Cloud Gaming has everything for the technology analyst. There are big networking requirements (all the equipment vendors), it pushes the envelope on device performance (which brings the chip companies into the mix), it is the definition of cloud computing (all the Internet giants), and it all revolves around video games (which makes it a lot more fun than RF semis).
What it does not have is a clear audience or business model.
For most of their history video games have required users to have fairly powerful ‘computers’ in their homes (or place of work). Whether a console, a PC, or smartphone, some local device needed to have a sufficently powerful chip inside to do all the calculations behind all those beautiful graphics. This was as true for Pong in the 1980’s as it is for the Last of Us Part II today.
With every other form of software moving to the ‘Cloud’, there is a logic behind the idea of moving game calculations to remote computing clusters as well. Under this model, the heavy math is done on really powerful computers (aka data center servers), which then sends the output to remote devices. The end device has to do much less work, merely displaying the graphics file it receives from the Cloud.
This means, in theory, that a user with ‘any’ device can play ‘any’ game. Modern tech companies are now all built around the idea of pulling data into the cloud, because once there, it lets them add on all kinds of new features and services. With cloud gaming, the central platform can provide identity management, social tools, ads, stores, advanced networking tools – the list goes on.
This also plays into gaming trends. Popular games today all have some element of online services, especially networked gameplay. A cloud gaming platform allows for enhancements to that online gamine experience. An important part of gaming today is streaming and esports, watching other people play games. And of course, these are all cloud services. The game companies are increasingly experimenting with ways to blur that distinction between viewing and playing – for instance letting esports viewers view any point in a live, competitive game or letting stream viewers jump into the action of a multi-player game with the click of a button.
All that being said, there are some problems with this vision of cloud gaming nirvana.
The first problem is that trade-off between on-device and cloud-based compute. The unspoken secret of modern software is that there are actually a lot of situations in which running software “on-prem” in your own servers is better than running it in the cloud. True, this is a minority of uses, but there are many situations in which having local compute matters.
This is especially true in gaming. As much as most people think of gaming as a ‘toy’, game development is actually one of the most demanding specialties in programming. The whole point of gaming is to push the technical envelope of the code to the limits, to make the games look and feel as good as possible. A game written for a specific hardware configuration is always going to look better on that special-built hardware than it will running on a generic server. For years, the game console makers, were engaged in a silicon arms race to have the most powerful gaming engines inside their hardware.
Arguably, we have gotten to the point where everything is just going to run on Nvidia and AMD GPUs, and whether they are in the cloud or in the living room should not matter. But there remains a risk that someone develops another purpose-built piece of gaming hardware that outperforms the cloud service. Look no further than Nintendo which has consistently defied the common wisdom of every gaming cycle as a living example of this.
Ancillary to this issue is the fact that many of the additional ‘network services’ which cloud gaming enables can probably be done regardless of where the actual compute for the game takes place. We already have identity management, load balancing, ads and all the rest today, and they work pretty well. Maybe some of the advanced viewing/streaming options can be done better in the cloud, but we still have a few years before we know for certain, and there may be perfectly reasonable alternatives.
A second problem is the ecosystem. As we noted at the top, for us, the interesting part of cloud gaming is how it draws in so many pieces of the Technology Industry. However, viewed from the perspective of the Business Development manager that sprawl is the recipe for a huge mess of partnerships, exclusivity bets, and branding wars. We will come back to this topic in a follow-up post next week when we walk through the ecosystem.
That being said, the biggest problem with Cloud Gaming is that neither the business case nor the target audience for it are entirely clear. Who really needs this? The idea of playing any game on any device sounds great, just like Java used to bill itself as write once run anywhere, but when you get down to the details the actual appeal is less apparent.
How many people want to play the latest Call of Duty on their smartphone? The mobile player, without a keyboard and mouse, will be at a permanent disadvantage to the PC player. Flipped around, no serious CoD player will be happy if the game strips its user interface down to level the playing field with mobile. Put simply, Candy Crush does not need cloud gaming, and Call of Duty does not want it.
Our impression of many of the early cloud gaming platforms is that a few big companies let teams of engineers loose without talking to any of the business people ahead of time. Some of them are clearly a solution looking for a problem to solve.
We can think of a situations in which Cloud Gaming might make sense. As we have noted elsewhere, Tencent in China, has built an interesting cloud gaming platform. Their target are players with bare bones phones. These phones can barely handle entry-level graphics, so Tencent has given them a way to play much more advanced mobile games. This is an interesting idea, but it remains to see if Moore’s Law moves fast enough to dampen the appeal.
We recognize that this all sounds rather negative. Our intention here is not to condemn Cloud Gaming before it starts. Instead, we want to challenge the assumptions around the idea. We think there is a case to be made for the idea, but everyone – all those participants above – need to think through the fundamentals if they want to see Cloud Gaming succeed.
(Note: We have been working on a lengthy report on Cloud Gaming for many months. We hinted at it in our look at Edge Computing a few months back, and our original plan was to publish a detailed note in April. The world has changed a bit since then, and that note hasn’t happened. We continue to work on that large piece, but we wanted to get the topic started here with a quick overview. Sometimes you just have to start writing.)