We are ruthlessly efficient. It’s built into the system. We’re trying to create a factory, and so it’s absolutely critical that we do not overspend. When we start ideas, we’re talking about maybe 50 percent of one person’s time… We’re really focused on speed and scrappiness and getting real-world feedback, and so we want to really make sure that we learn as quickly as possible whether or not a thing that we’re working on has an Achille’s heel…
…You’ve seen some of the prototypes. Some of those prototypes can be put together with just a few thousand dollars — we’re not talking about millions of dollars — just to test our ideas. And even at the other end of the funnel, our teams are no bigger than 20 people in some cases, and it’s incredible the progress that just a small number of people can make. My point is that radical innovation, especially in these earliest stages, can be incredibly cheap relative to what you might expect. Every step of the process through the funnel has to be ruthlessly efficient, because if we overspend early on, that just bloats the whole system…
One thing that came up a little bit our discussion at breakfast today is that X has been around since 2010, so that’s a 15-year stretch. And so many companies we see have these on again/off again labs or R&D groups where you do it for three or four years, then you maybe scale it back, or shut it down. So I’m glad that you’re here to kind of bring this role model example. Last year, we had Walt Disney Imagineering talking about decades and decades of work that they have done. I just feel like that persistence and consistent support is such a big piece of the puzzle here.
I completely agree. X was originally set up by the founders. They’re not involved anymore, and yet that belief in continuing to invest in innovation that is 10 to 15 years out, there’s absolutely sponsorship from the top. The other thing I would add is, in the early days of Google, there was a lot of focus on having an investment profile of 70/20/10. So 70 percent investment in the core business, 20 percent on things that are adjacent to the core business, and then 10 percent on the longer-term future. So X squarely sits in that 10 percent. I think some of the work that DeepMind does also sits in that 10 percent.
There is the continued commitment that magical things can happen If you wait it out. And the first project that I showed you is Waymo. That was 15 years in the making, and I’m delighted that we had the staying power and commitment to see it through, because it’s a magical experience for everybody.
You must be logged in to post a comment.