Dropbox, AirBnb, and Reddit: just three of the companies that began at startup incubator Y Combinator and have grown to have huge impact and valuations. Now the influential Silicon Valley institution is hoping to broaden its influence.

Most of the nearly 1,000 startups that have been through Y Combinator’s three-month boot camp have built Internet businesses of some kind, and more than 40 of them are now worth $100 million or more.

Lately the incubator has also taken on companies working on hardware, biotechnology, and even nuclear energy. And today Y Combinator announced it is setting up a research lab, YC Research, to work on “fundamental” problems in science and technology beyond the scope of startups.

The lab is initially being funded with $10 million from 30-year-old Sam Altman, Y Combinator’s president. The first project has not yet been disclosed but will comprise about a dozen researchers. Altman met with MIT Technology Review to explain how Y Combinator’s startups and new lab are intended to advance technology. What follows is an edited transcript of the conversation.

Why is Y Combinator establishing a research lab?

We’re doing this because funding and the environment for fundamental research keep getting worse. The government cuts science funding every year, academia is badly broken in many ways, and corporate research is not open enough, or too directed toward what that company needs.

If you’re a world-class researcher doing AI research or something you can go work in academia, or Google will tempt you away. We’re going to give people serious equity in YC in addition to salary so they don’t have to make that trade-off. They can come here and do research that’s offered free to the world. This is fully a nonprofit for us. We can fund work on science and technology with a 20- to 30-year time horizon.

How many projects will the lab work on?

We’ll start with one. If that works well, we’ll add more. We’re not talking about what that group’s going to work on right now; we’ll be ready in a maybe a month. The researchers will have full access to YC, they can come to any event we run for our companies and talk with them. A lot of the companies are going to donate data or resources.

Y Combinator is best known for companies like Dropbox and AirBnB, which used technology to grow fast and create new markets. But you’re now funding companies doing what you might call hard technology, like biotechnology and even nuclear fusion.

Dropbox is actually very difficult technology to produce. It’s certainly true that YC funds a lot of thin software companies; I don’t think Dropbox is one of them.

We’re in the business of funding valuable innovation because that’s important for the world and also makes a great return for us. But AirBnB changes the world and nuclear fusion changes the world—and I don’t think one is better or worse than the other.

As an organization your history and expertise is mostly around the Internet and computing. How can you help out companies working on fusion or biotechnology?

What YC gives advice about is not how to do engineering; it’s about how to build a great company. If we’re giving a software company advice about how to write software, it’s done. If we’re giving a nuclear fusion company advice about theoretical physics, it’s done. What we’re really good at is saying: “You’re trying to build a giant company; what do you do on day one, day 80, day 200.”

We try not to be limited by what expertise we might be missing. On a percentage basis I think we generate the best returns of any investor in Silicon Valley. I think that is because we make these long-term, somewhat risky-seeming bets.

Have you been making risky long-term bets for long enough to know?

I would say Dropbox is a great example. Everyone thought that was ridiculous. Everyone thought the whole model of YC was ridiculous, that you would fund these things early on.

The investor Peter Thiel, who helped fund Facebook in its early days, says that Internet and social networking companies have not significantly advanced technology (see “Technology Stalled in 1970”).

Peter is one of my closest friends, and I think on this point he is just like completely off the reservation. It’s easy to look back now from 1939 to 1970 and say, “Look at the progress we made,” and I do think that was an incredible 31 years. But when people look back from 1970 to 2015 they will say that the software revolution, the Internet revolution, is one of the great revolutions in the history of humanity, up there with the agricultural revolution and the industrial revolution.

What about outside of computing, say energy or health? Thiel argues that we haven’t made fundamental breakthroughs there.

Talent and capital sloshes around to where the best return is. We’ll make incredible progress in software and then that will power many other things. SpaceX was able to design rockets that were much cheaper to design and much quicker to design because of software. Both the nuclear companies we’ve funded are able to do their work in software.

Will the failure rate be higher in these areas?

Yeah, of course. If you try hard things you are going to fail a lot. I don’t let this bother me—177 YCombinator companies have shut down. That’s fine, it should probably be higher. But there is $65 billion in market cap of companies that have gone through YC. Most probably wouldn’t exist otherwise.

Some venture capitalists have recently voiced concerns about startups in Silicon Valley being overvalued and spending too recklessly. Are they?

Definitely that is true. I’ve felt valuations have been very high since 2011.

I cannot get my head around how some of these startups spend money. Burn rates have gone off-the-wall bananas. The money that has poured into the system has not made the startups any better. We really try to imbue our startups with this idea that if you have profitability within grasp, and you keep your costs low, you control your own destiny. If you’re dependent on outside capital, then you don’t.