There was a concept a long time ago that you would do a different type of reactor called a "fast reactor," that would make a bunch of another element called plutonium, and then you would pull that out, and then you would burn that. That's called "breeding" in a fast reactor. That is bad because plutonium is nuclear weapons material. It's messy. The processing you have to get through is not only environmentally difficultly, it's extremely expensive.

As we get robots becoming more sophisticated, I think we should worry sooner rather than later on how much they could take over, but I think it'll mostly be a positive thing. In terms of deadlines it won't be any worse than nuclear weapons.

If you rely too much on the people in other countries and other companies, in a sense that's your brain and you are outsourcing your brain.

There's life and death in every email.

When Paul Allen and I started Microsoft over 30 years ago, we had big dreams about software. We had dreams about the impact it could have.

Take agriculture, where we haven't done much, or sanitation; saying, "okay, we will be able to make a really huge effort there." It really energized the foundation and half of what we've gotten done in this last decade is because Warren [Buffett] trusted us.

I would counsel people to go to college, because it's one of the best times in your life in terms of who you meet and develop a broad set of intellectual skills.