It all started at E3 2018. After EA’s acquisition of Gamefly’s technology, Ubisoft and Xbox announced intentions to jump into cloud gaming. That got the gaming industry abuzz again with the promise of games being delivered to consumers from cloud machines similar to how you watch Netflix movies on your TV, smartphone, and computer. Nvidia and Google released cloud gaming beta products while Microsoft, EA, and Ubisoft shared their intent to do the same. With Project Stream, Google even demonstrated that their technology could work within a browser (we open sourced a demo of the techniques Google is probably using for this in a previous post, so you can see how it’s done). While marketers tantalize consumers with the idea that the Netflix of gaming is arriving soon, many significant differences have always existed between delivering video games and movies via streaming. These technical and economic differences are why consumers in the United States have had Netflix streaming since 2007, but have only been repeatedly disappointed by cloud gaming services.
Some of the constraints that felled early innovators, like OnLive, have been worked out, but considerable challenges still impede the mass adoption of cloud gaming. Cloud gaming as a distribution technology is exciting, but we won’t see mass adoption until game developers embrace the cloud and build cloud-native experiences for consumers. At that point, we’ll have unique experiences that aren’t possible on your local PC, and every gamer will want to try the new medium. When that happens, a new wave of creativity and innovation in content will result in exciting new experiences driving the $160 billion video game industry into the future.
The challenges facing cloud gaming
Whenever we talk about Parsec and the potential of cloud gaming, people always bring up OnLive. OnLive was an innovative cloud gaming product launched at the beginning of the decade. The founders and team at OnLive had an ambitious goal to replace consoles with streaming from cloud machines.
The economics and infrastructure required for this vision are challenging now, but they were next to impossible then. At the time it shut down, OnLive was reportedly burning $5 million per month on their servers with 100,000 monthly subscribers. Despite reaching a significant number of customers, they weren’t scaling fast enough due to issues that continue to impact the business models of cloud gaming companies today.
Startups have been pitching “the Netflix of gaming” to investors since Netflix streaming launched in 2007. It’s a sexy idea — eliminate the need for expensive hardware while delivering the most valuable entertainment medium in the United States directly to consumers via the internet. Increasing access to gaming through disintermediation of hardware — that sentence gets any venture capitalist excited — is a huge opportunity. After all, movies and video games are pretty similar and look how big Netflix and the OTT video business is overall. Although we provide game streaming technology and are really excited about the prospects of cloud gaming, we do not believe that the gaming PC days are numbered. In fact, we’re very bullish on gaming hardware in your home and believe strongly in people using their gaming PCs with game streaming technology for unique gaming experiences.
The truth is that there is one significant technical difference between streaming video and streaming games — games are interactive files, and movies are static. This difference underlies all of the technical and economic challenges in the market. You need to render a game on-the-fly, but you only need to render a final cut of a movie once. Rendering like that is computationally intense and requires expensive hardware. Latency is also a challenge based on distance from the rendered game, but this challenge is solvable for most games.
Cloud gaming requires expensive hardware
GPUs are expensive, which is one thing OnLive was running up against, and also what every cloud gaming company faces today. Not only are server-class GPUs expensive, but for now, virtualizing them doesn’t lead to significant benefits for a cloud gaming company. Theoretically, the best you can do is have a handful of customers on each GPU, but in practice, you don’t get much for this multi-tenancy approach. The more customers you have on each GPU, the less power each has access to for rendering the game graphics. If you’re going to just have low quality graphics delivered from the cloud, the consumer might as well use their low powered device at home to play the game.
Additionally, financing the leasing of these GPUs, co-locating them all over the world (because proximity matters), supplying redundant power to them, and paying for the outbound bandwidth (you’re using more than 2.5 GB per hour with 60 FPS 1080p streaming) is very expensive. It’s so expensive that the economics of just supplying the hardware in the cloud that gamers will use for playing their games eats up a significant portion of any potential revenue from cloud gaming.
According to the U.S. Census Bureau, an American who plays video games spends 2.4 hours of their average day doing so. For those who play video games, if you assume an all-in hosting cost of $0.15 per hour for computing (this is conservatively low for many companies), you’re going to need to charge at least $11 per month to cover hosting costs. Adding content licensing fees on top of that to compensate the lost revenue in game sales (assuming the games aren’t free-to-play) will result in a much higher monthly rate. Morgan Stanley found that consumers are willing to pay $14.75 per month for a cloud gaming service. At this point, that rate will not cover the hosting costs + content licensing costs. Since technology companies have a high cost of capital, the cloud gaming companies will have to charge a healthy margin on top of that to overcome their capital hurdle rate.
The counter argument to that is that cloud gaming will expand the market and reach customers who consume fewer hours of games and will pay a premium to not buy hardware. Let’s say you reach these new gamers with the technology, and they play just 30 minutes each day. If your costs are still $0.15/hour, you’d be able to serve games to them for just $2.25. That’s a really good margin if game licensing isn’t included. Adding $5 of licensing fees per customer per month still leaves the cloud gaming company with a 50% gross margin serving these new gamers — that’s better than Netflix. It’s unlikely, however, that a cloud gaming company could license a compelling library of content for just $5 per month per user without massive scale. Publishers will worry that you’re going to steal revenue from their core sales channels.
Convincing AAA publishers to license content is a challenge
To be the full “Netflix of gaming” solution, you’ll have to license content and figure out how to bring the hardware costs down on GPUs in the cloud. To make cloud gaming enticing to the largest publishers, they have to believe that providing a license to their game or going direct to consumer will be cheaper than through the standard platforms — Xbox, Playstation, and the Switch. The general rule is that the platform will take a 20–30% fee every time someone buys your game for their system. For each $60 game, that leaves about $42 of gross profit to the publisher before marketing expenses and other costs. If the publishers believe that they can get more gross profit delivering games through the cloud, they’ll try it. If they believe that they’ll capture “latent” gamers through the cloud (i.e. new revenue), they’ll try it. But if neither of these are true, the business of licensing to cloud gaming companies is less exciting.
Who is in a position to win?
In the next couple of years, we will probably see retailers (trying to save their gaming businesses from becoming irrelevant), telecom companies (touting cloud gaming for 5G), publishers (going direct-to-consumer), platforms (playing your console games on any device), and infrastructure companies (renting their GPUs for gaming) enter the market. They all bring unique advantages to the market, but I believe infrastructure scale is going to be a huge advantage. That’s where the infrastructure companies operating datacenters get a leg up in this game. The companies that have massive scale in infrastructure benefit from lower lease rates on their hardware, power costs for operating the GPUs, distribution across the world, and shared workloads. This gives the largest datacenter companies (Amazon, Microsoft, Google, Tencent, and Alibaba) a big advantage to bring costs down for cloud gaming machines.
Many of the companies with the largest datacenter footprints have already announced plans to work on cloud gaming. Microsoft has announced its effort to build a game streaming platform and Google gave players a taste of their tech with Project Stream. Along with these companies, I believe that the other big cloud datacenter companies will announce some form of cloud gaming or game streaming with their inherent infrastructure and scale advantages.
While scale matters, content matters just as much. Microsoft has a huge advantage over the other datacenter businesses (other than Tencent) with their original content, GamePass, brand recognition in gaming, and Xbox Live. These assets will be a formidable combination to thwart attacks from publishers going direct-to-the-consumer and from new entrants in the market, like Google.
Who will play in the cloud?
According to market research firm Newzoo, 2.2 billion people play games worldwide, and Nvidia reports that 1.2 billion of them are playing on a PC. Approximately 140 million current generation consoles (Xbox One, PS4, and Nintendo Switch) have been sold globally. There are roughly 110 million gaming PCs on the market based on dividing the total hardware sales since 2013 by an average cost per unit of $1,500. Assuming there are 2.5 people using each of these devices to play games, you have about 625 million gamers with an HD experience. That leaves 1.6 billion “gap gamers” worldwide who could be the target audience of cloud gaming. Granted, a lot of these people play games at LAN cafes in Asia, some of them play games on their phones, and others suffer through low frame rates on integrated graphics.
It’s likely that cloud gaming would also contribute to the expansion of the gaming industry as a whole. According to Morgan Stanley research, 10% of Americans say they would play video games if the hardware weren’t so expensive. Intent is certainly different than action, but this plus the “gap gamers” across the world are the targets for the cloud gaming companies.
While the goal for cloud gaming companies should be to expand the market and increase access to entertainment, the business will depend on reaching current gamers with an experience they’re willing to pay for. To convert the players who do buy traditional hardware, the experience in the cloud must be distinguishable from what they can do on their local hardware — otherwise, there’s no reason for them to switch.
A cloud-native gaming experience is the key to widespread adoption
As with every new distribution model for media, the breakout moment will likely come when the media starts matching the medium (i.e. someone builds a game uniquely taking advantage of the infrastructure in the cloud). This would likely look like a company building a game that uses cloud hardware to do something truly special for the game. EA has shared some ideas on this recently with Project Atlas, which sounds like a game engine taking advantage of cloud hardware. If this experience is truly amazing and provides something that local hardware cannot provide either by taking advantage of the proximity of the machines running the games, the access to unlimited CPU, or some other valuable aspect of the cloud, that would be a reason for gamers to try cloud gaming.
The issue is that for a gaming company to invest in this, they’d need to expect generous returns on their investment. Until there are millions of gamers on the cloud and a business model that delivers value to the game publisher, there won’t be much invested in these unique experiences.
At Parsec, we’re trying to deliver a unique experience to cloud gamers turning offline multiplayer games into online experiences through our Party System. This social experience allows people to play, watch, and share gaming as if they were sitting on the same couch without requiring cloud hardware and making every gamer the host of a virtual experience for his friends.
What’s the future of cloud gaming?
At Parsec, we are deep believers in the future of cloud gaming. We expect that it will encompass both a peer-to-peer network like ours, and a Netflix model like what Microsoft and Google are probably building. That being said, new distribution models are always exciting for consumers and drive a massive amount of innovation in content. When the television was first released, people were watching black-and-white versions of radio shows. When the internet was first gaining popularity, we were reading static pages ripped from newspapers. Today, we have close to an unlimited amount of content on TV and the New York Times has disrupted their newsroom to satisfy the demands of Facebook- and Google-driven audiences.
The future of gaming is really quite exciting in the cloud. I’m not a game developer or nearly as creative as any of them, but I know game developers will develop new content to reach the “gap gamers” and those who already access games in new and exciting ways. It’s up to the content creators and distribution platforms to develop the content and business models to make it worthwhile to consumers and studios to invest their energy, hopes, and dreams on the promise. 2019 will definitely be an exciting year for the cloud gaming industry. We can’t wait to be part of it!