Mary Lou Jepsen: The Full Xconomy Voices Interview

Xconomy National — 

For the inaugural episode of our new podcast, Xconomy Voices, we chose to speak with Bay Area entrepreneur and executive Mary Lou Jepsen. She leads a startup called Openwater, which is developing a new kind of wearable imaging device that might one day enable communication by thoughts alone. Of course, there are nearer-term goals as well, such as supplementing (and even replacing) bulky, expensive MRI machines.

Xconomy has followed Jepsen’s work over the years—from her role as CTO of One Laptop Per Child, to senior executive positions at Google X and Facebook/Oculus. In this recent conversation, we focused on her new ideas and strategies around Openwater, as well as lessons in consumer electronics, imaging, and startups from the past few decades.

Here is the full transcript of our interview (which includes a lot of material not in the podcast):

Xconomy: Thank you for being on the show, Mary Lou Jepsen. It’s great to have you here.

Mary Lou Jepsen: It’s great to be here. Thanks for having me.

Xconomy: Could you start off by saying a little bit about who you are? And then talk a little bit about what you’re up to these days.

MLJ: Sure. I’m Mary Lou Jepsen. I’m the former co-founder of One Laptop Per Child, I was its first CTO. At Google I founded and ran a couple of “moonshots” for Sergey Brin, and at Facebook I was executive director of engineering and worked on bringing virtual reality and augmented reality to the next level. Among the things I founded or co-founded were four startups—I’m very technical, also as a professor at MIT—and my most recent startup, the one I left Facebook for, is Openwater.

Xconomy: What is the driving idea, then, at Openwater? What are you trying to build there?

MLJ: We’re working on a wearable MRI system that doesn’t use MRI. MRI resolution “plus plus.” And the impact is targeted for two areas. Obviously one in health care. And not so obviously, the same device will enable telepathy, meaning communication with human thought alone.

Xconomy: That’s pretty provocative when you put it that way. But when you say MRI you don’t literally mean Magnetic Resonance Imaging, right? You mean an MRI-like kind of sensing of what’s going on inside the body.

MLJ: Right. A way to see inside of your body with equivalent resolution or even higher resolution.

Xconomy: We all know that MRI is expensive and uses these giant magnets and that’s why the machines are only inside hospitals. So is part of the idea here, too, to make it smaller and cheaper?

MLJ: Exactly. Put it into a ski hat or a bra for breast cancer detection or a bandage. Exactly.

Xconomy: OK. Well, how will it work? I mean, what’s the basic science behind this whole idea?

MLJ: So the basic idea is that the body is translucent to near-infrared light. That’s the light that’s wavelengths a little bit longer than red. Not that harmful UV light, but the stuff you can see with night vision goggles. And the body’s translucent to it, but it scatters it. And the basic idea is to use highly custom LCDs with embedded detectors to, first, take a hologram of the scattering of your body and, second, invert the scattering—this is a little technical—using a principle called phase conjugation. And this effect neutralizes the scattering of the body so that we can look at the differential optical signals like changes in the color of the blood which corresponds to oxygen use, changes in refractive index, differential scattering. And these collectively can give the equivalent or even more information than today’s MRI, but at a consumer electronics-type cost and scale.

Xconomy: Wow. So, can you give listeners a little bit of an idea, like a visual picture, of what these LCD arrays and detectors might look like? You mentioned a ski hat. So give us that image.

MLJ: Yeah. Line the inside of a piece of clothing like a ski hat. The lining is an LCD, an embedded detector illuminating with near infrared light instead of visible light but with objects that are quite similar to the LCD in your cell phone. But we can make them flexible. They’re harder in many ways but they’re also easier in many ways than today’s cell phones, [where] you can’t have one bad pixel in this device. It doesn’t really matter. You can have several, many bad pixels. The manufacturing processes, the design constraints are quite different.

But the idea is to fit them into clothing. One question I get a lot is will they be wired or wireless, and the answer to that is certainly our prototypes are going to be wired. We haven’t decided yet on our first products, whether they’ll be wireless or wired, but eventually we’ll go wireless.

Xconomy: OK, so they would be kind of a mesh or a flexible network of LCDs with detectors kind of mixed into that. So you’d be transmitting infrared from the LCD. And then there would be little sensors mixed in among the LCDs that will be picking up the light. Do I have that right?

MLJ: Yeah, like we did in the early 2000s before we did projected capacitive screens. There’s this really cool effect that Einstein explained, the photoelectric effect. And so with silicon you can get it to work in both ways. You can detect light when light hits silicon and the current changes, and we can measure that. So that’s been known for some time and used widely. And again, I don’t buy off the shelf LCDs. For about 20 years I’ve been working and shipping really, really innovative optical electronics components using the big, multi-billion dollar fabs of Asia, to ship billions of dollars of complex hardware/software systems that were really differentiated by the screens in the optical electronics. And so I’m deeply familiar with every single manufacturing process in these multi-billion dollar fabs of Asia. And very few people are because very few employees who are not of the companies themselves get access to these facilities. That’s the thing that I’ve done and I’m leveraging here to drastically reduce the cost and size of medical imaging devices.

Xconomy: So it could be sort of an unexpectedly welcome spin off of all of the development work that’s been going on in manufacturing for cheap, small, light phones and other kinds of devices with displays.

MLJ: Right. I was running advanced virtual reality and augmented reality at Facebook and Oculus, and myself and effectively my colleagues at the other large companies made a compelling story to the manufacturers who live at very slim margins to make many manufacturing processes, to make smaller pixels for higher resolution screens for the next generation virtual reality and augmented reality. And with those process changes coming down the pike I realized I could finally come to a solution of this other problem I’ve been working on for about 20 years. How could you make lower cost MRI-like resolution systems, and leveraging consumer electronics manufacturing to sort of democratize it and get it out to more people.

Xconomy: You gave a TED talk back in 2013 that has something like almost a million views, and you told a little bit of that story then. So could you go through a little bit of that now? I mean, you’ve been thinking about this problem for a long time.

MLJ: Yes. Since the late 90s I think. I had a brain tumor when I was in grad school and got it removed, and I only took a month out actually. I was going to drop out and I actually filled out the paperwork to drop out of my Ph.D. in optical physics at Brown because I was so sick. And then somebody sprung for the cost of an MRI. They found that brain tumor, I had it surgically removed, one month later they let me back into grad school.

I finished my Ph.D. six months later and co-founded my first startup. And I never forgot how close I was to death. I’ve been taking pills for the last 22 years. Every day a dozen pills. And missing those pills or not being able to get access to them profoundly changed my sense of self, my sense of my ability to think. I modulate myself in a way that almost no one else can, because they haven’t had that part of their brain removed.

Meanwhile I became fascinated by neuroscience, and as a matter of self-preservation read very widely on this subject for the last 20 years, and became very interested in advances at that time—in early 2011, 2012—advances in small magnet systems. And I thought that maybe we could make not bigger magnets but better magnets using devices there. Some of them [are known by] an acronym called SQUIDs. Not real squids but a kind of interesting magnetic phenomenon in small size. And I started to delve into that, and actually pitched that to Larry and Sergey when they were starting up Solve for X. I pitched this as a project about creating a system where you could communicate with human thought.

And I think Sergey Brin was smitten and acqui-hired me and my company. But he directed us to not work on that project, and instead I founded and ran a couple of moonshot programs for him that are still stealth. But I still wanted to do this.

Xconomy: At some point you pivoted from the idea of using these magnets, these SQUIDs, superconducting quantum interference devices, and you’re now using infrared. So what happened there?

MLJ: I read a paper in 2014, maybe I read it in 2015, about a group in St. Louis actually who used diffuse optical tomography—sorry there’s another technical [term]—but using infrared light to look through the scattering of your body. It matched the resolution of an MRI system to a couple of centimeters of depth in the head. They weren’t inverting the scattering, but using a fiber optic way and a refrigerator full of detectors, a refrigerator sized system. They were able with near infrared light to match the resolution of functional magnetic resonance imaging.

And I saw that paper and I looked at the pictures in it and have subsequently gone and seen the place and met the people. But I was just floored. I thought, wow. It’s not a magnetic thing. We could do this optically. That’s my thing! And these brilliant researchers and neuroscientists have made this major breakthrough, but they don’t know much about consumer electronics at all, shipping product. So maybe I should dive in and work on this on the side.

And at that point I was at Facebook so I was no longer banned from thinking on the side. Mark Zuckerberg was one of the reasons I joined Facebook. In my interview with Mark, I started talking about my desire to work in brain imaging. And there was a whiteboard in the room we were having dinner and I swear his feet didn’t touch the ground for the rest of the dinner. He was excited.

But at Facebook my day job, my full time job, was really doing advanced VR/AR stuff, running and setting up, well I’m not supposed to talk about what I did there. But anyway obviously it was in setting up interesting inventions and bringing those inventions to [market]. I’m an inventor and I’m also an executive. I’m kind of weird in that way. Most people that become executives in tech stop doing the coding or the invention or stop working in the lab. And I don’t. I don’t think you can really lead engineering unless you’re in it.

I wish I didn’t feel so much on a limb when I say that, but I’ve got more than 100 patents published or issued in the last five years alone. And I work and invent and have a lab at home and continuing to kind of tinker about ideas by working physically in the lab.

And so it occurred to me that after I left Facebook—I think we announced it last May and I left full time in August, but I was really part time since last February—and so I was able to work on my company from that time forward, and last spring I completely reinvented the idea.

I thought that I was going to use the simple approach of bringing in the light and just looking for a signal and being more clever and just dealing with all this scatter, the fog if you will, of your body. But then it occurred to me that I could instead take a hologram. I know a lot about holography. I made my first hologram as a teenager. I spent a decade making really advanced holographic systems including with a group of graduate students. And I co-created the world’s first holographic video system. And a holographic video system is what’s needed for your body. Because it’s scattering, and we need to take a hologram of the scatters of the fog of your body, then invert that. But then your body is moving, so you need to do this at a steady clip, somewhere between 10 microseconds and one second, depending upon what you’re measuring. You need to make a new hologram of your body and then by doing this phase conjugation trick you can invert the scattering and make your body transparent, and that has really, really profound applications.

But the first discovery of this basic physics effect was in 1965 by the very people, Leith and Upatnieks, who made the first display hologram that you could see with your eyes. It was of a train. What’s really interesting about the work of Upatnieks that made that first hologram of the train is the very next hologram they made, they put a diffuser, like a ground glass screen, in front of the train, made a hologram of it, and then inverted the image, as the phase conjugation. And they could see the train through the diffuser. That was like, whoa. That’s pretty amazing. But people haven’t really exploited that technique. I think it’s sort of a forgotten thing. It gets reinvented every decade or so in the field of math, working in, you know, microwaves or different types of wavelengths, and it’s become of interest lately in medical imaging.

Now there are a few scattered academic groups trying to work in this. They’re using off the shelf components, micro displays. They’re using components that are worse than stuff that we shipped in the late 90s at MicroDisplay Corp.

Xconomy: So it’s because you’re so ideally positioned, with your experience in display design and display manufacturing, that you can bring a perspective to this whole holography problem that nobody else has had.

MLJ: That was what occurred to me. And the strange thing is, it occurred to me last spring after I had already jumped off that cliff and started Openwater. But I completely threw out my old ideas on it and and moved forward with this holography thing. And device design, which is really the thing that I’m best known for. I put them into hardware, software, consumer electronic systems. But all my software systems have been really distinguished by getting into every level. Like when you look at a chip, there’s layers of metal and oxide and silicon and how thick they are, how long you bake them for, what patterns you put into them, what the design rules are, how you can move the design rules to a different space, break the design rules by making it a little longer. Or all kinds of different things. So there’s a very, very deep level of device physics and photonics and optoelectronics that can be brought to bear here.

After co-founding One Laptop Per Child with Nicholas Negroponte—hey, he and I made this little nonprofit and it catalyzed $30 billion of revenue for our for-profit partners and changed the lives of 100 million children in the developing world with a $100 laptop. A key part of the electronics was massively lowering the power consumption. And so we needed full custom chipsets for everything. And I designed and architected those. And as I came back to my team, I thought wow, maybe there aren’t enough MIT professors sleeping on the factory floors of the world figuring out how we can use these very slim-margin multibillion-dollar manufacturing structures to make more innovative products and skip a couple of generations.

And I could probably make the same components with the grad students or post-docs at the MIT labs that they could probably make, and maybe once and not repeat them. But if I make them on a $13 billion fab in Japan, once they make one of them they can make a million of them three months later. And I just got hooked on having that kind of impact. And so I left MIT to try to continue to make use of those 2-percent-margin fabs of the world and move them to faster innovation cycles.

Xconomy: Which you’ve done now for Oculus at Facebook and Google X, presumably. And now you’re bringing it all the way back around to your own company at Openwater. So what stage are you out, Mary Lou, at Openwwater? Are you building a prototype? And what do you think will be the first application areas?

MLJ: Yeah, we’re building prototypes. We’re actually spending about a year just building up prototypes, ripping them down and trying in parallel, you know a dozen different approaches when you get to the nitty gritty of the design. We haven’t decided our first product.

I keep looking at my little dog too. I’d really love to know what she is thinking. But following all the right protocols of course. But I’m thinking about that as a lab test obviously.

But the first product—there are so many people hitting us up. I mean, the implications are pretty profound for the billion people in the world that live with debilitating brain disease, be it mental disease or a degenerative disease, and they get an MRI once a year, for example. And the move to having a wearable with continuous diagnostic could be akin to the transformation of diabetes care before you could buy an off the shelf system, where you could do a blood stick and see what your insulin level was at that moment rather than a standard level of insulin every day. You can titrate it based on how you’re doing.

And also we can localize treatment because these systems that we’re creating in Openwater that make your body effectively transparent can do reading and writing. So you can localize treatment in certain areas of your brain or body. You can read things and write things so that as we get to thinking about telepathy, that also has some profound implications, if you think about learning and implanting thoughts and communicating at the neuronal level.

Xconomy: Do you think that you’ll try to develop the medical imaging applications and the thought reading applications simultaneously—that different people will partner with you and take it in both directions at once? Or are the so-called mind reading scenarios much farther out?

MLJ: I think the mind reading scenarios are farther out because of the ethical and legal implications. And the reason that I’m talking about them early is because they do have profound ethical and legal implications. The medical implications of having a bra that can tell you if you’re developing breast cancer are more straightforward, so to speak. I presume the medical applications will come first. But there’s a crossover when you look at mental disease, neurodegenerative diseases.

And that’s something where we will walk through and we are certainly in a lot of different conversations about right now as we’re developing the prototypes, trying to get in front of it. One of the reasons that I wanted to do this in a small company was so that we could talk about it. It’s just difficult for large companies to have programs that are in the very early stages be spoken about outside. It’s too complicated for their PR departments and for all kinds of different reasons.

And it was actually Peter Gabriel, the musician and human rights activist who kept calling me, I think, every week for about six months trying to convince me to take this project, make it into a startup and not do it inside of a big company, at least in the beginning, so that we could talk openly about the ethical and legal implications of this.

Xconomy: What’s Peter Gabriel’s interest in this?

MLJ: I’ve known him for a long time. But I saw him backstage at some conference. We were both speaking and I reconnected with him and he’s just very interested in it and wants it to happen. But wants it to happen in a way where it’s openly discussed with the ethical implications, and we’re figuring out how best to do that. There are a lot of existing organizations that work on this, and we may start a new organization. We’re certainly participating in a lot of the discussions but collectively, globally, our notions of privacy are changing every year right now.

And so the days of a committee sitting in a board room behind closed doors deciding what’s ethical seem to be over. There needs to be public discussion of it because of this rapid change of what our expectations and beliefs are on privacy.

There’s other people working on different approaches to telepathic systems. In fact right now if you look at it as I have, and I think others have, the National Academy of Sciences in the United States said one of the top five things you can work on as a technologist is reverse-engineering the brain. That’s true for many other countries right now in Europe, China, Australia, and so forth. So the question is—scientists don’t like to talk about the ethics as they’re going along. It’s not really in the education system right now.

Yet with all of these top bright minds working on this, what happens if we achieve it? What do we do? What’s right, what’s wrong, why are we not talking about what happens? And I think because so much of it is in the academy, and you don’t want to overstate anything in the papers that you publish for promotion, for getting the professorship. And yet the logical conclusion of all of this is we’re able to communicate with—we’re able to read a book by just tapping the book and it’s downloaded into our brain. That kind of stuff is the logical conclusion of this. And we’re not really talking about what that world looks like.

In many ways we see this sort of mass almost hysteria, I would say, about AIs taking over the world. But what if we do the opposite. I mean, when Marvin Minsky and John McCarthy first coined the term artificial intelligence, Doug Engelbart immediately said, “I don’t want AI, I want IA, intelligence augmentation.” How can we make people smarter. And so this technology makes people smarter.

And so maybe right now we’re really limited. Our input to our brains is pretty good. Our brains themselves are more complex than any computer we know how to make. A hundred billion neurons, each neuron having 100,000 different connections, and we don’t really understand how neurons work. As Paul Allen likes to say, there’s five Nobel prizes just to understand how a neuron works. They’re really, really complicated. So that’s pretty good. The problem is the output. We move our jaws and our tongues to talk or move our fingers to type. And what if we could communicate and dump images and music and thoughts and ideas directly to the computer or to each other, first mediated by computer. Or even, we can put a filter on it. Thoughts that you don’t want to communicate with others, you can filter them. You own your thoughts, you can delete your thoughts. These are some basic tenets that we’re working on. But if you’ve shared a thought you don’t get to pull it back.

Xconomy: You could be looking at the world’s best lie detector, right? No one would be able to mask their thoughts anymore.

MLJ: That’s why we need to teach people how to mask their thoughts. And so, because if the police or military makes you wear such a system—anybody’s system, I think it will be ours, but there’s lots of people working on these types of systems—I was just talking to the kernel.co people, I think they’re working on a totally different approach, it’s invasive, we’re doing noninvasive—but if the police or military, if you have such a hat and the police or the military makes you wear such a hat, it becomes our responsibility to inform everybody how to fool the system.

You have to want to think into the hat for it to work. That’s my goal right now. And be very responsible about introducing this into the world. And that’s, I think, the only way we’re going to release something, is if we can prove to ourselves we have ways to define what it means to be responsible about that and do that.

And I think that there’s been a lot thinking about this for the last 20, 30 years. I mean, I remember Jerry Yang gave those e-mails to the Chinese government.

Xconomy: Jerry Yang of Yahoo?

MLJ: Yeah, and was really surprised that those people were jailed. That was the first time that it sort of hit my consciousness of what could happen. I probably should have thought about it earlier, but it was a very, you know, jarring event. How do you create systems of privacy to protect people. It’s incumbent on us to do so and to see the development and evolution of that over time with the new technologies. Facebook, Google, they all struggle with that and have created some pretty cool systems but yet our notions of privacy keep changing. And so those companies are on the forefront of trying to create systems. They’re also in the crosshairs of evaluation of where that falls.

And so what I’m struggling with as as I start Openwater is, we like to think what’s in our hands is our private space. As we open that up, what are the responsible ways to do so. If you can’t speak, if you’ve just had a stroke, would you be willing to wear it and forego some of your privacy? If you’re brain dead, but not really—forty percent of people thought to be brain dead are not brain dead. They just can’t communicate. Well, what are they? What are the patient’s choices in this? One of the family’s choices, one of the privacy implications is, what are your decision making abilities in that state?

All of these questions need to be answered. I didn’t get my Ph.D. in ethics, so we’re reaching out and working with lots of different really big thinkers in this area and developing that conversation with them is going to take the same kind of time, probably a much longer time, than to just develop the technology. But we’ll wait until we can introduce this in what we believe is the most responsible way we can, and define that as we’re going along.

Xconomy: The longer you stay small, I think I heard you saying earlier, the better chance you have of keeping that conversation going. If you got snapped up by Google, let’s say, or Intel, it might be harder to keep this conversation going.

MLJ: Right. I’ve already turned down an acquisition offer. It’s an interesting thing to have a startup that everybody wants to fund. That’s never happened to me before. It’s a nice thing, but you still have to choose. And we’ll announce in a few months probably, what funding we’ve taken. Right now we can’t really speak about it. But we certainly did decide not to go the acquisition route.

Xconomy: I mean I love talking about the ethics, and I hope a whole fleet of science fiction writers get unleashed on this particular question, and ethicists and others.

But going back to the business—what existing businesses do you think this could disrupt, the Openwater technology, if it works, if it’s scalable, if it’s affordable? Who would be saving so much money from this that they would be willing to pay you whatever it takes? Who are your first customers?

MLJ: The thing is, when you’re reducing the cost of something from a few million dollars—an MRI system costs a few million dollars and a million dollars a year of upkeep, plus a dedicated person or two on the system—to something that at scale hits consumer electronic prices, and is a million times smaller, there’s your answer. I mean it it’s very clear that that enables doctors offices, ambulances, people with conditions that could benefit from more often looking at the inside of their bodies, a variety of conditions from clogged arteries, internal bleeding, to neurological conditions, a torn ACL, to see how things are. Those are some of the early very straightforward missions.

Xconomy: So, radiology departments, other medical offices? Sounds like you’re going to be going up against General Electric in the very near future.

MLJ: No, I don’t think so. I mean, why so? I mean why would somebody want to if you can buy something that is smaller and cheaper. This is sort of the $100 laptop problem, I suppose. Just another order of magnitude or two. The average cost of a laptop is $1,000 or $2,000 and we made a $100 laptop and that was disruptive to a couple of companies.

There was a “60 Minutes” expose once of Intel vs OLPC. It was really rough because some companies wanted to sell things for more money, and some people want to sell the same performance for less money and less size. And so, you know, sorry. Get better at your technology! But on some level the whole industry, I think, needs a shakeup. Right now about 2,000 MRI machines ship ever year and the average price is a couple of million dollars. And the access is very low. People are dying. We have 40 MRI machines per million people in the U.S. You get to Mexico, there’s two MRI machines per million people. You get to a typical African country, there’s one in the capital. People are dying.

And the reason health care in the U.S. and other places has gotten so much more expensive in the last 20 years is because of the technology. There’s a fundamental reason the technology’s gotten better and more expensive. And so let’s look at figuring, let’s make the technology better and at the same time massively lower the cost. You can do both if you take it on. And right now is a really great moment to do that because the consumer electronics industry is hungry. Most big consumer electronics companies, their valuation has halved in the last 18 months because the floor has fallen out of smartphones, which is the number one driver of consumer electronics right now.

And the markets have saturated. And so what you see right now in Asia, which is honestly where most consumer electronics are designed and made, is this incredible hunger that I haven’t really witnessed in the last 20 years or so to find what’s next. So they are willing, the companies that control these massive, very sophisticated factories, to make some process changes. These companies do not do R&D by large. And so it’s very interesting.

If that threatens the makers of the 2,000 MRI machines a year, well, they should get on the train. People are dying. We should be trying, I think, to save more people, diagnose more conditions early and treat more people.

So that’s where I come down on that. And GE, of all people, invented the light bulb. They were a leader in consumer electronics. It’s in their DNA somewhere to be able to move back to this.

Although you know some companies have gotten really good at selling things that cost more than a million dollars. And so it might actually just be more structural. And so there’s, you know, management solutions to that. If you’ve got a sales team that’s really good at selling big expensive stuff, you can get a different sales team and management operation that’s better. There’s a lot of talent around that’s going to other things. GE’s a very well-run company, I’m sure they can figure that out. And the other companies as well. I have a lot of friends at GE, by the way. I think they’re amazing.

Xconomy: And now they’re in Boston, so we love them. A couple of more questions and then we can wrap up. Do you feel like you’re under pressure to meet some milestones this year in order to make the company investable? You said you’re in the enviable position of turning away acquisition offers, and having lots of people come to you with potential funding. But you still have to demonstrate that the technology is real. So what do you feel like are your big milestones for this year?

MLJ: We’re invested. That’s not a problem. Really, it’s not. I just can’t announce what we’ve done.

Xconomy: Okay. So to put it differently, what do you feel like you have to prove in the next year to get to convince people that you really are on to something big?

MLJ: When we release products is when we have to convince people. Right now what we need to do—this is a Sergey Brin-ism—he says the best products are the ones that go through the most iterations. In software you can do that serially because you change some code and you can ship it out to a million people and see how it goes and then scale it out to a billion people at a Google or a Facebook. In hardware you need to do that in parallel, because hardware takes longer to change around the physical components.

And so what we’re doing, our main focus right now, is getting our hardware to go through a lot of iterations, but in parallel, so we build up a dozen different designs and look at the architectures and look at sort of the parasitics and how different different things work together. We’re trying to get the best signal to noise ratio in our system. We’re trying to optimize right now on depth and resolution and scanning speed and figure out exactly what the right combinations of those are.

We’re building up lots of different physics prototypes. See, the thing in hardware is, once you start engaging with a $10 billion factory—think about the amortization cost of a $10 billion factory over four or five years. Think about the per-hour cost of that. Once you’re in that factory you’re on a treadmill and you have to ship or you’re going to get thrown out of the factory.

And so what we’re doing right now is working with some small, little factories that are more patient, building up different designs. In developing the software and systems and so forth, what you try to do is bring all of that together at the right time. But it’s very much a hockey stick and we’re in the part of the hockey stick that hits the puck and not just the part that goes that goes up, going through all of the iterations so that we’re ready for the steep climb.

And this is my fourth startup working with all of these factories in Asia. I’ve learned a lot about how to do it over 20 years experience. And so basically I hire up a group of really extraordinary CTO, architect-type people with really good diplomatic skills, ideally that speak an Asian language. Because when you do hardware, within a month you can have 10,000 people on your team.

Xconomy: What is it like to be Mary Lou Jepsen right now? In a way, it seems like what you’re doing now represents a convergence of everything you’ve been dreaming about for 20 years or more, and all of the skills and experiences you’ve built up at your four companies and at these other large places. It must be pretty exciting.

MLJ: It’s really great. I think one of the reasons that I was able to do this is—I think everybody knows Google and Facebook pay really well, at the level I was at. And so I don’t need the money anymore. So I was actually able to make decisions that I haven’t been able to make for the last 20 years since 1995, because I had a brain tumor. I needed health insurance, which meant I needed a job that had health insurance. I had to give up arts and open source. I had to give a bunch of things because I really needed those things to live. And now I don’t.

I love the work that I do. I am very proud of much of it. If I could have chosen anything, this is what I wanted to do. And so now I get to do that.

I think of Seymour Papert, when we started One Laptop Per Child—we’re in Brazil once before our first head of state, President Lula, decided that Brazil was going to buy two million laptops, and everybody thought we were nuts, it’s never going to work, and I was having dinner with Seymour Papert in Brazil and Seymour called OLPC the dessert course in life. Like, what you do after you’ve done everything else. And I felt like that at that time too, that it was the dessert course in life. But this is, sort, of I don’t know, the cheese course after that.

And it continues to change as you have access to different resources and your life stuff becomes more flexible. When you have enough money to just pay for your own health care—for me that really changed the equation.

Xconomy: That’s such an unexpected answer, but so down to earth and so real. I mean, so many people are in that position of not being able to do the creative thing that they wish they could do because of instability in their income or their health care coverage. And for you to say that, having had having had the chance to work at Facebook and Google really kind of freed you to do something amazing and take a giant leap, that’s great.

I hope that many, many people get to enjoy your dessert course. This has been a wonderful interview. I want to thank you for joining us.

MLJ: Thank you so much for having me.