Gaikaiâ??s reality check

Gaikaiâ??s reality check

By Rob Crossley

June 11th 2010 at 12:30PM

Co-co founder Andrew Gault on how the service is shaping up ahead of its E3 push

Gaikai is heading to E3 with its most important year ahead of it.

The cloud-gaming company has continued to impress a huge chunk of the industry with closed-door demonstrations and explanatory videos.

But this year the service will be out in the wild. It will appear on websites you visit and deeply discussed in board meetings.

As the Gaikai team gets ready to set up its own booth at E3, Develop speaks to co-founder Andrew Gault about how the service is shaping up, and its prospects for the future.

You’ve just taken on a second round of funding, another $10 million. Does Gaikai hope to trade equity in a third?
At the moment no. We’re pretty happy with what we’ve got. A lot of people are funding us, and that’s through a whole lot more interviews and meetings we’ve done. With the money we’ve got we can pursue a launch while putting ourselves in a pretty stable situation. The expectation for another round of funding is not some time soon, certainly not this year.

Gaikai works with Flash, Java and Silverlight, but David Perry recently showed World of Warcraft displaying on the iPad. How is that going to work? What devices is Gaikai optimised for?
There’s two approaches here. One for the broader internet audience at large, where people will tend to use Gaikai on their PC. They’ll have Flash or Java; they’ll have something we can use.

And for the future we’re looking at HTML 5, which would mean just having a web browser should be enough.

We’re encouraged by HTML5, and we’re looking into if it can work, and how well it can work. There’s one guy in our office tearing his hair out right now trying to get Gaikai running at 30fps on the iPad. Our philosophy is to get games everywhere, we don’t want to turn anyone away because they use different platforms and different technologies.

On the iPad, we plan to have our own app, and it would allow optimized performance – something which is crucial for these low-level devices as we have to strain every last bit out of it to get the best visual decoding.

I would imagine that, if Gaikai’s main business continues to be demos by the time the iPad app is released, then the app itself will be free. But if Gaikai gets to the stage where it becomes a premium game subscription service, in a year or two, then plans will change.

In terms of compatibility, Gaikai already works with Slverlight, Flash and Java, but all these platforms have specific video codecs that you have to support. Does that present problems?
Well you’re right, we can’t use our own custom video codec, but not only do these have their own video codecs – they all have their own demands for optiisation. But we’re very pleased with how well Gaikai works.

Are there any perceivable differences to the user if Gaikai is running, for example, on Chrome on a Mac and Firefox on a PC?
No. We support every browser out there. There’s a lot of work going on in the background o ensure there’s no discernable difference in experience. It was never easy, but we’re proud with what we have now.

And even though the number of people using Linux is really small, we want to support that too. We want to support everyone.

In terms of hardware constitution, what does a Gaikai server look like?
Some of the specs I can’t tell you because they’re a trade secret in themselves. Getting a hi-def stream of a game that’s CPU-intensive is difficult, but certainly not impossible. Within months we got it working.

That’s not been the real challenge. Cloud streaming is relatively easy to do, but it has been expensive. So we have been searching for ways to keep costs down whilst keeping the streaming technology optimized. That’s why our solution is a trade secret.

We went through a huge amount of hardware combinations to find out what worked best. We’re using the very latest Intel six-core processors, and the most expensive graphics cards you can buy. For the current demo servers we’re setting up, we’re using GTX 470s. But we actually use less RAM than you would imagine, because of the way we virtualise the system, we don’t need more than one or two gigabytes of RAM per game.

And the cooling system is crazy. The amount of air coming out of the rig could blow-dry someone’s hair very quickly.

So how many players do you think can play a game from a single server?
It depends on the game, but around a dozen per server. But like I say, it’s all about costs. It’s a spreadsheet exercise. If we can come up with a different server system, that’s a lot cheaper and can serve four people but the costs per player are less, we’d be interested.

Though it’s fun to build insanely powerful servers, we have to be cost-effective.

Is overclocking an option?
We could if we tried, but it’s not something we use at the moment.

At what point down the line is Gaikai hoping to stream full games as well as demos? Do you have a timeline on a subscription plan?
Well we’d like to do it tomorrow but it’s all about picking you battles. I think we’d like to start it in a year or so, all the company’s founders would like to target that. But at the moment our demo service makes the most sense, and of course we want to move to the next stage.

But we’re open to what that next stage can be. Do we go for an arcade-style service where people buy virtual coins, do we go for subscriptions, do we focus on MMOs or action games?

And of course your demo business might be so profitable that it’d be hard to turn focus away from that.
Of course. And there’s a lot of adverts we can serve as compared to full game streaming. So we’re open.

Our policy is that the Gaikai demo will be the full game that has a cut-off point. For the developer, coming up with specific demo builds can be a real hassle – it’s usually the last few weeks a studio hits crunch and the publisher comes along and says ‘oh, buy the way guys, we need a demo!’

That’s putting a strain on developers, and we want to change that. So with Gaikai, the full game is offered with a cut-off point. That will save a lot of hassle.