Urs Holzle, senior vice president of technical infrastructure and a Google Fellow, spoke today at the Collision conference in Las Vegas about avoiding failure, not being lazy and when Google will "turn on the faucet" for its cloud business .
Holzle, who was interviewed on-stage by Forbes managing editor Bruce Upbin, is known as the man behind Google's vast infrastructure, designing its data centers and focusing on server efficiency and power usage.
"Being responsible for Google's tech infrastructure makes him one of the busiest, and most important, people in the tech industry," said Dan Olds, an analyst with The Gabriel Consulting Group.
Holzle, who was an associate professor of computer science at U.C. Santa Barbara before becoming the eighth person hired at Google, said the reason the company has done so well in the search market is because it has a better, faster infrastructure.
Zeus Kerravala, an analyst with ZK Research, noted that Holzle has played a significant role in Google's success but he's also had a good spot to watch an interesting time in tech's history unfold.
"Clearly, he's seen all the innovation over the years and used the right technologies to put Google on top and keep it there," he explained. "It was the infrastructure that let Google operate at a speed and with efficiency that no one else in the industry could.
"I remember when Google launched, no one really thought there was the ability to innovate in search , but Google did. Think how many search companies there were back then. They simplified the experience and did search better than anyone."
A sampling of what Holzle talked about at the conference:
When asked what it was like being the eighth person hired at Google, Holzle remembered some rocky times the company struggled through.
"I'm amazed we made it this far," he said. "We had plenty of chances to fail and, in a few cases, came pretty close to it."
One of those times came when the dot.com boom came to a sudden end in the early 2000s.
"A year after we were founded, the bubble burst and it was impossible to raise any funding," he explained. "We were profitable on an incremental search basis, but not overall, so we were slowly draining our cash. We were trying really, really hard to reduce our infrastructure cost and reduce all costs, our run rate, to get to the point where our volume was high enough where the incremental margin on every query actually paid for the company.
"We made it fine, but I think it was within a year of not working," Holzle added. "Had we taken one more year, it might not have worked out."
Holzle also called Google's size and success both a blessing and curse.
"A curse because you can afford to be lazy and you have to manage that specifically," he said. "That's really a culture question. Fortunately, we've spent a lot of time hiring really good people and they're motivated by accomplishment and not by having an easy life. That's one safety net."
Of course, the company's market dominance also gives certain advantages.
"It's a blessing too because for the first two or three years at Google, we couldn't really do anything ambitious because it was really about surviving the next week," said Holzle. "That constrains your horizons so much you have to make one compromise after another just to solve the problem of the day. With a little bit more self-generated money, you can really change the world."
Working so hard to survive in the early years makes people hungrier -- and they work harder.
Holzle also noted that Google has not picked up the pace of marketing for its cloud business because it wants to focus on the technology before opening the floodgates.
"We have definitely not turned up the marketing faucet as much because we want to convince people by the quality of the product and not the quality of the ad," he said. "We're actually growing very quickly and we're having trouble keeping up with the demand. You don't want to turn on the faucet and have people come to you and be disappointed."
Holzle also said the next five years will bring much greater advances to the cloud than the last five years have.
Google's infrastructure head also said that today's startups have it much easier than his team did -- and that's the way it should be.
"I think there can never be enough startups," he said. "Today you can push a button and have another hundred virtual machines. That has lowered the barrier to entry and that's very, very good."