Steve Perlman rushes out of a conference room, laptop under his arm, power cord dangling to the floor. He's wearing a lime green tropical shirt over his burly frame, his short thinning hair gelled to stand on end. It all gives him the appearance of Tony Soprano crossed with a geek from the A/V club.
After seven years in stealth mode, Perlman feels pressed for time as he closes in on the winter launch of OnLive, a service that threatens to upend the $60-billion-a-year video-game industry. He's funny about OnLive's prospects. As he strides through the company's Palo Alto headquarters, he caroms between circumspection and hubris. "I've always been a percentage hitter with my startups," says Perlman, who previously founded WebTV and Moxi Digital. "I've never swung for the fences. I just try to get on base." Twenty seconds later, he suggests that OnLive could eventually go beyond gaming and transform business.
OnLive promises to let users play high-end streaming video games over the Internet -- and the games are supposed to feel as responsive as those played on a console. For the first time, games from most major makers will live in the cloud, run on a server in a data center, and stream to players -- no disks, no downloads, no consoles.
The prospect is tantalizing. Users could play the processor-hogging Crysis on a sputtering old PC at work, and later pick up the action on a TV screen (HD or not) at home. OnLive's technology eliminates almost all lag from the high-def video sent over the Internet to your screen. Conventional wisdom said it couldn't be done over today's Internet -- until Perlman unveiled an OnLive demo at the Game Developers Conference in March. And earlier this year, the investment firm Wedbush Morgan reported, "We were given a private demo of OnLive. We were blown away." But OnLive still has to prove the system works at scale; at press time, it was launching a beta test to find out.
If OnLive catches fire, it could rip apart the game business much as downloads scrambled the music biz. "It's the advent of enormous changes in the industry," says Brett Close, CEO of game maker 38 Studios. Selling streaming games should cut production and distribution costs. And lower costs would allow developers to take chances on games that might not be blockbusters. "I'm excited about it," Close says.
But retailers would lose. The used-game economy could shrivel, hurting companies such as GameStop. OnLive foretells a doomsday scenario for game consoles; Xbox and PS3 have services that allow users to download games to the console, but OnLive makes consoles obsolete. And consumers would be able to play the most advanced PC-based games without having to buy a $2,500 Alienware machine -- bad news for PC makers.
Perlman, 48, has never attained superstar status in Silicon Valley, despite his habit of landing in the midst of interesting developments at interesting times. Early in his career, he led development of QuickTime for Apple and cycled through General Magic, which tried to single-handedly create the Web before the Web existed. In 1994, he cofounded Catapult Entertainment, which made modems for game consoles, enabling gamers in different locations to play each other. The next year, he cofounded WebTV, which Microsoft bought for $503 million during the dotcom bubble.
In 2000, Perlman created an incubator called Rearden (named for the character in Ayn Rand's Atlas Shrugged). It has so far produced three companies: set-top box maker Moxi Digital, which merged with Digeo in 2002; Mova, a motion-capture company that provided technology used to make The Curious Case of Benjamin Button; and now OnLive.
"They're all pieces of a puzzle," Perlman says. "It's about communications, graphics, audio, all moving toward creating an immersive experience." Much of his work has involved compression technology -- ways to break up and send audio or video in ever-smaller digital packages. Those schemes have always focused on delivering higher quality without hogging bandwidth rather than shortening lag time, or latency, as it's known in the sector. In fact, compression has always relied on latency to give the software time to anticipate the next frame of a video.
Video games, though, present a devilish challenge. There can be no perceptible lag between, say, pulling the trigger and the bullet firing from the on-screen gun. Latency has to be less than 80 milliseconds or the brain will notice it. Considering the messy path that packets of information take when moving from data center to home, crunching latency to less than 80 milli-seconds seemed impossible. "That was the hard question we needed seven years to figure out," Perlman says.
One big breakthrough: OnLive sends two streams to the gamer. One is opti-mized for speed. The images are crisp enough for HD while the game is in motion, but if you could pause it, each frame would look blurry. The second stream is optimized for quality; each frame looks high-def but lags the game by a split second. Gamers, though, will never know. Pause, and OnLive instantly switches to the high-quality feed; the frozen frame looks perfect. Resume play, and it switches back to speed feed.
This engineering success in 2006 enabled Perlman to reel in some all-star gaming and tech executives, though each one walked into his first meeting with Perlman as a doubter. Mike McGarvey had been CEO of Eidos, which created the Lara Croft Tomb Raider games. "They said they were working on a technology that would change everything, and of course, I said, 'Yeah, right,' " McGarvey recalls. "So I met with them, and realized right away it would be paradigm shifting." He's now OnLive's COO. Bruce Grove had led engineering for high-end gaming PCs at Dell. "When they told me what they were trying to do, I said, 'You can't do it,' " he says. "They turned a crummy Dell laptop around and showed Crysis playing and I had to eat what I'd just said." He's now director of engineering at OnLive.
One of the harsh truths of Silicon Valley is that cool technology doesn't necessarily equate with a successful venture. From 3DO to PointCast to Webvan, there's a long list of "game changing" advances that didn't succeed, and Perlman learned that himself at General Magic.
OnLive could fail too. It has no brand recognition. It has been expensive to build and will be expensive to roll out. OnLive has rented space in five data centers, costing millions of dollars. And potential rivals are showing up. Playcast, an Israeli company, claims to be doing something similar to OnLive, though its focus is on game delivery over cable-TV systems, not the Internet. Perlman says he's certain that OnLive is ahead. "We filed 100 patents around this technology," he says. "It's that big an invention. This stupid thing was very hard, but I knew damn well it could be done."
One of the keys to OnLive's success will be to price the service attractively. The company hasn't finalized pricing yet, but there will be a subscription fee. Users will get an Internet-interface box -- about the size of a deck of cards -- for their TV, plus plug-in controllers. Then they'd buy access to specific games -- for life, or just for a year or a month.
Wedbush Morgan suggests a subscription will have to be $5 a month or less to convince gamers that it's a better value than a one-time $300 console pur-chase. Casual gamers might relish the chance to play high-end games without buying high-end hardware, and they may be lured by OnLive's social functions. "Gamers will have brag clips -- recordings of the last 15 seconds of whatever great thing you did in a game -- which you can post," says Perlman. "You can watch the top players in the country playing a game you don't have."
OnLive will prove a tougher sell to hard-core gamers. Blog posts from them raise doubts that OnLive's service will have the speed of a console or a high-end PC. "I will be surprised if it does," says Jonathan Wagner, who writes the Musings of a Web Producer blog. "None of the hard-core gamers I know have any interest in it."
Still, if it survives, OnLive could be transformative -- such that Microsoft might want to buy it and turn it into the next-gen Xbox. (Remember that Perlman sold WebTV to Microsoft, so the connection's there.) And if it scales as Perlman hopes, OnLive's technology could also deliver business applications, especially those that rely on graphics. It's no coincidence that one of OnLive's investors is Autodesk, a titan in computer-aided design software.
Whatever happens, OnLive has already changed the very mind-set about video games. "Ultimately, developers will have to get their heads around this," McGarvey says. OnLive is opening a window to the future -- be on the lookout for cloud-based gaming. If it doesn't pull this off, some company eventually will, and the video-game business will never be the same.