2
Chapter 3
October 1996, Carnegie Mellon University, Pittsburgh, Pennsylvania
âWeâve got a scrappy little startup out here in Berkeley thatâs working on some very exciting things, perhaps things that could change the world.â
In October 1996, I got a well-crafted email from Todd Brin. âWeâve got a scrappy little startup out here in Berkeley thatâs working on some very exciting things, perhaps things that could change the world.â Would I like to be a part of it?
At that point I had been on the faculty at Carnegie Mellon University for four years. I was a research professor and had my own DARPA contract. DARPA is the Defense Advanced Research Projects Agencyâthese guys brought us the internet by funding Computer Science research back in the â60s. While having a DARPA contract to do research is a big deal, that alone wonât keep you employed at a top university.
I had the feeling I wasnât doing great at CMU, and I would be out of a job before too long. I didnât have a lot of publications, and my research wasnât making a big impact. Around the time of the email, I had had various meetings with more senior folks in the department who were trying to give me guidance. If youâve ever worked at a big company, you may know that there are certain procedures to firing someone. Part of that process includes some well-documented meetings where expectations are very explicitly set. These meetings had that sort of feel.
A great academic knows not only how to wow the graybeards but also bring in the grant money. I was a bit of a conundrum for the department. I brought in the money, but my work didnât have the academic flair that resulted in publications or academic kudos. They suggested I become a DARPA program manager, and I researched what it was all about. Basically you switch sides and decide which research to fund.
This was a very intense job and very political. It was a government position that entailed a lot of power over where big money was spent on academic institutions. You would hear all the latest research and try to figure out what could make an impact, and then you would fight for those programs. It was a stable government job with nice government benefits.
I liked the idea of being in the center of all the action. There was definitely a Star Chamber feel around the DARPA guys. But I didnât like the fact I wouldnât be part of the action, more of a facilitator than a doer. The Inktomi job seemed like the opposite of the DARPA job. With Inktomi, Iâd be on the ground making something new and exciting. That said, Inktomi was risky and underfunded. At this time my wife Loan was six months pregnant with our first child, and we were in the middle of renovating a house we had just bought. It didnât seem like the ideal time to change jobs or move across the country.
But Loan is very adventuresome and entrepreneurial. She and her family had escaped Vietnam on a decrepit overloaded boat after the fall of Saigon. They came to America with nothing and rebuilt their lives. Her risk tolerance was in a whole different category from mine.
As a child she saw explosions on the streets of Saigon. She had seen her family risk it all by moving halfway around the world to a very foreign country where they didnât know the culture or the language. She had experienced extreme risks in her life and seemed to take it all in stride. She has an innately positive attitude that things will work out.
That said, risk wasnât completely foreign to me either. I had grown up around risks, perhaps more subtle risks, but risks, nonetheless. My father had a very successful medical practice, but he had an entrepreneurial streak. He started a string of businesses, most of which failed, but the success of his medical practice carried us through these downturns without much problem. It taught me that failure is a very real part of risk, to be careful of your limits. Loan encouraged me to check it out, despite the fact we had several big projects in the works.
Amy Hanlon was the perky redheaded twenty-five-year-old head of HR for Inktomi. Her enthusiasm and resourcefulness reflected that of her employer. She arranged my visit and my travel. This was 1996 and you could buy airline tickets from people on the internet. The tickets were still paper. I got the ticket via FedEx, and it still had some other guyâs name on it. Amy assured me it wouldnât be a problem.
In todayâs post-9/11 world, this seems particularly sketchy. It was a risk, but what was the worst that could happen? They wouldnât let me on the plane? I flew out to SFO and rented a car and drove to Berkeley and checked into my hotel. I didnât know it then, but the next day would change my life forever.
* * *
I arrived early to the Inktomi office, so I headed across the street to the Sonoma Valley Bagel Company. I had never been to Berkeley; the weather was great, fresh, and sunny. Could it be an omen of things to come? Over my coffee and bagel, I thought over what Todd had told me. The search engine was only one of the first products Inktomi would produce. What else could they be cooking up?
They had a great start on search at that point. In May 1996, HotWired (the predecessor of Wired Magazine) launched HotBot, a search engine powered by Inktomi. HotBot impressed the world by having very fresh results. They could crawl the entire internet weekly, making sure the search results were the latest and greatest. This sort of deal is called white labeling.
Inktomi provided the back-end technology for search, but the front-end design and branding was controlled by other companies. HotBot was Inktomiâs first search white-label deal, and it was well received. Inktomi had received amazing press, being known as the search engine with the biggest (50 million pages) and most up-to-date (weekly) index of the entire web. However, they didnât think search was a big enough market. Search seemed important, but it wasnât obvious in 1996 it would be as huge as it is today.
At that time there were other ways of finding things on the internet. Yahoo! had an index built by humans and it was the most important website of the times. Inktomiâs underlying NOW (network of workstations) technology could be applied to more problems than just search. Their strategy was to provide multiple internet infrastructure products.
I had no idea what the next product might be. Search seemed like an exciting new business. I couldnât wait to find out what they might be working on. It could be something revolutionary or a total waste of time. Since nine out of ten startups fail and they already had what seemed like a successful product, why risk adding something else? In the end, that risk paid off.
It turns out they were working on a high-performance proxy cache. In those days the internet was growing like crazy, and there seemed to be no end in sight. From 1995 to 2000 the number of users on the internet grew from 16 million to over 300 million. Today weâre at 4.5 billion users on the internet. The frictionless economy and irrational exuberance were around the corner, but right then, bandwidth was a big problem, as the network had much more demand than the actual pipes could supply.
Netscape had a blockbuster IPO the year before. On opening day in August 1995, Netscape stock more than doubled from $28 to close at $58.25 after peaking at $74.75. The sixteen-month-old company was worth around $3 billion without having any profit on the books. Things were starting to get crazy. A proxy cache was a great idea. It was a software solution that could make the internet scale better and seem faster to the end user.
A proxy cache is software that runs on a server in the middle of the network and speeds up the network by caching parts of the web so they are stored closer to the user. Today they are an integral part of the internet. In 1996, few people were using them, and no one had a really good one. Inktomi was going to build a super-fast, super-efficient proxy cache and sell the software to internet infrastructure companies like @Home, AOL, and even Enron.
There was a huge demand for the caching product, and it made Inktomi much more money than search ever did, but in the end that market would have a dramatic collapse, taking Inktomi down with it. At the peak, Inktomi was making 60 percent of its revenue from caching. However, revenue went from $80 million to $40 million in a single quarter when the telecom market started to fall apart.
* * *
I thought I might end up being a millionaire if this whole startup thing worked out. I knew people got rich doing startups, but I didnât know the details. I also knew my professor salary wouldnât cut it in California. We had a good life in Pittsburgh where I could afford a big house. Most importantly, we had a baby daughter on the way. Was putting everything at risk worth it? For a scrappy startup with no real revenue at the time?
At that point only about twenty people were working there. I met with most of the key engineering staff. Everyone seemed very excited, very smart, but also very fun. At the end of the day, they decided to make me an offer. Jerry Kennelly was the new CFO. He sat me down to present me with my offer. He led with the stock, which was 50,000 shares. NaĂŻvely I didnât understand that this was very generous for employee twenty something. My PhD in Computer Science and CMU pedigree was paying off.
But the salary didnât impress me. I didnât know it at the time, but this is standard procedure for startups. You offer below market salaries, since employees stand to make out like bandits should the company have an IPO and the price of the stock rockets. Employees support the company with low salaries and get compensated later if all goes well. Of course, early-stage startups often fail, then the under compensated employees end up with smaller bank accounts for their efforts. Itâs a gamble. Conventional wisdom tells us itâs worth it.
Even if you donât make money off the companyâs stock, youâre getting valuable startup experience, which can lead to a better job at the next startup or perhaps a bigger, more stable company. If this company doesnât take off, youâll be better equipped to make the next one a success. This is mostly true. If youâre skilled enough to join a startup, those skills are highly marketable. If things go down the tubes, youâll be able to land somewhere else. While this is particularly true when the economy is humming along, it will be tougher when the entire market crashes, like it did in 2000.
After the dot-com crash, Silicon Valley emptied out. There is a corridor along Highway 101 from San Mateo down to Palo Alto known for billboards with ads for tech companies. In the good times, these billboards are filled with ads from tech companies, sometimes taunting each other. Itâs not uncommon for a billboard in front of Oracle to be used by a competitor to throw insults at the software giant. However, in the years after the crash, these billboards had a different flavor, pushing very different products, like beer and fast food. For a while, Silicon Valley became the valley of Budweiser, but it didnât last long.
Of course, I didnât know any of this at the time. I just knew that the salary seemed low. I called my wife and we decided that the salary just wasnât ...