Some revolutions take time. Four years after OnLive and Gaikai intensified a discussion that had already been going for at least half a decade, the idea that the cloud will be a vital aspect of the industry has started to lose momentum. Services like Netflix and Spotify prove what a disruptive and consumer-friendly force the cloud can be when applied to entertainment content, but its place in gaming still feels very much in the future, always a few years away, just beyond the boundaries of our bandwidth.
If the line-up of Cloud Gaming Europe's 2013 conference is a valid reference point, the contemporary landscape of cloud gaming is defined by vaguely familiar names: Ubitus, CiiNow, Transgaming, T5 Labs, G-Cluster. Most of these companies pre-date the current generation of consoles. A few have been chipping away at cloud gaming for more than a decade. Almost none of them are what you might call a household name. With one clear exception.
Thanks to a noisy entrance, a huge marketing spend and a spectacularly abrupt wake-up call, OnLive has shifted from cloud gaming's poster-child to its chief cautionary tale in little more than a year. Very few sessions on the conference schedule manage to conclude without at least one mention of its name, and precious few of those mentions constitute praise, or even constructive criticism. There is a sense that, rather than bring the idea of cloud gaming to the masses, the eventual collapse and subsequent resurrection of OnLive actually had the opposite effect, destroying its own progress.
"People get caught up in, 'this doesn't work today, for me,' rather than thinking about how these technologies grow"
Bruce Grove, OnLive
OnLive's Bruce Grove disagrees. Granted, the company's general manager for the UK is unlikely to encourage a dismissive attitude among his peers, but Grove believes that they, like the press, exist in a bubble. For a few weeks, the implosion at OnLive was one of the biggest talking points in the industry, the source from which dozens of news stories and innumerable tweets flowed. But it's not the sort of news that passes the high walls around the industry's perimeter. OnLive didn't go offline for a single day. For its admittedly small group of customers, the lights were always on, and have remained on to this very day.
"We were the first to market, and I think we're still the only company offering AAA titles direct to consumers. People like to treat us as some kind of whipping boy, but really, we have achieved a huge amount," says Grove, with the practiced ease of a man often called to defend the sins of the past. "OnLive never dropped its service. We worked really hard behind the scenes to keep things running. We'd spent years trying to educate people about why the cloud was so good for gaming, and the last thing we wanted to do was destroy that perception."
To be frank, the perception of OnLive within the games industry has been a lot better, and the fact that Grove is unwilling to discuss any user figures suggests that the service's "core loyal following" hasn't grown by a significant amount. But if Grove's opinions accurately represent those of his employer, the new OnLive is an altogether more pragmatic and patient company than the one that slipped into the abyss under the auspices of Steve Perlman. The cloud, Grove says, is creeping into mass culture more and more every week, and OnLive is ready to play the long game.
"The key for [our] following is convenience - not worrying about updates, and being able to just switch on and start playing. But who's the Spotify audience? Who's the Netflix audience? Or BBC iPlayer?
"Around the same time that cloud gaming started to make some noise in the U.S., Netflix launched its streaming service. Now, here was a very successful DVD rental service that, to a small handful of people, offered SD video and a back-catalogue of a few thousand titles, most of which weren't really that interesting. Everybody said it was stupid, that it would never go anywhere. But two years later, Netflix has a huge streaming catalogue, it's high definition, it's 5.1 surround sound, and it has a much, much larger audience. The same thing happened with the iPlayer.
"People get caught up in, 'this doesn't work today, for me,' rather than thinking about how these technologies grow."
Speaking to the attendees of the Cloud Gaming Europe conference, it's clear that OnLive's initial decision to target AAA gamers is regarded as a decisive factor in its downfall. Superficially, it made sense: streaming AAA games from remote servers is a cutting-edge concept, so why not target the part of the audience that defines itself by proximity to that cutting-edge? But by targeting core gamers, OnLive was trying to convert the audience with the most searching questions and the highest standards to a technology powered by values like accessibility and flexibility. OnLive fell short in the disciplines they most valued, and excelled in areas where they were happy to accept a little compromise. It may be counter-intuitive, but the logical first stop for cloud gaming should have been the people that scarcely know about technology at all.
"The cloud gaming industry needs a horizontal architecture. Without that this industry will not reach 100 million users"
Peter Relan, Agawi
"The family audience wants a casual, social kind of gaming, and that sort of gaming is not constrained like AAA console," says Peter Relan, co-founder of the cloud-gaming start-up, Agawi. "For two years, Agawi has been streaming those games to millions of users [via its Cloud Play service], and I've never seen friction with that. That section is very amenable, but OnLive went after the other extreme: hardcore gamers, who are the most difficult to please. They will compare you to their current experience on console.
"I don't think Netflix would have succeeded if it had focused on people who love going to see movies at the cinema. They were never going to get those people. It was an accessibility and convenience play for those who say, 'I'd rather give up the HD experience to watch my movies anywhere at any time."
Winning over the AAA gamers involved a "bits and bytes" conversation that didn't flatter OnLive's service, and probably wouldn't have been necessary with a more casual audience. Relan is convinced of one thing: the market doesn't want technical specs and explanations; it wants products. If something works, it can only be harmful to spend time explaining the processes involved. To use Relan's analogy, neither the person making the toaster nor the person buying the toaster spend any time thinking about electricity.
"They're not necessarily even capable of that, and why should they be? In cloud gaming, the consumer, and even the content provider, must be isolated from those details. That's Agawi's job."
When it comes to online technology, Peter Relan has seen more than most. He introduces himself to the conference crowd as a "networking guy," but that's too loose a term. Relan has been based in Silicon valley for the more than 30 years. He was an ARPANET researcher back in the Seventies, he co-developed one of the earliest iterations of video-on-demand while at Oracle in 1993, and he helped found the incubator that spun out OpenFeint and CrowdStar. From PC to online to mobile to social, one thing held true through every cycle of boom and bust.
"It takes two or three tries to bring the knowledge and the wisdom to do it right," he says. "If OnLive had got it right, I'd have been shocked."
OnLive fundamentally misunderstood the potential of the cloud for gaming, putting the most important leap in accessibility since the foundation of the world wide web at the service of a niche market with a rarefied taste in products. In Relan's view, it was hugely ambitious in the wrong direction, targeting a market of tens of millions rather than the hundreds of millions captured by successful cloud platforms like Netflix, Hulu and Spotify. "Who's happy with a 10 million user market?" Relan marvels. "That's not even the size of a small European country. We have global markets here, and global scale."
The contradiction is that OnLive tried to build a vertical structure for cloud gaming. It had to do everything, from assembling and maintaining the data centres, to signing content deals, to providing a good service to the consumer.
"It doesn't scale," says Relan. "$100 million isn't enough if you're going to do the vertical business. The vertical businesses of today are the iPhones. It takes Apple, it takes Google, maybe Microsoft, to build those businesses. They cost billions of dollars. It's just fundamentally flawed.
"Past attempts at building cloud gaming haven't worked because the capital expenditure required to build this model isn't feasible. Everything is proprietary here. It's expensive. It's highly local, because it's a data centre intensive business and latency is a real issue. To get global coverage, you need data centres all over the world."
In his talk at Cloud Gaming Europe, Relan made the persuasive argument that, for cloud gaming to become a mass market proposition, it needs the sort of horizontal architecture that comprises the compute cloud - where companies like Amazon and Rackspace sell and rent the servers that support a huge number of web and mobile apps. The compute cloud is open, it's inexpensive, it's globally available, and it's been accessible to developers of all budgets for almost five years.
"Cloud very specifically means elasticity, not just being in the internet. Don't confuse the cloud with just having a bunch of servers in a data centre somewhere," says Relan. "There was no cloud in cloud gaming. That's the reality. That's why we've struggled. We've been operating for two years on the thesis that the cloud gaming industry needs a horizontal architecture. Without that this industry will not reach 100 million users."
Relan believes that we have arrived at the "inflection point" for cloud gaming. The next two years will herald the emergence of a similar platform to the compute cloud, but based on high-performance GPUs. And, as is often the case with impassioned executives, he has a dog in the fight: Agawi True Cloud, which will offer content creators a choice of server architectures built on Nvidia's Kepler GPUs. With a minimum performance target of 720p resolution at 30 fps with less than 100 milliseconds latency, Agawi True Cloud will cost 39 cents per user hour, per stream. With the release of Nvidia's Maxwell processors in 2014, the price should drop below 20 cents.
"We strongly believe it's about new and different experiences: How to use this technology to go beyond the kind of experiences you can have on existing platforms"
Julien Merceron, Square Enix
"This year and next year are Cloud Gaming 2.0," he says. "I think that in five years we won't even think about it any more. It'll be like Spotify or Netflix. It'll just be the way it is."
Agawi and similar companies may prove to be successful, but what that means for games is less clear. To date, cloud gaming has essentially meant streaming games-on-demand, but that really sits somewhere in the middle of a much broader spectrum of possibilities: background downloads, active and passive spectating, instantaneous cross-platform play, a new lease of life for the back-catalogue of every content creator in the industry, and a degree of elasticity that will ably support a wider variety of business models than ever before - full releases, micro-rentals, free-to-play, subscription services, you name it.
At Cloud Gaming Europe, Naked Media's Steve Cottam painted the ability to play Borderlands or Homefront at the touch of a button - OnLive's key selling point - as almost quaint. The extreme of the cloud's potential could herald the creation of games on a scale difficult to conceive: terabytes in size, with AAA production values, supporting tens of millions of concurrent users. But that won't happen tomorrow, or even in the next five years. Indeed, by the reckoning of industry consultant Nick Parker, the widespread streaming of full AAA games is unlikely to be a reality until around 2020. For now, content creators are looking at the implementing the cloud in more subtle and pervasive ways.
"We strongly believe it's about new and different experiences: How to use this technology to go beyond the kind of experiences you can have on existing platforms," says Julien Merceron, worldwide technology director for Square Enix, which operates a cloud-based gaming service called Core Online. "For us, the huge advantage on the content side is to make our products more dynamic. One of the problems with dynamic problems right now is that users have to download updates and patches and wait for things to happen instead of playing. In the cloud, you can have very, very dynamic and secure products without making the user frustrated.
"On the technology side, we're trying to look into what is actually feasible: Where to start, and where we want to go. We have a lot of ideas, but we are almost redesigning products for the cloud, rather than just adapting or porting them."
In that sense, the feeling among the massed developers at the Cloud Gaming Europe conference is that Sony has judged its implementation of Gaikai's technology just right. The PlayStation 4 employs a little of the cloud in a lot of different ways, from instant demos to ambient updating to sharing video - the sort of features that will fundamentally improve the console experience without the need for long explanations about reducing latency. Sony gave no concrete details of its plans to stream full games, and that's telling. Even David Perry's vague ambition to stream the entire PlayStation 3 back catalogue seems more fanciful than realistic. Indeed, in a conversation between sessions, Graham Clemie, the founder of cloud gaming pioneer T5 Labs, told me that the idea sounds a little like "bullshit" - thanks to the PlayStation 3's unique cell architecture, that sort of backwards compatibility would be a difficult and expensive undertaking that the won't be feasible until Sony's customers are past caring.
"Most of us don't know what's on offer. We don't hear much about cloud gaming. The Xbox guys don't talk to us about cloud gaming - ever"
Simon Humphreys, Codemasters
Does Microsoft have more ambitious plans for the next Xbox? It's far too early to tell, but the smart money is on it being at least as cloud-facing as the PlayStation 4. Microsoft has the sort of resources and scale that Agawi's Peter Relan believes would be necessary to create a successful vertical cloud gaming business, and the truth is it has been more ready for the cloud than Sony for many years. The notion that none of that learning will trickle down to the next Xbox is borderline absurd.
However, in a roundtable hosted by Microsoft at Cloud Gaming Europe, Simon Humphreys, director of digital content at Codemasters, indicated that the company is not making its best effort with developers. "I think you need to work more closely with developers and big publishers, and incentive them to use the cloud features you're providing," he said. "Because most of us don't know what's on offer. I'm not talking financial incentives, but engineering time, and assistance in getting some of the features that would work better on the cloud, on the cloud.
"We don't hear much about cloud gaming. The Xbox guys don't talk to us about cloud gaming - ever."
When it comes to companies the size of Microsoft and, to a lesser extent, Sony, these small inefficiencies and arbitrary barriers are inevitable, but when it comes to the widespread implementation of the cloud in gaming they could prove decisive. Once the infrastructure is in place, the GPU cloud will reach billions of devices with games of every kind - greater reach and more flexibility at a significantly lower cost. The PlayStation 4 and the next Xbox will no doubt be designed with a degree of flexibility in mind, but if Agawi and its peers can establish an affordable GPU cloud in the span of time Relan suggests, the console manufacturers may be in yet another race they just can't win.
"The console is faced with the classic innovators dilemma. You have to take care of the thing that built your business in the first place, but by doing that you're limiting yourself," says Relan. "They might sell, what, 30 or 40 million of these things in a year? But more tablets or connected TVs are going to sell in one quarter. After these next Sony and Microsoft consoles I don't think there'll be any more. Cloud gaming is still incipient, but not for much longer.
"They have one more shot, but five years from now that'll be gone."