Frank Thomson Leighton (born 1956)

Wikipedia 🌐 F. Thomson Leighton

Saved Wikipedia (Dec 30, 2020) for "F. Thomson Leighton"


Frank Thomson "Tom" Leighton (born 1956) is the CEO of Akamai Technologies, the company he co-founded with Daniel Lewin in 1998.[1] Akamai has become the top content delivery provider in the 21st century with the arrival of dedicated techs. As one of the world's preeminent authorities on algorithms for network applications and cybersecurity, Dr. Leighton discovered a solution to freeing up web congestion using applied mathematics and distributed computing.[2]

He is on leave as a professor of Applied Mathematics and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL) at the Massachusetts Institute of Technology (MIT). He received his B.S.E. in Electrical Engineering from Princeton University in 1978, and his Mathematics from MIT in 1981.[3] His brother David T. Leighton is a full professor at the University of Notre Dame, specializing in transport phenomena.[4] Their father was a U.S. Navy colleague and friend of Admiral Hyman G. Rickover, a founder of the Research Science Institute (RSI).

Dr. Leighton has served on numerous government, industry and academic advisory panels, including the Presidential Informational Technology Advisory Committee (PITAC) and chaired its subcommittee on cybersecurity.[5] He serves on the Board of Trustees of the Society for Science & the Public(SSP) and of the Center for Excellence in Education (CEE), and he has participated in the Distinguished Lecture Series at CEE's flagship program for high school students, the Research Science Institute (RSI).

Awards and honors[edit]

In 2018, Leighton won the Marconi Prize from the Marconi Society for "his fundamental contributions to the technology and establishment of content delivery networks".[6] In 2017, Leighton and Lewin were inducted into the National Inventors Hall of Fame, for Content Delivery Network methods.[7] He was the first winner of the Machtey Award in 1981 and is a Fellow of the American Academy of Arts and Sciences and a member of the National Academy of Engineering. In 2008, he was appointed as a member of the United States National Academy of Sciences. In 2012 he became a fellow of the American Mathematical Society.[8] He received the IEEE Computer Society Charles Babbage Award in 2001. He was elected as an ACM Fellow in 2018 for "his leadership in the establishment of content delivery networks, and his contributions to algorithm design".[9]


  • Introduction to Parallel Algorithms and Architectures: Arrays, Trees, Hypercubes (Morgan Kaufmann, 1991), ISBN 1-55860-117-1.

  • Complexity Issues in VLSI: Optimal layouts for the shuffle-exchange graph and other networks, (MIT Press, 1983), ISBN 0-262-12104-2.

  • Mathematics for Computer Science (with Eric Lehman and Albert R. Meyer, 2010)

Frank Thomson Leighton


October 28, 1956 (age 64)



Alma mater

Princeton University (B.S.E.)

Massachusetts Institute of Technology (Ph.D.)


Marconi Prize (2018)

NAE Member

NAS Member

AAAS Fellow

Scientific career


Applied Mathematics


Akamai Technologies

Massachusetts Institute of Technology


Layouts for the shuffle-exchange graph and lower bound techniques for VLSI (1981)

Doctoral advisor

Gary Miller

Doctoral students

Peter Shor, Mohammad Hajiaghayi, Robert Kleinberg

2011 (Sep 13) Interview - F. Thomson (Tom) Leighton PhD ’81

INTERVIEWER: Today is September 13, 2011. I'm Chris Boebel. As part of the MIT150 Infinite History project, we're ta

Transcript (below) - [HE003S][GDrive] ( video - )

INTERVIEWER: So let's start at the beginning, as it were. Tell me a little bit about where you grew up, how you grew up, your life as a as a kid.

LEIGHTON: I grew up in Arlington, Virginia, right near Washington, DC. My dad was in the Navy, worked in the nuclear power program, and then continued that work, working for Admiral Rickover as a civilian when he got out of the Navy. My mom spent most of her time raising us kids and then became a librarian when they needed to get some extra cash to be able to afford college for myself and my brother. It was a nice environment for growing up. Parents cared a lot about us and a lot about our education. Made sure we had all the best opportunities they could afford for us. Went to good schools. My high school, Washington-Lee High School, was a very good public school. That's before the days when they had Thomas Jefferson in the Alexandria area, which is now a magnet school for all the science and math talented kids. And then went to Princeton for college. Majored in electrical engineering, computer science, and then found my way to MIT for graduate school in applied mathematics.

INTERVIEWER: So tell me a little bit about growing up as a kid. Were there moments that stand out in your memory as realizations that you had a particular aptitude for math or science?

LEIGHTON: I always just loved mathematics and really liked science. As far back as I can remember, there was no question that's what I wanted to do. I suppose when I was little enough, I wanted to be a train conductor. But once I got past that stage, I just wanted to do math, and science related to it. I didn't really know what that meant in terms of careers until much later. But it was very natural for me to evolve along the graduate student research track, become a postdoc, become a professor, because I just liked doing that stuff. I had a lot of help as a kid, starting with my parents, of course, but also through organizations like science fairs, the then Westinghouse Science Talent Search, now the Intel Science Talent Search, the International Science and Engineering Fair, teachers that went out of their way to help, and actually people in industry that went out of their way to give me special access to large-scale computing back then. Of course, the computers back then fit in my wallet today. But it was a special to have access when I was a kid to the old UNIVAC machines or whatever. And I was able to use the computing cycles to do research on number theory, which I enjoyed a lot as a kid. And that helped encourage me in the direction of mathematics and applied mathematics and computer science in later life.

INTERVIEWER: So when you were a senior, you were a finalist in the Westinghouse Talent Search, which I recall from my days in high school-- I think I went to high school before it was the Intel Talent Search-- was a pretty big deal. Tell me about that experience and how it shaped you.

LEIGHTON: It was a fantastic experience. You get to meet 40 other kids, or 39 other kids, with similar interests. It's where I met Eric Lander, who's probably somebody else you've interviewed here. In fact, that year Eric won first place, and I won second place. And of course, I've known Eric to this day. I met other folks there, in fact, including one of my college roommates my first year at Princeton. And I know several of those folks today. We're still friends. So it's a wonderful experience to get together with other kids who have similar interests in science and math and who really care about it. And so today, in fact, I'm now a trustee for SSP, which is the new version of Science Service, which used to run the Westinghouse Science Talent Search. And today, Science for the Society & the Public runs the Intel Science Search. So it's a chance to give back a little bit in thanks for the impact that they had on my life.

INTERVIEWER: I also wanted to follow up on something you mentioned, this sort of early exposure to computers and early programming experiences. Tell me a bit more about that. Where did that happen? What kinds of problems were you dealing with? I have a vague memory of this myself. Were these stacks of punch cards you were working with?

LEIGHTON: I am old enough to remember the paper tape, believe it or not, in my junior high school. And we were pretty fortunate to have any computer in junior high back then. And then it evolved into punch cards. My science project then was looking at some very famous conjectures. The first is Goldbach's conjecture. And about 300 years ago, Goldbach conjectured that every even number is expressible as the sum of two prime numbers. A prime number being a number like 3 or 11 that has no factors other than itself or 1. So for example, 12 = 5 + 7. And he conjectured that it was true for all even numbers, you could write them this way. And so I was trying to prove that. Now, I didn't realize then I had no chance of proving that, because a lot of serious mathematicians have been working on this for centuries. But I did computer studies that gave evidence to a theory of if you viewed numbers probabilistically, that the bigger the number gets, the more representations it has as the sum of two primes. And if you argue that primes are somehow random, then you could make an argument that says, yeah, it should be true for all of them. That's not a proof, but it gives-- today I'm not even sure I would say it gives evidence that it's true, but back then I thought it did, as a high school student. But doing the analysis involved programming with large computers back then to get this evidence that would back up the theory. The other problem was, again, a long, open question of number theory, which is that there's an infinite number of pairs of primes, p and p + 2, where p + 2 and p are both prime. They're called twin primes. And I was trying to give evidence for the fact that in fact there were an infinite number and that, in fact, for any arithmetic sequence of primes, either the pattern occurred once, or it would occur an infinite number of times, which was a new conjecture. I'm not sure anybody really picked up on that conjecture, but it's something I worked on in high school. And it got me introduced to computers and what they could do. And I learned how to program. And that's where people in government and industry reached out to me and gave me access. I remember a fellow named Carl Hammer who was the senior chief scientist at UNIVAC in Washington way back when. And because he was the chief scientist of a big computer company, he had special access to state-of-the-art equipment, and he lent me that access, which was wonderful. Here I am, 16 or 17, and I had access to the state-of-the-art computing back then.

INTERVIEWER: So what about the importance of those kinds of mentors or that kind of encouragement in encouraging your early career and, I guess, by extension other people's careers too? We'll talk later about your own work as a professor and a teacher.

LEIGHTON: I think mentors are incredibly important. They become role models. You want to become like them. You see career paths. And they help you. They encourage what you're doing, which, I think, is good for kids to be thinking about math and science problems and spending a lot of time working on them. I think that's really great. It helps launch them on a track as a researcher or as a scientist and mathematician. And they tackle real problems more seriously later in life. And it can make a difference that way. So I think mentorship is very, very important.

INTERVIEWER: Were there other important mentors to you that we haven't talked about, maybe in your high school years or at Princeton before MIT?

LEIGHTON: Yeah, I think there's a series of people. I remember there was a teacher of graph theory at Princeton. And he just was a great teacher, and he had a love for the field. And it becomes infectious. And you just love working extra hard to show him what you can do and to be able to have a dialogue with him. So it's very motivational. Here at MIT, there were some excellent professors-- good teachers, good advisers. And you want to become like them when you grow up. And that makes a big difference. It helps you evolve into the next level by seeing people that you have a lot of respect for and that are really good at what they do.

INTERVIEWER: So at Princeton, you majored in engineering. Tell me a bit about that, how that became your focus rather than mathematics or other directions you might have gone.

LEIGHTON: Well, a couple of things. First, I was afraid of real analysis. And if you were a math student, you had to take real analysis. Math at Princeton was a little scary. Pretty tough. And it was very pure. And my interests were more on the discrete side of mathematics, which MIT is great at, the world's best at. They had their token discrete mathematician, who's the guy who I liked as my mentor at Princeton at that time. And I found that the kinds of things I liked best actually were done in the computer science or EECS department back then. And in fact, that's where I started learning about theoretical computer science-- algorithms, complexity theory. And you can almost think of it as a branch of discrete mathematics. It's in computer science, but the tools are all mathematics, and the kinds of problems are all discrete mathematics problems, combinatorial kinds of things. And so I really gravitated there, and they had a very flexible policy. You could do pretty much whatever you wanted. And so I just ended up going there for my degree and ended up getting lots of graduate courses and just taking everything they had, because it was so much fun. It was really great stuff.

INTERVIEWER: And then you made the decision to come to MIT for graduate school. Tell me about that.

LEIGHTON: MIT is the best place in the world for discrete mathematics and the best place in the world for theoretical computer science, so it was a perfect fit. And when I chose between the two, I decided when I came to MIT to do the applied math track. So I didn't do math as an undergrad but did do as a graduate student. Also, I had some encouragement from my father when I was an undergrad to go the engineering route. And somewhat joking, I think he felt that if you're doing mathematics, you'll never get a job. Engineering, well, that's okay. He himself, of course, was an engineer. I'm exaggerating a little bit. But then, by the time I got to graduate school, at that point I'm doing research anyway, and it doesn't really matter anymore to me. I'm not thinking about getting a job. I'm thinking, okay, I'm going to be a professor. Well, that's a job, but it's not the same kind of job. And so let's do the applied math thing, because that's where folks like Danny Kleitman, Gary Miller, Len Adleman, Richard Stanley-- those guys were all in the math department. And that was just a fantastic fit, because I could do everything. I could do the math I liked. I could do the computer science I liked in that department. It was very flexible. And I had a great time doing that.

INTERVIEWER: So what year roughly did you come to MIT, and what was the environment like at the time just generally on campus but also within your department?

LEIGHTON: I came in '78. I did notice when I came, there was tension in the department between within the applied math between the discrete side and the continuous side- - the guys that do partial differential equations and so forth. And this was relevant because we had to pass our oral exams. And the prior year, all the students on the discrete were failed. And being on the discrete side, that was of some concern. And I do remember that was a big issue, and there was a tension between the two sides of the department. Now, I think they since got beyond that. I think things on campus, I don't really have any recollection one way or the other. It was a fun experience. It was good. I really liked it. I could do research. Great faculty. Good colleagues. That was long past all the stuff with kicking ROTC off campus and all those kinds of things. So the campus, I think, was very peaceful when I came. The question was, are you going to win your intramural softball game this weekend? That was the extent of it, as I remember anyway. Maybe I was just oblivious.

INTERVIEWER: Is there a way of explaining for someone like me discrete versus continuous?

LEIGHTON: Continuous is calculus. You do an integral for discrete mathematics. You compute the area under a curve-- sorry, in continuous mathematics. In discrete mathematics, it's summing a series of numbers. What's the sum of 1 + 2 + 4 + 8 + 16? That kind of thing. You have differential equations on the continuous side. On the discrete side, you have a recurrence, that the running time for the algorithm on an input of size n is twice the running time it takes for an input of half the size plus n more steps. And the techniques really down deep, there's similarities, but the actual day to day is different from continuous mathematics and discrete mathematics. Counting things is discrete. How many ways are there to put five red balls and six green balls and pick from them-- that kind of thing is discrete. Continuous is complex analysis, real analysis, calculus, and stuff like that. Proof methods are different. You use induction when you're doing discrete mathematics. You use different approaches when you're doing pure or continuous mathematics.

INTERVIEWER: And what is it about discrete mathematics that you found attractive and appealing, that kind of clicked with you?

LEIGHTON: I found I could get my arms around it better. The definition of discrete mathematics is it's dealing with numbers, things that are countable. Doesn't mean they can't be infinite, but they're very constrained in how they're infinite, like 1, 2, 3, 4, 5, 6 to infinity. That's very concrete, get your arms around. Continuous mathematics, there's an uncountable number of points in the unit sphere. I found it conceptually more challenging to get my mind around it. I liked it. It was good, but I liked the discrete side much better. Also, when you get into the applications, things like computer science and theoretical computer science, it's the discrete mathematics that matters. And so you can have a chance to have impact with your applications from discrete mathematics. Structures like graphs, points and lines, or points and edges that connect them, nodes, arcs to represent communication problems or network flow problems-- that's all in discrete mathematics.

INTERVIEWER: And what kind of problems really attracted you as a grad student? What kinds of things were you working on and really drove your interest?

LEIGHTON: I think they primarily revolved around graphs-- using networks to model things and proving facts about them. Some were just pure graph theory. And most came up in the context of a computer science problem of some kind. Ultimately I wrote my thesis on problems-- the field then was called very-large-scale integration, or VLSI. And in practice, that referred to the design of tiny microprocessors on chips and how to lay out the transistors in the wires to fit as much as you can on the device. In my mathematical version of that world, it was all about taking a graph and embedding it in a grid in a way that was efficient somehow. And that's pure mathematics at that point. So that was sort of the focus, the sweet spot when I was a graduate student.

INTERVIEWER: And who were you working with, both in terms of faculty and even other grad students at that time? Who leaves an impression in your mind as being a particularly important influence?

LEIGHTON: Gary Miller was my adviser, and he was a great adviser. He was very giving, very unselfish. He would get me good problems to work on. He'd encourage me. He'd listen to me. He'd give me good ideas. He wouldn't even try to put his name on the paper. He wouldn't try to take credit in any way. He was just a really good guy. And that's a great role model. If you have nice parents, better chance you're going to be a nice parent yourself. So that was great. Danny Kleitman, fantastic influence. Again, super guy. Really nice. Very smart. Great to talk to. Great problems. Very encouraging, supportive. Always putting the student first. So they were probably the two largest influences on me as a graduate student, and they were both in applied math. Len Adleman comes to mind. I worked a little bit with Len. He left MIT pretty early on. He was a great teacher. And so that's the kind of thing you try to emulate on the teaching side of things. He made material clear. He made it be interesting, made you want to think, wow, this is important. I want to work on this stuff. And of course, you got guys like Ron Rivest. Very famous, doing huge things. Sort of a funny story. Today there's a thing called the RSA algorithm. It is used for the encryption on everything you do online. So very important thing. Early on, it was very revolutionary. Nobody thought that, oh, that can't be a good idea to base an encryption protocol on the difficulty of factoring numbers. Before it even became public, when I was a undergrad, I was at the then National Bureau of Standards, now called NIST. And because of the Westinghouse Science Talent Search, I got a summer job there. And I was working with a mathematician there. And the RSA paper was sent. The government got their hands on it, and they wanted to evaluate it. And so it went to the mathematicians at the National Bureau of Standards. And so my summer project was to evaluate the RSA algorithm. It's a little scary the government is relying on some undergrad in a summer job to do the evaluation, but it was really cool. And then later, I come to MIT years later, and, wow, there's Ron Rivest. Pretty cool.

INTERVIEWER: Did you share your evaluation?

LEIGHTON: My evaluation was that I sure as heck didn't know how to break it, that it wasn't clear that anybody would know how to break it. But also there was no proof that it was unbreakable and that if we relied on it, people would try to figure out how to factor better and that this could cause a problem. And in the end, people have learned how to factor better. And so instead of using hundred-digit keys they were doing back, now you got to use thousand-digit keys, because people have got better factoring algorithms. And still nobody knows how to break it, but there's still no proof that they won't figure out how

INTERVIEWER: So it's just a matter of staying ahead?

LEIGHTON: Today? Yeah, that's right. But a lot of people have been thinking about it now, so factoring seems pretty tricky.

INTERVIEWER: So tell me a little bit about making the transition then to being on the faculty at MIT.

LEIGHTON: That means you've got to teach and you've got to get research grants. Now, I had, I would say, a very cushioned experience joining the faculty, because I got a postdoc first. And it was a two-year postdoc. And after the first year, I joined the faculty, but I had the postdoc, so it meant that I didn't have to teach, but I was on the faculty. So that gave me a breather to help adjust. And then when I did start teaching, the field, the community at large, started getting pretty excited about the field I was working in-- large-scale networks and doing large-scale parallel computation and distributed computing. The area I had gravitated to became hot out there. And so I got to create a graduate course on that material. So it was based on the field that I cared most about, and it took me a ton of time. I spent a week for every hour and a half of lecture. But it was very worthwhile for me, because it was my field of interest. And you never really learn something until you have to teach it, at least I find. So it gave me a chance to really learn even at a deeper level what was happening in my field. And so a lot of good research came out of the fact that I was teaching this class. Students that were interested in doing research in this new field took the class, ended up doing research with me because I'm the teacher of that class and designing it. And so it became highly productive, as opposed to my teaching a class that's unrelated to my research interests that takes a lot of time to do a good job but doesn't provide any side benefit in my research program. So it worked out very well for me. It helped boost and accelerate my research. And it worked out, I think, good for the students at the time and probably for, I think, the department, because now they get a new field being taught. And students are actively interested in it, and research is happening at MIT. So it was a very good experience moving into teaching and the faculty position at MIT.

INTERVIEWER: So you mentioned that your research topic became hot, which clearly is related to the rise of computer networks and the importance of computer networks and ultimately the internet. I was wondering if you could talk about that, especially the internet, and at what point the research you were doing sort of became of paramount importance to that.

LEIGHTON: Yeah, that's interesting. I'm a theoretician. I certainly was a full time theoretician before. And so the quality of my work was judged based on the depth of the mathematics and the perceived importance of the problem that's being solved, not based on its perceived applicability at all. I'm in a math department. I'm in a field of theoretical computer scientists. The conferences I go to are other people like me, and we're all mathematicians underneath the covers. Now, the field itself talks about large-scale networks and doing things like routing communication problems on those networks, avoiding congestion. And so it's not a stretch to say, hey, this might be relevant to practice. And so the first brush with practice I think came with the advent of companies like Thinking Machines, which is sort of an MIT spin-out nearby here, that were actually building large-scale parallel computers and starting to use the ideas that were in these theory papers that we'd developed. So a close colleague of mine, Charles Leiserson, actually spent a lot of time at Thinking Machines. I consulted there and taught courses there in the summer with him. But that started getting a brush with practice. Unfortunately, Thinking Machines went broke, and the practical side of that field sort of faded away. There never really was a impact in practice, I would say, to the work in the '80s and early '90s. Then the next brush came actually through encouragement from DARPA. DARPA was was, and probably in some sense still is, a big funder of research in computer science at the leading universities and at MIT. In fact, back in the early to mid '90s, it was probably 2/3 of the LCS research budget was DARPA. And theoreticians always had a hard time getting research money, because we're doing math, and DARPA's interested in things-- they're out there. They like research, but they want things that are going to really be applied-- the guys building systems and stuff like that. There was always encouragement in the lab for computer science for guys like me to at least look a little more practical so that you could get funded. So I'm doing work on very large-scale networks. People are waking up that the internet's an important thing. And gee, some of the stuff we're doing seems like it might be relevant. And so I did take a step a little bit to the right and start thinking about, with the encouragement and support of guys like Tim Berners-Lee, who's the father of the web, to start thinking about problems directly in the context of the internet. Using the mathematics from the large-scale network work, how would you now tackle the problem of routing around congestion in the internet? How would you solve the problem of getting a hot piece of content or piece of news to millions or billions of people in a short amount of time? And this is the kind of problem that was right up our alley in terms of expertise-- my group and the stuff we were doing. And well, boy, if we worked on that and made some progress, we might actually get funded. So we did. And we did get funded. Now, we weren't really thinking that we're going to change the world at this point. That started happening two years after the research began. The research, we're still publishing in theoretical conferences. It's still being judged based on the mathematical depth of the work. But there's this thing at MIT called the 50K contest. And my lead graduate student on this stuff, Danny Lewin, had a next-door neighbor who was a Sloan student who knew about the 50K. And they were talking, and Danny's going broke, because he's got all these student loans, and he's a theoretician. And we didn't get paid that much, especially when you're a graduate student. Kids in private school. So Preetish talked to him about taking the work in his thesis and turning it into a business plan for the 50K competition. And so then Danny talked me into engaging with them on it. And we thought it would be a good learning experience. And so we entered the 50K process, starting with the three of us and ultimately growing to 35 people. And through that experience, we really got exposed to the real world. It's hard to imagine now, but I didn't know what a VC was. I didn't know it stood for venture capitalist, and even if I did, I wouldn't know what venture capitalists did. It seems shocking today. We learned. But as part of the 50K process, we met VCs. We met industry pundits. We ended up talking to potential customers for our pretend business plan. And so we got exposed more to the real world, and we began to understand that wow, our technology really could work in practice and make a difference. And that was a first, a fundamental change. Because people who do mathematics and do theoretical computer science, that's a very rare outcome. Now, over the history of the field, there's been huge advances in practice from work on theoretical computer science. It really is a vital field to the economy and to the industry. But a lot of the work takes a long time before it's relevant, if ever. And this was sort of an eye-opening experience for us to see that, wow, we might be able to do something in the real world.

INTERVIEWER: So I know that at one point, with the work that ultimately became Akamai or on which Akamai founded, at one point you had attempted to approach various ISPs, internet service providers, and interest them in the work and had kind of gotten the brush-off. Was this before or after the 50K experience?

LEIGHTON: This is during and at the end of the 50K experience. The original idea of our plan was to take our technology and, well, in the business plan, sell it to ISPs with the idea that they could get better performance for their subscribers and lower their cost. Now, we actually went to go talk to some ISPs. We went to Coolidge Corner to talk to HarvardNet. We talked to several of the local ones, called up UUNET down in Reston. And we quickly learned that they had no interest in what we were talking about. They said, look I can get caching software for free. I got much bigger problems to worry about. I'm going broke. And distributed computing, everybody knows that doesn't work in the real world. That's an ivory tower concept. Please go back to your ivory tower and let me get back to saving the company. They're all up in flames. They're in the process of going broke. That was not encouraging, so we modified our business plan over time. At the end of the 50K, we decided not to form a company. We were actually approached by some VCs that said, hey, look, you didn't win the contest, but we liked what you're doing. It's MIT. It's sexy. It's the internet. Let's go. Let's make a company, and in six months, we'll flip it. We're all going to go home rich. And again, showing you how naive we were, we said no. We didn't want to do that. I liked being a professor. Danny wanted to become a professor at MIT. And he was smart enough to do that. The business plan had been fleshed out. And we didn't know anything about really starting a company. The 50K thing had been as a learning experience and for fun and a challenge. We did keep thinking about it. We did have to decide about are we going to patent things or open source it? I remember having discussions with Tim and Michael Dertouzos, head of the lab, about that. As part of that, we talked about hey, can't we just get the companies to use the technology? Because we wanted it to be used. And even if we made it free, they didn't think it would work. They think they had other things for free that if they needed to go there, they could go there. And there just was no takers. And so at the end of the day, we decided that this is our chance to make a difference with technology. It doesn't come along very often in what we do. And we'd modified our business plan, changed it around, so we felt it could work, changed it to selling to people with websites-- the big websites of the world. Talked to some of them, and they started expressing interest, which was better than we'd had before with the ISPs. And so we said, the only way we're going to do this is to do it ourselves, so let's give it a try. So we spun out of MIT at the end of the summer of '98, a year after the 50K started and got offices over in One Kendall and then called up those VCs and said, hey, we could use some money. Turned out to be harder than we thought to actually get the money on terms we wanted, so we had a period of about three to four months where we were funding it ourselves with friends and with some angels, professional angels. We had friends of friends we knew to bridge us to the first round with the VCs.

INTERVIEWER: So stepping back for just one minute, what was the big idea behind what ultimately became Akamai? And why do you think there was so much resistance among those initial people, the ISPs that you approached? You mentioned sort of this ivory tower attitude that they had. But just expand on that a bit.

LEIGHTON: It's not one idea per se, but it's an approach. And the approach is mathematical and with depth in technology. And that's not the way any of those guys would approach it or even understand it. I still remember-- it seems funny in retrospect. Today I'm a professional salesman. I go out, and I give talks to customers or at keynotes at major conferences and explain the future of the internet and stuff like this. Back then, we would go talk to people. I'd get out my deep technical slides on here's the proof that consistent hashing, our brilliant new algorithm, does all these great things. And I'd start giving them a proof, mathematical proof, to the VCs. They didn't have a clue what I was talking about. out But somehow they had some belief that we were going to do something good. So it was based on deep technology. Not the deepest math ever done for sure, but in terms of this field, bringing some depth in mathematics to the field, that was different. It used distributed computing. It used algorithms. The algorithms were expressed as a function of n. n? What's n? So it's just very different than the way they would think. They'd build something for size four and then build a new one for size eight. And here we're just thinking at a whole different level. Ultimately it is that capability which made us be so successful, I think, in addition to getting really strong people who worked really hard. But having that technology, I think, made a real difference and still drives the company today. It allows us to give a lot better services than the dozens of competitors that have sprung up over the years. So I would say it's not so much one idea but a philosophy. One person I remember-- actually a well-known person in the field-- said to us years later, the only reason we were successful is because we were too naive to know that what we were trying to do was impossible. Because we were naive, and we did things out of the box. And we weren't constrained by the thinking in the industry at the time. Because we weren't even practical people. Here we are now building systems without a clue how you're supposed to do it. That means we made a lot of mistakes, but we also did things differently, because we were mathematicians building a large-scale system. And so we came at it from a very different angle, and that helped in some ways. Today, of course, we've evolved to the point where we have the world's best systems developers working with the mathematicians to develop our systems. You need both to really do it right. So it was sort of a philosophy, almost a naivete that we did things differently. We had a very open mind about how to approach the problems. It used things like distributed computing, which really weren't used and practiced so much back then. And we didn't know even at what level to explain things to the ISPs. Me going in and giving some proof to the guy at the ISP, I laugh at that now. No wonder they wouldn't be interested. It was what we were doing.

INTERVIEWER: I also wanted to just ask a bit more about the 50K experience. I guess first of all, for people who might not know what the 50K is, maybe just provide a little bit of context. But also you mentioned you didn't win, and yet it was also a valuable experience. Just tell me a bit more about that.

LEIGHTON: Yeah, the 50K contest, it's now called the 100K because the prize money has been increased. It's run by students in the Sloan School. And you submit a business plan. Actually there's several rounds. The first round is more of an elevator pitch. It's a one-page thing. But ultimately, as you go towards the later rounds, you flesh out a business plan and submit that. And they pick the ones they think are going to be the most successful, most likely to make money. And they use the money to fund the company. Now, in our case, we had no intention at that time of forming a company. It's a pretend exercise. And we were a bunch of folks-- Danny Lewin in particular-- that when you enter a contest, you like to win. So we got books out of the library, how to write a business plan. We didn't let our lack of experience hold us back. And we competed hard. And when we got to the final round, we saw the other finalists, and we just saw how much better they were than we were. They were very professional in their presentations, just a whole different level or two levels beyond where we were. And that was a great wake-up call, because we realized, wow, we don't know how to do it the way they do it, but we know we got to get to that level if we're ever going to be serious about it. So a slap in the face like that was good for us. At the time, we decided not to form a company. But when we changed our minds later in the summer and did start the company, knowing what we were missing was very important. And knowing what we weren't good at yet was important. Because then we would either get better at it or go get people who were really good at it. Now, in the case of the ultimate leadership of the company, both Danny and I decided that we weren't going to be the CEO. I told Danny, I'm the professor. He's the student. I told him I'd work for him if he wanted to be the CEO. I had tremendous respect for his capabilities. Tremendous leadership. Tremendous smarts. Driven guy. But he didn't want to be CEO either, because he was smart enough to recognize that he wanted to go get somebody really, really good. And we did. We got two people who are really, really good. We got Paul Sagan, who started Road Runner, and then George Conrades, who was number two at IBM, had gone to BBN. Bunch of academics there. Converted them into a real company. Sold it to Genuity very successfully. And now is an entrepreneur in residence at a local VC. So because of that experience, I think helped us understand where we needed other stuff, other people, or to improve ourselves. And we went out and got the best.

INTERVIEWER: So about a year after the 50k, you decided, okay, we're going to dive in and form a company. How did you manage to attract those kinds of people to something that was such an early startup, something that was the sort of getting bootstrapped? That would clearly seem to be a very early indicator of success. I'm just curious about how you did bootstrap it.

LEIGHTON: It was MIT. It was the internet. It was sexy. It was the bubble. We had a lot of things in our favor. We met some VCs through the 50K. We knew some people to start calling. We did research on who else to call. So the timing, the environment was great for that. Now, we also had really smart people, not in the business necessarily, but at algorithms and computer science. And I think the VCs sensed that. I think they also sensed we were pretty clueless about business. But that's their job is to help people talented at something get the help on the business side they may need. And I think they sensed that we were motivated and committed. We wanted to have the technology be used in practice. Once we made the decision to do it, we wanted to make it be successful. We wanted it to work. And I think that helped a lot. Now, Battery Ventures, Todd Dagres was the lead for us there. He became convinced that it was a good investment for him to be engaged in. We had a second VC as his partner in the first round. And the very last minute, as the funds are supposed to be transferred-- we'd had the handshake, the signed term sheet, the deal all worked out, all ready for the final signature and the $4 million to get wired. That was happening at the Monday morning. On Sunday night, I get a call from the managing partner of the firm letting me know he's pulling out, because he just didn't think it was going to work. And that was devastating for us, because we'd already said no to all the other consortia that we'd negotiated with. And we were scared to death that that could put Battery in the driver's seat to really screw us, because we really needed the money at that point. And to their credit, Todd said, we'll go through with our share of the deal, same terms, and we'll help you find somebody to fill it out. Very important. You hear all these bad stories about VCs. That was great. So they did help us, and we went and got Polaris to fill out the round, because George Conrades was there. And we made this list of people we'd like for CEO. He was the top of the list, of your fantasy list. And he was there. And we said, okay, look we'll take Polaris if you put George on our board. And then we got him on our board. And then we seduced him into being the CEO.

INTERVIEWER: This is probably a good time for me to ask sort of the larger question about this mind-bending transition from being a professor to being an entrepreneur, to sort of be dealing with all of these things-- VC funding, and first round, second round, finding a CEO. What did that feel like?

[... continued ... Transcript - [HE003S][GDrive] ]