Cloud gaming’s history of false starts and promising reboots

During the summer of 2011, Robert Stevenson and Tim Wilson of Gaikai, a cloud gaming company in Aliso Viejo, California, boarded a 13-hour-plus flight bound for Samsung Digital City in Suwon, South Korea. Their goal: to wow a room full of almost 40 Samsung employees and executives with playable demos of several games, including Mass Effect, running off a server in California over 5,700 miles away. But smoothly pulling it off at Samsung’s high-security headquarters required ingenuity, given that the meeting room was isolated from the internet behind a layer of security, making unfettered web access problematic. To get around this bottleneck, they MacGyvered a solution by having someone lean out a window and run an Ethernet line down the side of the building, while someone else several floors down plugged it into a router.

“You do all this demo prep, but this was one of those things you don’t quite anticipate,” recalled Stevenson, Gaikai’s former chief product officer and executive vice president. “But we did it, and by the hair of our chinny chin chin, the demo worked. We played these games, and it was quite compelling.”

Gaikai’s demo that day clinched a deal with Samsung that allowed the South Korean consumer electronics giant to serve up game demos, running on servers across North America, on Samsung televisions. The high-profile partnership became one of many, including deals with Facebook, Walmart, Best Buy, LG, and Google, that made Gaikai an attractive target for Sony, which acquired the company in June 2012 for $380 million. The startup’s core streaming technology became the foundation of PlayStation Now.

In May 2020, Sony announced that PlayStation Now has 2.2 million subscribers, but that number is modest compared to the PlayStation Network’s 103 million monthly active users and the 106 million-plus PlayStation 4 consoles sold so far. The PlayStation Now subscriber base reflects the stark reality of cloud gaming overall.

Many cloud gaming services over the last 15 years, like OnLive and GameFly, have come and gone, dangling visions of high-quality, low-latency gaming experiences crunched on remote servers and piped to PCs, consoles, tablets, and smartphones miles away. But often, technical issues and a lack of game exclusives prevented those services from offering cloud-based experiences that rivaled games running locally, and a healthy dose of consumer skepticism ensured these services stayed on the fringe. While a wave of newer cloud gaming services — Google Stadia, Nvidia’s GeForce Now, Microsoft’s xCloud, and Amazon’s recently announced Luna — aims to change that, the battle for mainstream acceptance may be as crowded and as difficult as ever.

“The biggest challenge with cloud gaming is the experience has to look like playing a game without actually buying a console,” said Michael Pachter, managing director at Wedbush Securities, who advises clients on making decisions in their investments. “Everything you do has to be processed in the cloud, so as to travel over the internet to the server and rendered back to you and me so we both see it, so massive amounts of data are going back and forth. A small amount of packet loss can completely screw up the signal. If you can’t figure that out, everything else is a non-starter.”

---

To understand when cloud gaming as we know it began, we have to go back to March 2009, when Steve Perlman unveiled OnLive at the Game Developers Conference. Inspired by several technological advancements at the time, including better data and video compression techniques and the increasing ubiquity of capable smartphones, Perlman felt the time had potentially come for gaming to move into the cloud — and at GDC, that’s essentially what he promised. With OnLive, gamers could play “games on demand,” paying a subscription fee, plus additional fees to rent or buy games, for access to titles running in resolutions up to 720p at frame rates up to 60 frames per second.

“The desire to create narrative experiences that were just going to be completely photorealistic drove me to go and say, ‘We’ve got to build this thing,’” Perlman told Polygon in a recent interview.

By 2009, Perlman had already carved out an impressive reputation for himself in Silicon Valley. He’d helped create multimedia software that eventually became QuickTime, and he’d created and sold WebTV to Microsoft in 1997 for $425 million. But in the tight-knit gaming community, he was a relatively unknown commodity with a lot still to prove to companies and gamers.

“I think it was hard for Steve to navigate the industry because it’s so insidery, and he was on the outside. Steve’s ego couldn’t allow for the fact that people didn’t respect his genius, kind of, even if it was as simple as executives from a gaming company coming over to say ‘hi’ to me before him,” explained a longtime game industry veteran familiar with the matter. “The games industry is really difficult to get in, and then once you’re in, it’s sort of like going to high school in a small town: You see these people over and over again, and you recombine in different formats.”

Could Perlman become an industry insider himself, and, in the process, convince an entire industry his cloud service was worth embracing?

He certainly tried, using expensive marketing campaigns, big booths at GDC and E3, and an aggressive push inside the Palo Alto, California, company to land games for the OnLive platform. Although game publishers like Ubisoft, 2K Games, and THQ contributed to a catalog of 10 or so titles for OnLive at launch — a number that grew to more than 100 within a year — landing more AAA games in the months and years after launch proved challenging. Many other publishers, long accustomed to the concept of selling games at fixed prices upfront, were wary of OnLive’s Netflix-like model and what it could mean for the industry if the startup became popular.

Publishers were more receptive toward Gaikai, a cloud gaming service run by David Perry, a well-liked industry veteran behind games such as Earthworm Jim, MDK, and Enter the Matrix. Gaikai’s focus on offering game demos rather than full titles made it a non-threat from a publisher standpoint. As Gaikai landed more deals and partnerships, OnLive struggled to sign new top-tier titles to play, and three former OnLive employees told Polygon that OnLive generally came to view Gaikai as a serious competitor.

OnLive launched on June 17, 2010, and Gaikai announced a multiyear deal with Electronic Arts that same day. Perlman threw a fit in a conference room, according to former OnLive employees, shouting and demanding that they immediately yank EA’s games off the service, including titles that were already being tested in beta form. Employees told Polygon that they had to close the conference room door and turn up the music in the room to drown out Perlman’s yelling. Perlman, for his part, didn’t recall shouting, but he did admit to being upset on a call later with EA’s then-CEO, John Riccitiello, about the situation.

While talking with Polygon, Gaikai’s David Perry reflected on the rivalry between OnLive and Gaikai then as a particularly intense time.

“It meant that every time I walked out of a meeting, some OnLive person would be sitting in the lobby for their meetings, meaning we were meeting with the same people,” recalled Perry. “It was quite intense — it was real competition, but it was also motivational.”

Gaikai has never publicly disclosed the number of people that ultimately signed up for its demos. Stevenson suggested that because of its free-to-play demo streaming focus, there wasn’t a “signed-up user base” in the traditional sense. OnLive, for its part, had 275,000 monthly active users, 150,000 weekly active users, and 50,000 daily active users, according to internal data obtained by Polygon. Those numbers fall short of PlayStation Now’s 2.2 million subscribers, and likely didn’t help OnLive’s case in the eyes of any game publishers that were already wary of putting software on the platform.

John Spinale, former senior vice president of games and media at OnLive, compared the startup’s content problems to the chicken-and-egg conundrum.

“How do you get publishers and developers to support your platform when you have no users, and also you’re not going to have users until you have some content?” said Spinale.

In 2012, HP approached OnLive about acquiring the startup. The tech giant even extended a $15 million bridge loan to the company, which was burning through $5 million a month at the time, with roughly 20% of that going to server costs and the rest largely spent on paying its 200 or so employees. But in July 2012, HP backed out of the deal and offered little explanation at the time, leaving Perlman and OnLive on the hook to pay back the $15 million loan and find another buyer. One month later, Perlman found one in Gary Lauder, managing director of the venture capital firm Lauder Partners. He purchased OnLive’s assets in August 2012 through a newly formed company called OL2 for $4.8 million — less than 1% of OnLive’s reported $1.1 billion valuation two years prior.

Although OnLive initially said that Perlman would stay on as CEO, he departed the company “to work on myriad projects” that same month. Lauder then worked with acting CEO Charlie Jablonski, OnLive’s former head of operations, to salvage the company; in an effort to reduce spending, they laid off two-thirds of OnLive’s employees. Further attempts at a turnaround included the launch of a new feature called CloudLift that detected games users had already purchased and attempted to stream those games from virtually any device. But the pivot didn’t click.

“At the time that I saved the company, you know, cloud gaming was still a hot thing,” recalled Lauder. “But partly as a result of the negative press associated with the layoffs and the transition — which, I can say in retrospect, we didn’t handle properly — cloud gaming became considered one of those areas that was kind of a failure.”

Sony acquired OnLive in April 2015, only to shut it down four weeks later, while Gaikai had become pivotal to Sony’s development of PlayStation Now. Looking back at OnLive years later, and its attempt to popularize cloud gaming, Perlman remains proud of what the startup accomplished as a pioneer in that space and wistful about its demise. As several former employees lamented, OnLive was a classic example of a product or service that was ahead of its time.

“We were a small ragtag team of people, who in a very short amount of time during the Great Recession, advanced the state of the art a decade earlier than anybody else could do it,” said Perlman. “It’s just too bad that we made it within inches of the finish line, or within inches of safety, because if the HP [acquisition had happened], of course, we could have kept going on forever.”

---

Despite OnLive’s flameout, cloud gaming efforts soldiered on. Sony debuted PlayStation Now in beta form in July 2014. Roughly during the same period, Nvidia and Google began exploring cloud gaming services of their own.

In 2013, Nvidia began testing a service that took a different tack. Instead of streaming games players didn’t own, Nvidia Grid allowed users to stream games they had already purchased on storefronts like Steam. The service, later renamed GeForce Now, worked on Macs, Chromebooks, Windows PCs, Android TV, and mobile devices.

“PC gamers told us they didn’t want another walled-garden game store, and most already have Steam, Epic and Uplay accounts with games that they own,” said Andrew Fear, senior product manager for GeForce Now. “And, increasingly, they play free-to-play games.”

Since GeForce Now exited beta in February 2020, the service has signed up over 4 million users who now stream for 15 million hours every month, according to Nvidia. But GeForce Now’s unusual model also raised copyright concerns. When a user purchases the digital key to a game, it’s usually with the express understanding that the game will be played in a specific format. But GeForce Now changed that dynamic. Yes, the purchased digital game is technically the player’s to use, but was it kosher to play it in a different format other than how the publisher intended? That’s not to mention other possible issues, like the challenges for developers to fix bugs in the game that arise because it’s being played on a different platform.

As a result, GeForce Now’s new model didn’t go over well with some developers and publishers. Activision Blizzard and Bethesda Softworks pulled their games from the service weeks after it launched, offering vague statements suggesting they were displeased with the lack of a revenue split for their games appearing there. Meanwhile, Raphael van Lierop, the game director and writer of The Long Dark, pulled his game in March, frustrated with its inclusion in GeForce Now without his explicit permission. (Two months later, van Lierop relented, and The Long Dark returned to GeForce Now after Nvidia shifted to an opt-in program, letting developers fully control whether their games appear on the platform.)

“Today’s world is getting complex for devs, with lots of platform changes and shifts to streaming, so devs have to be able to plan a strategy for how their games will appear and where, as a means of running a business,” van Lierop wrote in a tweet in March.

Shortly after Google launched its Chromecast line of streaming-media devices in 2013, the Chromecast team began prototyping ways to enable gaming through them. At the same time, the Google Fiber team was also building a cloud gaming prototype, exploring what was possible with high-bandwidth, low-latency connectivity. When these teams joined efforts in 2014, they ended up working on what eventually became Google Stadia.

In October 2018, Google launched a closed beta for its in-development cloud gaming service, which it called “Project Stream.” It let users load up Assassin’s Creed Odyssey in a Chrome browser. Project Stream proved popular among people in metropolitan areas with high-bandwidth connections, and so at first, the entire venture seemed promising.

When Google Stadia was announced at GDC 2019, expectations ran high. At least some gamers expected an actual console announcement from Google. Instead, what they were shown was another streaming service, albeit a promising one, as well as a proprietary controller and a lot of onstage chatter about the technical wizardry powering the service.

“Stadia is the first major platform introduced to the industry in fifteen or so years, so there are a lot of behaviors, habits and ideas that gamers have when it comes to how and where they play their games,” said Majd Bakar, Stadia’s vice president of engineering. “So, convincing them that, yes, you can play your favorite AAA level games on virtually any screen has been something we have faced.”

Google finally launched Stadia in November 2019 to a mixed reception. The service was lauded for having higher image quality than other rivals and making the gameplay transition between PC and smartphones a seamless one. But Stadia, like many Google services at launch, felt incomplete. A number of promised features, including wireless Stadia controller support and 4K resolution via web browser, were missing. Concerns about input lag remained, as well as fears that Stadia would be a short-term effort, doomed to eventually join the digital graveyard of other discontinued Google services like Google Plus, Google Reader, and Google Express.

Google, for its part, told Polygon that building out an ambitious service like Stadia takes time. As the service approaches its one-year anniversary, one of the biggest concerns around Stadia is its small number of compelling exclusives — an issue the company is working to fix with new games developed by the Stadia Games and Entertainment division, run by Ubisoft veteran Jade Raymond.

Meanwhile, after nearly a year of public beta testing, Microsoft launched its streaming service, xCloud, in mid-September. xCloud is currently available for Xbox Game Pass Ultimate subscribers. (Update: This paragraph has been changed to clarify the availability of xCloud.)

The xCloud service was the brainchild of Kareem Choudhry, Microsoft’s corporate vice president of gaming cloud and a 22-year company veteran who had previously led Xbox’s software engineering side. In early 2017, Choudhry began having conversations with key members of his team about the possibility of developing a new service that leveraged what he liked to call the “three C’s” — “content,” “community,” and “cloud” — so people could play high-quality console games streamed from a Microsoft Azure data center to their mobile devices.

“If you had told me five years ago that I’d watch a whole season of Stranger Things on my smartphone, I’d think you were crazy, but that’s kind of reality now,” said Choudhry, who points to 5G as a potential future catalyst that will make high-bandwidth connections more accessible and cheaper around the world. “And you know, music has made the jump, video has made the jump, and from a consumer expectation perspective, I think it’s now gaming’s time.”

Convincing publishers to let their games onto Microsoft’s service took time. At that point, Choudhry’s team had not yet deployed its xCloud hardware — modified Xbox One S consoles — into Azure data centers. So when Choudhry and his team made the pitch to partners, often in hotel rooms, they traveled with they called internally their “cloud in a box”: an early prototype of the hardware that they’d set up in the next room.

“There would be one of my engineers with the ‘cloud in the box’ in the next room, and we would be in the suite preparing for the meeting,” Choudhry recalled. “Then the CEOs and creators and founders of various game companies would come in ... and we would just start those meetings by handing them a phone, and a tablet with their games on it.”

There’s a lot to like about xCloud, which Microsoft folded into its $14.99-a-month Xbox Game Pass Ultimate subscription. The service provides access to a revolving library of more than 100 downloadable games — including, as of this writing, Doom Eternal, Forza Horizon 4, and Halo 5: Guardians — for Windows PC and Xbox One. And with the arrival of xCloud, subscribers can stream those same games to Android devices. It’s a strong offering, but the lack of support for iOS devices (for now) will disappoint some. It’s also just plain odd that the streaming aspect only really applies to Android devices; playing on a PC or Xbox means having to still download the games.

Barely two weeks after Microsoft rolled out xCloud, Amazon announced Luna, which will runs on Amazon Web Services and stream games at a resolution of up to 1080p at frame rates as high as 60 fps. (The company told Polygon that 4K support is “coming soon.”)

It’s unclear when Luna will be widely available; Amazon is currently offering invitation-only early access. However, the company aims to undercut its competition with an “introductory price” of $5.99 a month during the early access period, which gives gamers a library of more than 100 games, such as Resident Evil 7, Control, A Plague Tale: Innocence, The Surge 2, and Brothers: A Tale of Two Sons. While Luna resembles Stadia’s approach in some ways, Amazon is pricing content differently, bucketing different catalogs of games into separately priced “game channels.” The Luna+ channel costs $5.99 a month, while details are scarce on the Ubisoft channel, which promises access to ultimate editions with downloadable add-ons for some titles.

For all these services, the true test will come in some shape or form when next-generation consoles arrive in November. The leap in graphical fidelity of next-gen games, and the desire to have PlayStation 5 and Xbox Series X titles eventually streamed on Sony’s and Microsoft’s services will once again push the limits of what’s possible in cloud gaming. Likewise, cloud-based services will have to introduce exclusives that prove a pivotal draw. But whether those streamed experiences will mean that more players embrace the cloud is anyone’s guess.

“That’s really the part you have to do your own gut check on, is if the experience is as good as local, and it costs way less,” offered Gaikai’s Perry. “Will that make cloud gaming OK? Because it’s coming. It’s definitely coming.”