PAUL SAGAN, President and COO, Akamai Technologies, Inc.

MODERATOR: Hope everybody has enjoyed the day so far. We have a little bit left. I'd like to introduce our afternoon Keynote, a guy who's back to some home turf, a Northwestern alum. Paul Sagan, President and CEO of Akamai. (applause)

SAGAN: Good afternoon. Thanks for having me and for staying on. It's always dangerous to be near the end of a conference. I guess the bad weather probably will keep you all here a little while longer.

Things have changed a little bit. There's a restroom back there and I tried the men's room door and it's locked, and so of course without thinking about it I walked right into the women's room and I heard a voice, "there's a man in here." I said, "OK, now there are two." (laughter) I think things are much more conservative. In the ‘70s they weren't even labeled, they were all co-ed, so things are a little duller I guess.

I want to talk to you about Akamai and it's a pleasure to be here. They asked me to tell you about the company so I'll give you the pitch. I'll try to be brief and leave lots of time for questions, so if you want to talk about other stuff about the Internet, I'd be happy to address it when we get to the end. Akamai is a fairly new company. It's gotten a fair amount of notoriety recently and I'd like to tell you what we do, how the company was created, and why we think it's significant to the Internet.

Our tag line in marketing the company, not so much in advertising but we kind of go out and explain who we are, is rocket science for the Internet. Our company is based on math, comes out of years of research in the lab at MIT, and basically algorithms are what enable our business, and I'll explain that.

Our mission is the high performance delivery of content and applications over the Web.

Quite simply we do this. We distribute thousands of servers around the world at the edge of the Internet and we link them in our own, if you will, private network with IP connectivity, and we go to Web businesses, and they pay us to move their content from their sites out to the edges and to deliver their content and applications more reliably and more quickly to end users than sites can do it themselves. And we do it in a totally unique way using technology that no one else has that we think is virtually impossible to replicate.

And we make the Web work better. Essentially solve the problem of erratic delivery and slow delivery of Web content. And we think that that's a huge market opportunity and solves a big piece of the problem.

Our business plan is very simple. We think of it as a virtuous circle and it has only three components.

We try to get customers, lots of them. Web businesses like Yahoo and CNN and Martha Stewart and Brittanica.com, and the more customers we have the more valuable we are to networks. And we try to extend our servers across as many networks everywhere in the world as possible.

And the more customers we have the more valuable we are to networks because they all want great content inside their network. And so we take lots of content to networks and we get more and more networks, which makes our customers happy because we have broader and broader deployment of our technology, and we mix in our secret sauce. New development out of what we call Akamai Labs, and then we've got broader network deployment and better technology, which leads to better services, which gets us more customers. And we have more customers, we'll get more networks and around and around. And that's our basic business model. That's what we do and I'll describe it.

I'm going to tell you a little bit about how I got there. About 20 years ago when I was here, I foresaw the development of the Worldwide Web and the Internet so I majored in journalism to bide my time for 15 years (laughter) until everybody caught up. And I, an undergraduate here and minored in Poly-Sci and Economics, and I had a lot of fun doing journalism.

I worked at CBS for almost 15 years, and then at TimeWarner, and had like most people on the Web and accidental evolution to doing businesses on the Internet. But I had the lucky opportunity to be on the development team that launched one of the first sites in the world with advertising, and I was one of the four founders of Road Runner, the world's first cable modem service.

And what I came to understand, now first building Websites was that the first model was important. You had to have a Web site that performed well, and in the early days it was real easy. We thought. This is amazing. We've got a Mac and a computer line and a phone line and we're on the Internet and we can serve everybody.

Well, we learned that you couldn't scale that simply. You had to build bigger and bigger Web sites, but we learned a lot about building the first mile, if you will.

And we thought, this is great. Wow, we get horrible delivery at the other end. It's about the last mile. You've got to get faster last mile connections, which is one of the things that let me and a couple of other people to launching Road Runner.

And then we discovered a really ugly problem. For all the issues of making great Web sites and solving the last mile problems, there's a third component to the Internet, which I think of as the middle mile. It's everything in between. It's actually the Internet itself where the inner connection of all the networks that make up the Internet. And you have to solve problems in all three places for the Web to scale, for it truly to be a global marketplace open 7 x 24 for real time commerce, real time news and real time entertainment.

And there are lots of people who work on solving first mile problems. Database providers, software, massive hardware providers and consultants. And there are a lot of people working on the last mile.

Well, Akamai's mission is to solve the middle. It's to untangle the mess created by the way the Internet was built. Because actually, it's not built for what we're all trying to do on it.

And that's what I'm going to describe to you today. How we do that. What we do in the middle of the Internet between the Web site and the last mile, how we solve the problems in between.

Akamai is very young even by Internet standards. It was actually founded in the fall of '98, incorporated in September'98, funded Thanksgiving of '98, so we're about a 15-month-old company.

But what's very important to understand is that we -- the foundation of the company is based on patent pending research that was begun five years ago in 1995 at the Lab for Computer Science at MIT.

And there was a very fortunate coincidence. Around that time, Tim Berners-Lee left Geneva, after he had invested the Web, and moved to Cambridge to found the Web Consortium, which is the Standards Body of the Internet.

And by great luck, the Web Consortium moved to the third floor of the Computer Lab, which is also where the algorithms group at the Computer Lab is located, run by Tom Layton. And Tom and his colleague's expertise is theoretical math and the application of algorithms to theoretical distributed network problems.

And Tim and some of this colleagues suggested to Tom and some of theirs -- we like to say it was over a Cappuccino in a corner -- it's not exactly that true, but it's a great story -- that they ought to think about taking these theories and applying them tot he Internet, because the Web is a giant distributed network problem and there's nothing theoretical about it.

And so Tom's group got DARPA grants. You can think of Akamai's angel money as defense department grants.

And they began work on applying their theories to what Tim Berners-Lee calls hot spots, where you get congestion on the Internet and performance degrades and ultimately breaks down.

And what Tom's group discovered after a couple of years of work was that their theories actually solved Internet congestion problems, and they began thinking about how you take these theories and make a business out of it. How you create software and what kind of business model you could create to form a company. And they solved that problem.

And so in the Spring of '98, they entered a student competition to introduce their idea and get it judged by students and Internet experts and they didn't win. They were runners up, but they met a lot of venture capitalists at that conference. One of whom changed their business model from being a software company, to a service provider.

We in fact are a service provider. We don't sell software to people, we sell a service to e-businesses.

And the founders spent the summer and early Fall of '98 getting funded, getting incorporated and creating a business.

They got a license from MIT for the technology. We have a royalty free, in perpetuity exclusive license.

MIT has, this may be of interest to some people here, because I don't know how NU handles it. MIT has a very unusual and creative technology licensing deal. They say the faculty members who develop IP, Intellectual Property, on campus -- that they will licensee it to the faculty on this exclusive basis for a teeny piece of stock at a tiny valuation, leaving enough to go to venture capitalists or many here in the audience, I think, so I'll be very, very generous, to help add value at a fair price and still leave enough equity with the founders and the employees.

And then MIT does something -- I'll do a short plug for them, because I think it's relevant, maybe here. MIT takes their stock and then they cut it into three pieces and they put a third in the General Fund. They give a third back to the department to make them feel OK about faculty leaving to start businesses, because if this thing hits, the department gets new funds, and then they take a third of the university's stake and they give it to the students and faculty who worked on these things when they were research projects, because often those people have gone off somewhere else, so they can't become an employee and hope for an IPO pay out.

Fortunately for us, almost all of them have remained employees, but there are some who left, and they have stock in their pocket in Akamai, and it made them feel good for when they were probably paid 22 cents an hour as a researcher on something that looked like it had no potential, that they weren't cheated later.

And I think that's an amazingly open approach that MIT has adopted, and today they're shareholders, and we're opening our new headquarters. We actually moved yesterday. We will be 50 paces from the Lab for Computer Science. We will recruit by the hour for students to come code (laughter).

So, we got funded at Thanksgiving of '98 and one of the other unique things about the company is, because there were five years of work going into it, the thing worked. It wasn't the typical startup that said, "I need some to build my alpha and then my beta. It worked in the lab. The alpha was done. The beta was done, the technology worked.

So, we took the first $8 million at the end of '98 and started building a massive global network. And so in January of '99, a year ago -- we opened for business with about 12 customers who really trialed our service by putting their live front page content up on the Web.

Took them an extra month or two, but in March of '99, less than a year ago, we went live on the homepage of Yahoo, which is very usual for a startup to get that far.

And we started charging people, then in April -- less than a year ago, we had our first customer, and we were in business and I'll sort of describe what's happened in the last eight months.

Today, we are 375 employees. More than 200 customers.

Our network is very large. We've distributed more than 2,000 enterprise-class Pentium servers, primarily running Linux around the world. They sit at data centers and points of presence on over 100 distinct networks in more than 40 countries. And we probably have the largest distributed computing platform on earth.

What's the problem we solved? Well, its a simple problem. We solved the Internet. Let me describe the problem that you're all familiar with.

For every hit, the users HTTP request goes from wherever they are in the world, back to the home site, contents dragged across the Internet. For every user at the same site, it's dragged back and forth, over and over and over again.

And on a busy day or for a big event like a major sporting event or a lingerie show or some of the things we're familiar with (laughter), some of the better uses of the Internet -- sites crash.

So, first you could have congestion at any of these points. Think of these as connection points, routers, networks handing off traffic or back to the home to silo. You get either poor performance or no performance.

And in a world where more and more Web sites now outsource their hosting to companies like Frontier and Digex and Exodus, you'll often be affected by your neighbor does, because at some point you're sharing an Ethernet connection into a router, out to a pipe outside. Your network can hose the whole data center and there's not much you can do about it. And that's a pretty ugly problem.

Well, the founders of Akamai thought up what I believe is a really unique solution. They weren't going to fix the Internet, they would finesse the way content and applications are delivered to end users, and what they created was technology, that allows us to locate Akamai servers. These Pentium based machines running primarily Linux that I described. The edges of the Internet. Outside of central core, outside of where somebody's silo is, if you will, the hosted site, and what we do is we put these servers in small clusters, pretty much anywhere where we can get Internet connectivity where there are end-users connected or we can get to them.

So, for example, they're in every Exodus data center. All of the Digex data centers. They're at AT&T data centers. They're at UUNet POPs. They're inside smaller ISP's all around the world. In some cases, we buy co-location and in some cases we get free co-location and bandwidth.

And so, that's what we've done physically with machines. And now let me walk you through all the things that we do that allow us to create a service that solves this problem of delivery.

And I think what you'll see is that each one of these categories -- I think -- you could found a business on. I think you can found a once in a lifetime kind of company like Akamai, where you can link them all together.

And what we do is, the home site still understands where the traffic is going. We don't take control away from our customers. From Yahoo, for example, they can still place cookies and put the right ad in, but the content is delivered at the edges. So, the first thing we have to do is distributed server deployment. We have to put thousands of machines, ultimately, tens of thousands of servers, at the edges of the Internet.

That's actually not that easy. I guess if you have a really, really, large credit card limit, you can call up and you can order servers and have them shipped. But actually getting them contented and being able to talk to them and manage them remotely isn't that easy.

We manage the machines from Cambridge. We never go and visit them. We don't see them and water them, and they're all over the world.

We change software remotely and we do it without ever taking our network down. We don't have down time for maintenance and we do that all remotely. But it's the easiest thing we do.

The next is connectivity, being able to track how they're connected, whether they're connected. The quality of the connection. The amount of connectivity they have in real time. The next issue is, you have all these machines that you know are there and that you can talk to. You can control them, what do you put on them?

We don't mirror Web sites. It would be impractical. Instead of 2,000 servers today, we would have to build 2,000 data centers, if we took all of our customers and copied their content.

So, we have unique algorithms that allow us in real time to know what which of our customer's content needs to be located on which specific machines in quantities to sufficiently handle demand in real time. That's a pull model it's not a push model. There's nothing predictive about it. We don't have to figure out what content will be popular next.

Our customers don't even have to know. They can Akamaize all of their content, and wherever the demand comes we just handle it. That's a very complicated mathematical problem that our system handles.

The next is mapping end users everywhere around the world. If you're going to locate servers at the edge and then serve people locally, meaning on a machine that's in a network sense topologically close to them, you have to know where every end user connection is in the world. And we do.

Now that doesn't mean I know if you're surfing now in your pink slippers. I don't know who you are individually, but I know where you drop on and off the Internet.

And we update that in real time so that we believe we have the most accurate global map of end user nodes on the Internet, and that's through a proprietary math and mapping algorithms.

The next is never want to send you content on a down link or a dead link or a -- or a network connection that's congested. And the reality is that network performance varies in real time all the time everywhere around the world. A link's that great right now could be horrible by the time I'm done speaking to you today, and I promise that won't be too long a period.

So we have unique technology that allows us to monitor the state of the Internet in real time, and we do it because we're sitting on top of 100 networks, essentially triangulating them all and have proprietary technology that allows us to understand essentially the capacity of links. Not whether they're up or down but their capacity.

And that's not easy to do. The simplest way to sure if you think about it what's the capacity of a glass, well I'll fill it up till it overflows, and then I know it's full. Well if we did that to all links on the Internet we would just crash the Internet. So it's not an effective way of doing it, so we have an efficient way of knowing in real time network status.

And I think each of these things is pretty cool. That's really not the magic of -- of what we do.

What we really do is what we call content and application routing. It's the ability in real time to take a user request and to say this content on this machine on this link to this user.

We now do that 25,000 times a second for our customers. We handle a load today that exceeds one billion hit requests a day on our network, and we do it accurately every time, faster, more efficiently than our customers can do it using any other means on their own. And we guarantee that we will do this or they don't pay us. That's not easy to do, and that's what our company is really about. So it allows us to have this business model that I talked about.

So who are our customers? Most of the search sites on the Internet. Many of the leading e-commerce sites on the Internet. Top financial sites.

We stream, not just static content. We've done several million streams already for people like Apple and Microsoft.

Nike's new introduction of a show has changed their whole business model. They started running commercials last week. They pushed people to the Internet to pick the ending of the commercial, and at the ending you can buy a shoe on the Internet, and they use our network to guarantee that they can take a TV audience, drive it to their Web and get high performance delivers so that they can ultimately sell shoes. It's a completely different way of thinking about selling a consumer product, marrying broadcast television Internet with a unique delivery mechanism that now insures it doesn't break down in the last piece.

And finally major media companies, all of the large networks, CBS, NBC and ABC are our customers. Major publications -- I shouldn't in this audience -- I should have put the Chicago Sun Times, the New York Times, one of our customers.

And people are using us for applications, think of software downloads. Think today how you update software on your site. Often you go and you go to Web site, and you click through all the pages to find the plug-in or the patch you need, then you get to the page that says where do you want it from? I always pick the site that sounds like a great vacation that I can't take right now. So I try to download from New Zealand and then server's busy, you get poor performance.

Contrast that with the QuickTime player today. If you get the message that says there's a new version do you want it. and you say upgrade, that's it. Our network detects where you are, what server has the software on a network that can deliver it to you and it's downloaded to your system seamlessly. It's a way of moving an application from a central site under a globally distributed platform, and improving performance and satisfaction for you the end user and the customer who's paying us to do it.

Using kind of the marketing vernacular, we're in the tornado on our first service offering which we call Free Flow, the delivery of Web content.

When we went on the road for our IPO this summer we talked with 49 customer, November we talked about 100, at the end of December we had doubled that in six weeks to over 200, and we continue to grow. So in a traditional marketing sense we're just in a tornado on this. It is just flowing at us, we are scaling the company and dealing with all the problems that you study here of how do you tell people about what you do, and then handle them when suddenly they believe your message.

Why do they use what we do? What's the pay-off? Why do they respond to our message?

Know that people believe on the Internet there's a lot of hype which is sort of get people to do almost anything for awhile. You have to deliver something that's of value. They are spending real money and building real businesses, it's not just for fun.

In our offering, we have a very clear message, and I'll tell you how we pay off on each of these things.

We demonstrate using third parties that if you use Akamai, there's a difference that at minimum makes your site twice as fast, often 10 times -- actually with e-commerce sites usually faster. And they can see it and we can deliver third party measurement to prove that that's right.

That's important but the game is really about making sites more compelling. We enable them to deliver richer content.

Think about the Apple homepage, not your typically fast performing site, but it is almost in the Keynote Top 40 the second fastest site on the Internet. Why? Because they can take huge rich graphics and deliver them more rapidly on our network and they can do other things.

I talked to you about the applications of software downloads or streaming media. All through an integrated solution that's easy for them to implement.

We give them capacity to make peak demand. You'll remember the Blair Witch phenomenon. This is a -- Keynote is a third party measurement group and this is the kind of chart you get. It's abstracted to simplify it. But this is tracking an image on the homepage in real time on our network and off. And that nice pretty flat green line on the bottom there is that image being delivered in an average of under half a second on our free flow service.

The line above it is the same image at exactly the same time running off of the site's central location, not Akamaized as we say. And what you see is a huge performance difference. And before they Akamaized all those spikes were crashes they simply couldn't handle the load.

So we walk in with a third party validation, very typical kind of sale. Somebody says prove it to me, we actually prove it. We walk in with proof of how their site will perform before they become a customer. It's a very, very compelling marketing pitch.

We give them a real time look into what their content is doing. We track where we deliver content by state and by country for example, and they can look at real time, and for the first time know precisely where their users are.

We give them a view into the Internet because of our technology they can't get on their own. And none of those other slides mean very much to them and you won't get business unless you leave the Web business and control the customer relationship.

So they have to be able to see every hit, they have to control ad insertion, they have to be able to do secure and authenticated content, or it's a very, very short term kind of solution. They have to be able to personalize content, they have to complete a transaction, complete it securely. They don't want me messing with their database. That's their proprietary business. And they always have to serve fresh and dynamic content. Our service allows that and that's crucial in the way we sell what we do.

And then also very important we are an outsource solution, and you're going to see more and more Internet provider solutions are outsource solutions. In the early days we built everything ourselves. From ad server, to acquiring bandwidth, to hosting, to even figuring out what our back-up power was and how much UPS we were going to have in our own private data center Web sites I built.

More and more people are moving towards outsource solutions, to publishing systems, to acquisition of bandwidth, to hosting and in our case it's content delivery. You got a very efficient model when it's pay per use, when it's scalable, when it's a shared resource, rather than a central approach where you have to buy your own bandwidth, you have to buy your own servers, and you have to keep adding people to manage more and more demand. So we come to them with a very compelling cost proposition.

And then there's the question of marketing pay-off. Can you go back and ask them why can they give you a positive answer, and in our case we get very positive answers, and from an on-going sales position that's crucial.

So CNN enhances their capacity simply by becoming a customer without spending any more money, buying more servers. Motley Fool gets better customer experience for their Website at a lower cost. Britannica, and I'm sure being a Chicago company, you're aware the problems that they had in terms of performance and reliability, we actually them the ability now to serve the traffic they've generated through their own marketing.

And then I that this conference is really about e-commerce and e-business. Well one of the things that we found -- it was our original pitch and we hoped we could pay-off on it -- is we said if your site performs better and you're an e-commerce provider, your business will go up. You get people to have a richer, more reliable experience they're not going to go to your competitor, and they will buy more. And we are now seeing that. Which is amazing validation, and I think long term for an infrastructure provider or anyone selling e-commerce solution, if you cannot demonstrate to make more money using a solution you're not going to sell it. And what we've seen is that we help people make money by outsourcing problems to us.

So KB Kids is seeing increase conversion rates in sales that they attribute directly to using our solution. EBags has an increased conversion from visit to sale of a quarter of a percent, which actually a huge number. That is money then can measure going into their pockets. Discovery isn't selling things, but they're selling ads. They have doubled their ad target based on using our technology. Which changes their whole view about the potential of their business.

Since we're here at a marketing conference I thought I'd show you our advertising, because this is crucial to building an Internet business because you have to build a brand. You have to build a customer base, and you have to talk to current or potential investors, and we as a company have marketed incredibly aggressively.

We have talked an advertising budget for Year 2000 that will exceed $20 million. It's a lot for a very small company. We advertise both in the daily press, and we've done a branding campaign nationally. This was actually a Q4 branding campaign, and I put it up here to sort of challenge your normal thoughts about where you might market to. We actually ran these ads in Sports Illustrated, Time Magazine and the New Yorker.

Now why did we do that? Well we need to get out of the clutter of all of the dot com ads that run in all of the dotcom magazines, all of the standard trade rags.

Also we have several audiences we need to reach. We need to reach the head of Web Ops who reads the trade publications, but we also identified that we needed to talk to what we call the CXO audience, that's the CIO, the CTO, the CEO and on and on, of companies who are in the Internet business, because they are fundamentally struggling with how do they make their Web business successful, whether they're a traditional company migrating to the Web or someone who's Web only. For a solution that is as sort of dramatic and scary in some ways, you have to talk to that audience who actually makes the decision, or if things crash is yelling at the people down the line. So that's why we've done something that I think is unusual in the Internet business, and you may have seen our advertising showing up in places like the back cover of Sports Illustrated and saying why is this here?

Well, by looking at where the CXO audience also puts time into media we found some unusual spots where we could talk to them and there are no other dot com ads around. None of our competitors or people on the Internet are advertising there. We found that very effective.

Other ways we add customers and build our business, the Internet is about partnerships.

Forester now is talking about hyper-partnering and measuring return on connection. What they mean is to identify a niche, partner with everybody in the space, competitor, friend, foe, not sure, sign them all up and then measure who returns on that connection to you. Where do you drive business, and it may sound cruel, you ignore everybody else and you move on.

But one of the key strategies is to find a channel kind of relationship or partnership and sign them all up, don't spend a lot of time early on trying to figure out who's the best one, because nobody knows. They don't know either whether you're a good partner. Sign them all up and natural relationships evolve, measure the return and go where there's return.

We've signed up many, many channel partnerships, and we've been able to measure effective return. And one of the things we've found we didn't think we'd do reselling, but we found great return, so with the blue box on the top, Digex Frontier Global Center and Navisite are reselling our service to their customers, which is finding us business much more quickly than we could find it.

Below that blue box are people where we have a no money changes hands relationship. They're people who advise folks on building Web sites or provide functionality applications. Their customers want Akamaize.

Vignette needs to be able to Akamaize efficiently, that's good for them because their customers say I'm going to find a publishing solution that can Akamaize content because I'm going to use Akamai, or they can also go and say have you thought about Akamaizing, that's going to help your site perform better and we can make it easy for you to do it.

The flip side is a customer says I love you Akamai I want to Akamaize, I really don't know how to do it. We can give them a list of people who can make it happen quickly.

We've expanded our company internationally, which is another phenomenon you see on the Internet. I think there is a conception that's pretty accurate about this being get big very fast and it's not just a US phenomenon.

We have expanded and announced just in the last couple of weeks the opening of offices already in Germany, France and England and expect to grow our staff there very rapid

And the next question is how do we get networks? We have a staff that does nothing but work at growing out network reach, which has gone from nonexistent 13 months ago to about 1,500 servers in the Fall, now as I said 2,000 on more than 100 networks on a truly global basis.

Because for Yahoo their users are literally everywhere. And when they think about an Akamai solution it is unacceptable for me to say, "Well I'm in five countries or 10. I have to be able to deliver their content quickly everywhere in the world, and that goes for most of our customers. So it's critical that for this kind of solution you have global deployment.

And the kinds of places we deploy tier one networks, ISPs, telcos and CLECs, I won't talk too much about that unless you want to get into in the Q & A, but you can ask, "Well why do they want to partner with Akamai?" Well, some of them want to sell me co-locations for my servers and bandwidth.

We actually have a much more interesting pitch which is why on the Internet you really have to think about how you partner, and we have a philosophy which I try to impart to anyone who says can I have an Internet business, which is you have no vendors. You have customers and you have partners. Part of our business exists because I don't dig trenches to lay fiber, and I don't build data centers. I have to partner with people who do, and I bring them benefit.

By bring an Akamai free flow server inside their network I solve two issues that they try to solve on their own. One is how do they have great performance to their end users, people who pay them for Internet connectivity. Well, some of them all go to Yahoo for example. I bring them greater performance by bringing Yahoo content inside their network.

And then all networks have a variable cost, how do they get content off the Internet inside their network? Well, instead of having to go across the network every time to get Yahoo, it resides in there because Akamai brought it in, and in fact, I put the servers in these networks for free and I mange them. So we can actually partner with people who normally might just be a supplier, which creates the ability to grow a symbiotic relationship much more quickly than in a traditional business approach.

And then the last piece of our business is technology leadership. This Rogue's Gallery represents some of the founders and scientists of the company. I -- I think it's a little unusual, we really refer to our engineering department as Akamai Labs. We have 40 Ph.D.s in the company, and they think about solving complex theoretical math problems that all apply to the Internet.

And then we have a staff of really product pioneers, who understand how to build better solutions for delivery of content, applications, streaming media. They've already built based on customer feedback products three and four, services three and four, which are now in beta test. They're unannounced, we'll talk about them later in the quarter.

Unannounced actually one of them is fully commercialized with a first customer, and that was the result of going back to our customers and asking them, "Gee, what else would you like? What else could you do on a platform like ours in building new services," which means you need to create in an Internet company very rapid feedback loop with customers. If you just sell them and move on, they're going to go somewhere else. The next time you come back you've lost them. Guaranteed. And we split research and development, 2/3 is in Boston and 1/3 is in Silicon Valley.

One of the other things that we did as a company was try to create a world-class place we could get advice, because you can't solve all of the problems yourself, and in our case just to feature the first two, when you can turn the guy who invented the Web and the guy who built one of the all time great Websites for advice, you're in great company. And I would advise anybody starting an Internet company to try to look to have a Board of Advisors as opposed to a Board of Directors who can help you smartly figure out how to build your company.

I'm going to you a little bit of a different take on what we do and where we're going next, and then we can talk about anything you'd like to in the Internet.

The first problem we solved is taking our customers in all these categories, we broadly call them content providers, and marrying them to Internet, networks, all the ones on the right hand side.

And today they don't talk to each other. They don't effectively communicate. An Internet company isn't going to go to 100 networks and say talk to me about getting my content delivered. And any one network that wants to go talk to content providers can't provide a global solution. The largest, most dominant players of the UUNets of the world carry small percentage of all of the traffic. There are thousands of ISPs out there. So there's no way until Akamai to bring these two together in a solution set where we solve the network problem and we get the customers and link them in the middle. That was we thought the first great opportunity in the first business we launched.

Now we think we've identified another interesting problem. It's not content getting to end users across networks, it's e-businesses trying to figure out what applications they can run to actually make their sites better. How do I do content transformation based on bandwidth of end users? How do I do better search on my site? How do I do customized ad insertion?

The reality for all of this functionality on the right to a Web site is that almost everything today is a one-off installation, and it's very slow and cumbersome. It makes growth of and penetration of better functionality into e-businesses slow and cumbersome and very much a handwork kind of problem, because there's no way to have a easy to integrate solution.

We think there's a way to do that we call it Edge Advantage. By pushing thousands of servers that run in a completely distributed fashion, in a completely fault tolerant way because of our math, we think what we've created is essentially a processing brain at the edges of the Internet on which people could fit applications, and then somebody who has a content transformation tool or an anonymous user profiling service. If they can fit it into our platform we can go to all of our customers and essentially say this application is Akamaized all of you who want it can do it by simply checking a box. We've done the integration work for you.

And we think that this is a huge opportunity for us as a business, and a great opportunity for Web sites to have better functionality and people who build applications to drive their business more quickly. And I'll give it to you kind of in a different visual take.

Think of our system as this platform and the first two services, Free Flow, which is how people deliver things like ad banners, or Free Flow Streaming how they deliver streaming media on our network is the first two applications, and we're going to write some ourselves.

I talked about two new services that we have in the field today, and then we believe that if you can create an open standard it can be taken to the IETF for example, you can let third parties who build applications, and you can think up anything from ad serving to search, could hook them into our network through either standards or open APIs so that content and applications can begin to talk to devices on the Internet. Our network is a series of devices that all talk to each other. You can actually change the way Internet content and application is conceived of and the way sites perform and we think that that is a transformative kind of idea, that it changes the Internet and actually accelerates the growth of the kinds of businesses that everybody's trying to do today.

As an investment we think that Akamai was a new opportunity that we identified. Nobody had talked about content delivery before. Everybody thought about Websites, bandwidth and end-user problems.

We came with a world class technology that defined a category which I think is important for Internet winners. We got some of the top sites as our initial customers. We are working to create new standards through partnership, which we think philosophically is the only way you can do it.

We have an attractive business model. What that means is we know how to make money. We make money on all of our sales. We have a profitable margin. We don't have a profitable company today, you wouldn't want that on the Internet I think, at least not in the short term. That's a joke for the tape. I don't anyone to be mislead by that later. But we actually know how to make money and do it.

We've built a large company, put a lot into R & D, we have to build out under the employee base, but fundamentally we don't, as Bill Girly (sp?) said sell dollar bills for 85 cents and make it up on volume. That's not our business model.

We have enormous barriers to entry. If I haven't convinced you after walking through all of those boxes of the things we do, that this is hard to duplicate, I need to get the mathematicians out from Cambridge to walk you through some of the algorithms. And that's a huge barrier to entry, and then we have a very diverse management team.

And so let me leave you and then take questions on one of our favorite quotes about the company from Wired Magazine. Thank you. (applause) In the back.

AUDIENCE: Have you thought about going into the content industry. It sounds a lot like a television station actually.

SAGAN: Right. No we will never be in the content business. We don't intend to ever compete with our customers, and while I have a media background, no one at Akamai knows anything about building content. We know about delivering it but we are not out to try to aggregate an audience for ourselves. We want to help our customers build bigger, more successful audiences.

And in some ways we're more like -- we're a distribution network. Think of it. The analogs I would point out would be in the old media world, Time Magazine for example the first copy is produced at Rockefeller Center, and that's not something that's likely to ever change. But Time doesn't duplicate, they turn to somebody like Donnelly, and they say on Saturday at midnight, here's one copy, you have to get seven million out there by Monday morning. We're in the position of a Donnelly or a FedEx that actually does the delivery, but we don't care what those bits are. We do those billion hits a day, we don't care of know what nay of them are. And we have no expertise in knowing what they should be.

(Question from the audience)

SAGAN: We think we complement them. Some of their competitors are our customers. Ad Force, for example, who's a direct competitor, is a giant customer, and they feel that what we do is add value, because they're really in the business of getting higher CPMs for more ads, and one of the ways you do that is higher performance and click-through and we help them deliver more accurately. So someone like a Double Click clearly could run Dart as a service which is -- they're in the service business and get enhanced performance by using our technology underneath it to make sure that the ads are delivered faster and in a more targeted way.

(Question from audience)

SAGAN: I didn't get to -- because it's kind of an old slide the who are your competitors. But Accelera is a caching company. Let me answer that in a broad way and cover several companies, because I'm here saying Akamai makes Internet perform better. Well there are a number of other people who could come up here who would say the same thing. And in fact, they do and that's a good thing, and I'll talk about each one of them.

First bandwidth, right? You could have someone who's kind of a wholesaler, a Level 3 or a Qwest, or somebody or somebody who has more intelligent bandwidth like AboveNet or InterNAP who can sell you kind of a smarter, connectivity to multiple networks, and in fact I'm a customer of all of them. I buy as much co-location and bandwidth as I can, and I'm if not the largest, I'm one of the largest customers of almost every large bandwidth provider on earth, because all of my customers need Internet connectivity, and they need the best they can have. That doesn't solve content delivery at all. That just gets it out better, and then it's in that mess in the middle of going through -- from coast to coast to get bits you, probably on the Web you probably go through 24 router hops to get an average object on a page. And none of those networks or solutions can solve it, so we need better connectivity for sites to grow, it's not a piece of it.

Caching or mirroring is a technology that allows individual networks to have better performance. We are caching agnostic. We in fact at some level cache. Cache is something that gets content, serves it, and can reserve it if it gets an identical request. Your PC has a cache for smarter delivery to your screen of stuff off the hard drive for example. Our machines are caches. We have a partnership program with most of the major caching vendors where we're trying to work an API that would allow their boxes to be nodes on our network, for example, because we're actually agnostic off of what box we serve, our cache software, as long as we can control that and report back to our customers.

And then hosting I touched on but then hosting is also seen often as competitive, but we don't host. We don't do source data management. If you're an eBay or an Amazon you got to have a great place for your Web site that's hardened, that has reliable power and air-conditioning and redundant access to the Internet, and people who can swap machines. We don't do any of that. What we do is get it from their end users, and hosting companies can't offer that.

So all of those verticals are people who do make the Internet better, and without them also progressing there's a cap to how large we can go if we don't all make the Internet work better. And then Accelera specifically is a company with a mirroring technology. Most of our big customer mirror their sites.

Yahoo for example is duplicated on both coasts and they better be, because no matter how good a -- either your own hosting center or an outsource is, they go down. Guys with back hoes do really bad things in the parking lot to your fiber sometimes, and you simply better not be, if you're mission critical, in one location.

____: According to you (inaudible) by customer, what am I giving up. Some sort of functionality (inaudible)?

SAGAN: No you're giving up higher costs, headaches, core performance, bad reliability. (applause)

____: No downside?

SAGAN: No, no downside at all. Nope. (laughter) No and in fact that's crucial in those slides of the pay-off. If we -- and we had some competitors. There were people who actually had sort of a matching pitch, who say we do content delivery. Well sometimes if what you say is you have to mirror the site, you lose the view of the customer. That's a bad thing. If you mirror, you've duplicated and taken away. If you put a box in front of the Web site and say that's how we'll do it you lose sight of the customer. If you say move the content off and you can no longer do customized ad insertion or match cookies, then you lose.

In our case we allow them total control of their business and we accelerate the performance of the delivery. It's the right question, and I think the reason we are getting customers so fast is they don't find a downside. They save money, they don't lose control, they get tools that give them better information actually than they've ever had before, and it's reliable and it's guaranteed. So it's a good question, but the answer honestly is that they don't lost anything.

AUDIENCE: Let's take a customer who has a data center at Exodus. What is the process that customers go through to Akamaize their firm?

SAGAN: OK. These are great. These are all set-up questions.

MODERATOR: Could you repeat the question?

SAGAN: How does a customer who's hosted somewhere, your case was Exodus, Akamaize their site. Well, the nice thing without going too deep into technology is there's no capital cost, there's not projecting how much equipment to buy for example.

To Akamaize content you sign a contract with me and you change the URLs of the embedded objects to what we call ARLs, Akamai Resource Locators. Because in a Webpage, (laughter), good marketing, but on a Webpage every embedded object, if you look in the HTML, everything on the page has a unique name. It has a URL. In our case, all you have to do it rename objects, or in fact, you don't rename. Most people are always publishing new content. You simply name them correctly as ARLs, and our network identifies and serves them.

So there's no cost, you don't have to predict which stuff, just name everything and whatever's in demand we serve it. It still can be hosted at Exodus. You'll get great performance there. There's a fabulous company. You get better end user performance, and you didn't have to buy nay more machines or try to project, simply outsource to us the delivery.

We have a very small -- it's a couple hundred lines of code something we call Launcher, we've authored it in several dozen versions, there's a TCL version for Vignette and there's a C++ version, so that you can plug it into your publishing system, whether it's an off the shelf like Open Market's or homegrown. We have some technical consultants that will help you do it, and then automatically every time content is pushed out, republished by your Web group, it gets named properly. Because everybody has a naming convention. We help you change the naming convention, that's it. And you're Akamaized.

So we had -- we've had some customers talking about, "Well, we might use you. Oh my God, we're down. Something happened. On the phone we took the largest sports site in the world and Akamaized the most crucial content in 15 minutes, because we manually told them what to do. And they were up and they never went down again, so it's really simple. Oh, here's your hand before and I kept forgetting. OK.

AUDIENCE: What would render Akamai obsolete?

SAGAN: Horrible customer service would render Akamai obsolete. We are building a service company. We're doing it with unique proprietary technology, but fundamentally we're providing and outsource service.

The greatest risk I think to us is failing to understand that and scale our customer support. To bring on 200 customers, most of them in the last three or four months, is a massive problem. It's really easy to lose their contracts, not answer their phone calls, because even if the network works, they have questions. "I see something. Is that you? What is it?" We fail to respond once, they are gone. I don't care if I have a no cut contract. They're send me a check and tell me to go away if they think I'm hurting their Web site. So I think the greatest challenge to growing a large service, am outsource service company is customer, the way you handle customers. Do you find them? Do you tell them the truth? Do you deliver exactly what you said when? Do you love them to death, and when you know what hits the fan, for whatever reason, do you solve the problem? And we have invested millions in our IT infrastructure.

So we're not doing stuff by hand. You could do the first 12 customers by hand, you can't do 200 bills of -- essentially, we're sharing a global resource. We are charging for bits delivered for each customer on every server. We have to aggregate by object name every customer's content across all those machines and give them an accurate bill of their utilization. It has to all be automated and massive.

I've been doing customer visits. I go in with a printout of every time they called our 24-hour number, what the issue was, who handled it and how quickly that ticket was closed.

We have an engineering group that includes some Ph.D.s, we call the White Hat Group in R & D for serious technical service issues that can't get handled by customer care, it goes right to Ph.D. level people.

We just solved a massive router problem for one of the largest search and portal sites on the Internet. They were just having a massive problem. They threw their hands up. They called their networking people, their outsourced hosting people and us and started just -- they were beside themselves. And we were sure it wasn't us. Our group within hours, overnight on a Saturday found the specific router in a tier one network and said it's this machine, get it fixed. And we solved their problem.

They will never leave us. They won't. (laughter) They won't. I mean they were down and we solved the problem. It had nothing to do with us. We didn't say we confirmed it's not us, and there's no charge, and it's because we have a customer focused enterprise that goes all the way through the R & D organization. I think that's how you solve that. Yeah.

AUDIENCE: Describe your pricing model and (inaudible) model (inaudible) I can show you what you did and what other options that you might (inaudible)?

SAGAN: Sure. Describe our pricing model and how we choose what we did. What we do today is we charge upper network utilization.

We bill in what looks like a bandwidth bill. We charge people for megabits per second of data delivered on the course of a month.

And we do it on what's called a 95th percentile which is the way people buy bandwidth. Networks measure how much traffic you're delivering at five minute intervals, you then sort them by highest to lowest at the end of the month. You slice off the top five that's free, and you pay me the 95th number. So you may -- our rate card if $1,995 a month per megabit per second, so if you contract for five megabits per second, it's essentially five times 2,000 you pay me $10,000 a month. We measure your traffic and if you go over it we charge you 25% bursting premium and that's how we charge.

And the massive infrastructure issue is how do you keep track of all of these customer's content flowing through all of these machines and aggregate it back. It's a huge data "munging" problem that our guys have solved. And you need to do it, where sometimes machines are down, so you have to make sure you go back and aggregate it accurately.

And it's a shared resource. So in every box Yahoo is sitting next to Lycos, is sitting next to CNN, is sitting next to Martha Stewart so we can't just say that's a Yahoo machine what went through it. We actually have to trace every bit through every machine.

So that's how we bill today for our basic services of object content and streaming. Some of the new services that are coming are based on monthly recurring revenue. Because we really like that model, you sign them up, and keep getting it, but they're not on bits delivered because it's a different kind of intelligence, so I think the margins there could be even more attractive. What was the fundamental though decision it wasn't this sort of -- was it 95th or 90th percentile like bandwidth. When the founders went to the 50K competition they thought that they were in the software business, right? They wrote code, it was really cool, it made networks better, so we'll sell software to networks.

One of the things they learned and this was very helpful from some of the venture capitalists, actually the one who wound up leading the first round, a good lesson to venture capitalists about getting in early and actually giving good advice and following it up, Todd Davis at Battery did. He said, "Software's an OK business, but service is a much better model and think about the law of network math. One massive network with more nodes is more powerful than any number of small networks with fewer nodes that actually might be an aggregate the same size." So the founders changed the business model from licensing software to creating a service and selling it to content providers.

(Question from audience)

SAGAN: No. No. No, actually, well, I disagree. We actually like cable modems. The question, if you can't hear the back is, to paraphrase it, and put it more crassly, is will broadband cable modem obviate the need for Akamai and in fact no. In fact we see huge growth potential, because we charge for utilization of our network. The more bits people want, the more utilization.

And in fact broadband creates a bigger problem in the middle of the Internet. The more people want faster, the more congestion there is in the middle. Remember, we're addressing the middle problem.

What we see is that in cable modem homes you get the greatest benefit from an Akamai site because we're delivering close to you, and the only way -- it's not a question of that the eye can't comprehend it, we're now moving -- we're not talking about static pages, right? If I make static pages fast enough over narrow band, why would you need Akamai over broadband? It's not about static pages, it's about streaming media, it's about applications and stuff that needs to look more and more like television on your computer screen, and the only way to deliver that many bits is at the other end of your modem connection if you will, where line speed becomes the gating issue.

There's not enough bandwidth in the center of the Internet to do streaming media for many users. If you took just a couple hundred thousand simultaneous users at a broadband connection speed of 100 or 200 kilobits per second, which actually isn't a very high speed, you'd basically melt down the Internet. You're probably match the total capacity at the center of the Internet today. That's pretty scary. So the only way for the Internet to become this truly rich, real time medium, you have to move content out of the middle and not traverse the middle.

And that's actually why we think the greatest growth potential for our kind of solution is when more people have broadband, not a problem. Way in the back on the wall there.

AUDIENCE: (inaudible) where you're serving up more functionality and applications possibly it seems like you guys are more towards a hosting company, and if so does that -- does your current relationships with let's say Digex or Exodus, is that a temporary marketing channel at that point?

SAGAN: No, I don't think so, because take content transformation, how do I serve a broadband user versus a narrow band user in real time. Hosting companies can't solve that problem at all. How can I access real time customized ad insertion off of a third party provider?

Well, today the third party is not doing it at your data center of the hosting company anyway. If we can make that work better, everybody's faster. So I don't think that that in any way challenges hosting companies or people who build people's Web sites or serve them at all. I think the partnerships will keep growing. Yup.

AUDIENCE: Can you go over what value people with MBAs add at organizations like yours? (laughter)

SAGAN: That's -- that's too easy. I spoke at -- you guys have heard of Harvard right? That's another business school. I spoke, I gave the Akamai story there to their tech student group, and the hero of this year's class is a guy named Chris Turner. Chris dropped out and is head of channel sales at Akamai. Finished his first year and I don't think he's going back, that's to depress you. Sorry.

No, there are a lot of MBAs and there are a lot of JDs at Akamai, and it is great experience. You could also say what would a journalism degree do for you? We need a wide variety of skills.

We are doing all sorts of business development. We announced our first acquisition this week of a company in the Valley, for $200 million.

We need MBAs and Internet companies need them, but I guess what I would say is what Internet companies need is people who think outside the box. So whatever training you have, if you come with too narrow of a view, you're going to be road kill. You reinvent things everyday, and the key is really bright people with diverse experience who work collaboratively and are willing to reinvent your model every 24 hours and reinvent the company. So we clearly hire MBAs and JDs as well as Ph.D.s.

Our company has a particular bias towards Ph.D.s in computer science and theoretical math. There aren't a lot of Internet start-ups that say they will hire an infinite number of theoretical mathematicians. If there are any Ph.D.s here in theoretical math you're hired. That's all I need to know. If you're an MBA, you have to have an interview. (laughter) Sorry. Yes.

AUDIENCE: (inaudible) About two weeks ago, I came across this article in the New York Times and I saw your name there and saw that you went to Madille (sp?) and --

SAGAN: It's true.

AUDIENCE: -- I noticed today that one of your clients happened to be in the New York Times. My question very shortly is -- is Wired Magazine too.

SAGAN: No Wired Magazine -- well Wired doesn't actually have a site anymore, right? Wired sold to -- no, I guess I should take that back because they sold to Lycos? Altavista? Lycos. And Lycos is our customer, so my guess is Wired.com is now Akamaized, but the magazine doesn't actually have a Web site anymore. So there's -- that's my truthful answer. Yes.

AUDIENCE: In terms of competitive advantage. You said before that you think there are barriers to entry, what -- I mean, what about the companies that puts 10,000 servers out there with like $200 million of investments, brings 200 Ph.D.s and pays them for a year to do the algorithm that you do, maybe even better.

SAGAN: Well we're paranoid as hell about competition. I worry about it 7 by 24, I don't -- I don't miss a day where I'm not worried about competitors I know about. I worry most about the ones I don't know about, because what if somebody's come up with a better idea?

And there's no lock on brains. We certainly tried to lock up as many in our space, and Tom Lane I joke with him is the smartest living mathematician on earth, and he hates it when I say that, and he may not be, but he's pretty darn close, and we've hired his best students who all went off to be tenured faculty in places like CMU, and so we have some of the smartest people thinking about it.

Now it doesn't mean that in a garage somewhere or a lab there isn't a better solution which is why we view ourselves as a service company first. Go out -- just lead the market and just put your foot down on the whole space and take the whole hill. You get all the customers. You get 200 that fast and you try to keep doubling it and doubling it and doubling it. And what you realize is, to answer this question again over here, is you are a customer service enterprise. The model has to be like IBM or Cisco.

The reality is you could probably point to any product in companies like the line and say somebody has as better one of those, somebody has a better one of those, but they dominate their spaces because their customers won't leave them. They love their customers, they hear what their customers need, they get feedback into R & D and bring a solution, nine times out of 10 faster than anybody ever does, so they get bigger and bigger and that's our goal.

That's why our next service was developed by one of our customers. We listened to them, we brought them a solution much greater lock in. Likelihood that they would ever switch just dropped again because of this new product, because of how valuable they found it, and we think others will as well.

So I worry about competition. Our goal is 100% market share. We never want to lose a sale and so far we're pretty close to never having lost a sale -- to anybody. Over time we can't stay there forever as the space gets as big as we think it is, and other people have some pretty cool solutions, but if we do the best job at customer service we win and so that's how we focus on staying ahead technologically, but actually winning even if somebody else comes up with some cool -- cool ideas.

AUDIENCE: In a world where the --

SAGAN: Are you a VC? You have a tie on.

AUDIENCE: Oh. I hope to be but I'm not right now. No just kidding, just kidding. In a world where the first mile, all of the miles in the middle and the last mile are all optical, and are being switched to optical switches, where is Akamai going to be?

SAGAN: I think we still have a great role in a world where everything is optical. I think you still have the same problem of, no matter how big we build our freeways, they keep getting filled up, and think of it this way, you can't double the speed between here and O'Hare by doubling the speed limit. Right? If it's made 110, it's still going to take an hour to get there because the roads keep getting filled up.

The more fundamental architectural reason why I think the need for Akamai doesn't go away is because we're moving to a packet switched universe and leaving a circuit switched universe. We're not going towards the day where my connection at home is a piece of glass that I can point from site to site.

When I make a phone call I get a dedicated circuit to my mother, to the store, to the airline, and it's mine and I can move it. But I get a dedicated circuit switch. That's not the way the Internet works, and in fact voice is moving to the way the Internet works, the packet switching, and when you get that jumble of everything hitting each other, you need a way that sorts it out, and we believe that fundamentally that's what we do. We manage the movement of the data to the edges, and then the secret sauce is in managing the data at the edge, and more glass in the middle is going to be more stuff swimming around. It still has to get to the edge, and it has to be managed, and we think fundamentally that's where our value, that's where the algorithms really work, is managing data at the end. And more fiber is going to get more bits that need to be managed at the end, so we actually welcomes things like Sycamore and stuff like that, accelerating the adoption of optical stuff in the network. Yes.

(Question from the audience)

SAGAN: We don't do what they do. The honest answer is they don't really like us very much. I was asked to handle a lot of questions, and I understand that they backed out of coming here, so you'll have to get them next year I guess. They don't like us because they're positioning, and some of their perceived value, I think, is they're the cool software company on the Internet with lots of Ph.D.s, and now frankly there's another one. So that's probably not -- they don't like that.

But the truth is, we don't do what they do at all. We don't ever see them in sales, and think of it this way, this is a triangle, the top of the triangle, top of the pyramid is content, that's what matters. If there's no content, there's no Internet. We're at the top of the triangle, and we go to content providers, and we say we will get through all the applications and all the networks to the end users everywhere. And we will solve this problem for you and you're our customer, and they find that very valuable. We can sell there.

Caching companies, to put this more generically, sell down here to individual networks. They sell a hardware or software solution to individual networks to improve performance on a network by network basis. That's a really good business, and that's one company that's doing really well there, NetApp is another one and there are others. They do really well and I think they're going to make a lot of money. How much? I don't know I don't study that space that much.

They don't sell to content providers. They don't provide a global answer for the delivery of content and content applications everywhere, and so fundamentally we offer a value and a different part of the chain, and we never run into them. So in some ways, positioning meaning we want to be seen as leadership in designing the way the Internet operates is clearly kind of competitive messaging, but in terms of businesses and who we sell to, we never see them and I don't think ever will. But there clearly is a mind share competition and we want a big piece of that mind share.

AUDIENCE: You have a lot of hard to find, high-powered people. I wondered what sorts of things you're doing, since there's going to be -- there is a battle for talent, what kinds of things Akamai does to retain those people as long as possible, and of the things that you do do, what do you think is the most important?

SAGAN: Well, one of the ways we're able to recruit effectively is by having people like Tom Leighton as founders. He's a magnet in a certain part of the computer world to just get the best people. He's just known in a small community.

One of the things we've done to find more people is open an office in the Valley, there's clearly not -- there aren't enough people in one place on the Internet I think to build massive technology groups. If you look at IBM, CISCO people like that have campuses, Sun, in a variety of places. Often it's the Boston area and Silicon Valley or the Bay. We've mimicked that if you will by opening our San Mateo office.

My guess is that we'll open other R & D centers over time. Not quickly, but over time is we get as big as we think we have the potential to.

To attract and retain people we're very generous with options. Everyone in the company is a shareholder and right now a very, very happy one.

We pay competitive wages, it's not cheap, we have people who are employees in two of the most expensive places in America right now to live, and stock options are great but they don't pay off for a long time in a young company, so you have to be able to own a home or pay their bills or rent a nice apartment, and so we pay people competitive wages.

But I think most important if you have to have an environment where they want to come to work. Because in this current economy -- I think somebody told me that Boston now has negative unemployment among college educated people. It means there are some people are holding jobs and an offer and trying to decide which one to do. It's really competitive hiring.

So lots of companies with great options and opportunities, and so I think the way you keep them is to make it a great place to work. They have to feel that they're productive, that they count, that they can have an impact, and that they're doing something that's fundamentally important. And our pitch and our advertising is delivering a better Internet, inside the culture it's inspired by Tim Berners-Lee, who's on our advisory board.

That's a pretty cool idea and if we make it fun, which means lots of free pizza and snack food, it means a game room and foosball. It means set your own hours, it means take work home. It means work in spurts and collapse. It means you can roller blade in the office and nobody's going to chase you away and say, "The insurance says you can't do that." Those are the kinds of things that are critical in today's environment. You can wear anything you want to, just wear something. (laughter) But that's -- that's kind of the -- I think that's the way you really do that. Yes.

AUDIENCE: Yes, Silicon Valley I (inaudible) a lot of attention, can you just briefly describe the Cambridge atmosphere?

SAGAN: The atmosphere in Cambridge? Well, the weather's more like here than Silicon Valley, I think it's a great place to find technology people.

Boston had a hey day a long time ago, and then with Wang and Digital and kind of the downsizing of that model, and kind of the ebb and flow in medical engineering. Hit some relatively hard times. Those are way past. There is a hotbed of Internet start-ups spinning out of MIT. Great minds and people who fled out of Digital as a place to recruit more seasoned engineering folks. There are a lot of incubator type places you can rent a cube by the day and keep expanding and they'll get you a phone and fax kind of service. The kind of stuff you start to see in the Valley. Places where people hang out and talk about their businesses.

We're now in a cluster which has two big companies, Akamai and Forrester and the Computer Lab, and miscellaneous other floors in some of the buildings are people like Frictionless and Curl and a Viant office and Sapient is down the street and there's a place called -- what's it called, it was called One Kendall, now it's called Two Main or something, but it's a rent an office place, and there are 50 companies starting there and many tier one VC-funded companies.

So it is much like the Valley, but it hasn't gotten to the point where we have to do cleaning for our employees so they stay at work, or whether they get recruited with flyers in the parking lot. So it's not quite the stories you heard about Netscape in the craziest days, but it's a pretty intense, Internet-focused kind of environment.

It's also geographically a little more spread out and you get just that intensity in the Valley and Sycamore is out in the suburbs and Site Path is in Waltham and Direct Hit moved to Wellesley, which is 45 minutes from us. They were an MIT spin-up. But they're a bit farther away so they're sort of intermingled with other Boston businesses. I think it's a great place to live. Yes.

(Question from audience)

SAGAN: In terms of utilization -- utilization can be very, very small, because our machines -- these enterprise class Pentium servers are very inexpensive, so the return on thousands and thousands of them is huge. So we don't have to be very efficient. And often our co-location and bandwidth comes with no minimum, so we can go to places and then if there's no traffic we don't lose a lot of money to do it, so we can become profitable at a very small percentage utilization. It's like a telecom model you can't start selling cell coverage until at least you think you've built all the cells, so the phone works everywhere. And so first you get great performance and then you've got to keep expanding the cells or at some point you can't get a dial tone. But it's much the same in the first step, We had to build a global network before Yahoo will do anything with us. Then the key is we keep ramping customers. We have to keep building it fast enough to have much great -- very, very, very large overhead so that spikes in demand don't effect performance at all.

MODERATOR: Take one last question.

SAGAN: One more. OK, one here.

AUDIENCE: How does your service compare to that of Digital Island's?

SAGAN: Well Digital Island is a hosting company that bought a company called Sandpiper, which is I think why you'd ask. Sandpiper has a head-on marketing message, they sound exactly like us. Dead-on competitor. They really represent a solution that's off the shelf. They buy caches and software from some people. They have a lot of fault in tolerance in their network.

We have an integrated solution. I could tell you how our algorithms are better, and I fundamentally believe that. The best answer though is our customer list. Look at who selected us. They're all expert buyers. Yahoo doesn't do anything without testing. Everybody. Most of the names of our customers did head to head shoot-outs. We never lost one. If you look at their announced customers and actually look at some of our recent announcements, NBC is an investor in Sandpiper, is our customer today. Ad Force was an announced customer is now an Akamai customer.

So I think the market has voted and totally on the slide return, but the fundamental answer is we're focused on superior technology in the marketplace and superior customer service, and that's how I think we beat all competitors who come in this space. Thank you very much. I enjoyed it.

MODERATOR: Paul -- Paul we've just a small token of our appreciating, making a trip back to Evanston, the other side of the campus, thanks a lot.

SAGAN: Thank you.

MODERATOR: Appreciate it.

SAGAN: Thank you. Thanks.

END

Transcript by Fantastic Transcripts

[ Home ]