The Evil Economics of Pixel Streaming

[This article is the 2nd part of a 3-article series "Is Pixel Streaming headed for another epic fail?". Part I can be accessed here. Part III here]

Often when talking pixel streaming solutions like Google’s Stadia, the majority of questions asked by journalists and industry participants still remains focused on latency and image quality.

While that those are indeed some of the leading challenges for pixel streaming solutions attempting to compete with established platforms, another challenge, namely the economics behind such a service, are often forgotten or treated as easily solvable by the likes of Google.

Nevertheless, they do remain an important topic (if not even more important than latency) when dealing with this kind of technology and something that even Google is probably spending a lot of its own resources to address.

One could say that Google has deep enough pockets not to care about losing money for a few years, but remember that at some point someone will eventually question whether or not a certain line of business is actually worth it. And for Google, that wouldn’t be a first, right? right?

So what are the economics aspects we need to focus on for Google Stadia and other Pixel Streaming services? There are actually 3 angles here:

  • The first is the economics of the content, something we will save to address in the next article.
  • Then there are the economics of the high-end gaming servers that Google needs to set up for Stadia (trust me on this one, those are definitely not cheap… even for Google).
  • Finally you have to consider the economics for the user at this kind of service, after all, someone has to foot the bill eventually.
Google’s original home built server. Ingenuity was always their core strength. :)

While Google is known for their server building abilities, high-end gaming hardware is an entire different species on its own, with the GPU representing the majority of the investments as on any gaming hardware.

For that reason Google has partnered with AMD to build a custom gaming GPUs to power its Stadia servers.

Naturally one would expect Google to get quite a good cost for such a large order for Stadia’s users, right?

Let’s see…

During the Stadia announcement if you use the showcased map as a reference (see picture below), let’s assume all those 58 zones would have Stadia’s servers capable of handling at least 50.000 concurrent users each (let’s be conservative for now) or around 2.9M total concurrent users (just for the sake of comparison, Steam gets on average around 10 to 15M concurrent users every day [source]). For Google that would mean 2.9M custom AMD GPUs, a considerable amount for any OEM (Nvidia must be drooling over this).

Google’s infrastructure map during the Stadia unveil

However, guess who else is also a huge AMD customer for gaming GPUs? That’s right, our old gaming pals at Microsoft and Sony. Both of whom have been using custom AMD GPUs for several years now (namely the last three generations of the Xbox and the latest generation of the Playstation).

While Google may be ready to order a couple million of AMD GPUs, Microsoft and Sony are on a whole different level here, having a combined historical console base of over 300 million units, of which their latest generation sold over 100m units, and all of which have a custom AMD GPU not unlike the one Google is ordering.

I doubt therefore that anyone would consider Google any more specialized in high-end gaming hardware than Microsoft or Sony. So if you consider that Microsoft already publicly admitted that an Xbox One costs them upwards of US$400 to build and are in fact sold at a loss [source], it wouldn’t be far fetched to believe Google will be spending at least as much on their side.

Also consider that Google is (maybe smartly, as we are about to get a new console generation from both players above) looking to go beyond here, by promising 4K (and even 8K) game streaming at 60fps not so far into the future and therefore already employing GPUs that are at least 60% more powerful than current consoles. Even at $400 a piece per concurrent user, we are talking at least US$1.1Bi in hardware investments alone… and this is where things start to get expensive… even for Google.

Considerably faster than latest generation consoles… that has got to cost something, right?

Finally consider that those investments are solely for the hardware side of the service. Don't forget that operating expenses are also considerable part of costs for this kind of service. It's all nice and dandy to have 11 teraflop GPU, but those things drain power… a lots of it. Add to that the physical space costs (non-existent on users side), maintenance & engineering teams, and the constant bandwidth costs in streaming those 4k signals to users every time one of them decides to play.

If even Google has already mastered managing those costs with their cloud infrastructure, in the end what matters is the cost/min per user of the service. At an estimated $0.51/hour per user (or $10/month) just to maintain the service (take a look at Bruce Grove's, former Onlive partner, keynote on this here ), it's not so cheap now, is it?

See where I’m getting here?

We shouldn’t be surprised at these numbers, after all the whole premise of games streaming from a server is to abandon the need for hardware on the users side…

Unfortunately we are talking hardware that remains quite expensive, needs to be always close to end users (connecting to other regions would likely result in high latency) and needs to be updated every few years to cope with new games ever increasing graphical demands.

On the upside though, Pixel Streaming solutions also imply that its hardware may now be shared amongst different users. That means that those millions of GPUs and operating expenses could now be shared with more users (in the same server region), that is assuming not all users play together or at the same time, something that is not usually the case with gamers.

At peak hours over 15M users play simultaneously on Steam.

Another possibility is if Google was sucessfully able to implement some of CiiNow’s patents, namely one that would allow multiple instances of games to share the same individual hardware. While this would mean a considerable cost reduction per concurrent user and make Pixel Streaming economics far more manageable, its not exactly clear if that's realistic with the high end games running at 4K as Google is advertising.

Who foots the bill?

How about the users side of this equation then?

I will assume for now that the service is charged as a subscription at say, $15/month (for comparison sakes PlayStation Now costs $19.99/month in the US), that would indeed mean potentially that those investments would be paid back within less than two years. Something quite bearable by the likes of Google. 👍

Well, not to get into the ugly economics of hardcore gaming subscriptions (I promise a future article on this one), consider for now that users will be willing to pay for a monthly subscription for Stadia.

Another monthly service…

This will likely mean that users will be spending around $15/month subscription + the optional Stadia controller, which will probably retail in the $40 to $60 range (similar to an Xbox One or Playstation controller price). Cost of games will likely be on top of that (more on this later).

That amounts to $220 on the first year, which is just slightly cheaper than what you could spend for an Xbox One S during the last Black Friday and about $80 cheaper than an Xbox One X (not counting the $60 / year for the Xbox Live Gold subscription).

If you consider a console lifetime of about 5 years, that means users will be spending over $940 on Stadia versus $400 for a latest generation console (or $700 if you add $60/year of XboxLive/PS+ subscriptions). Even a good gaming PC can be bought nowadays for cheaper than that.

Lifetime Ownership Cost *

* Let’s assume that Stadia’s users will opt to buy at least one of Google’s Stadia controllers here, but won't have the need for a Chromecast or other optional devices for now.

PS: I know that PC Gamers are going to bash me for not considering the savings of cheaper PC games, but bear with me for now. ;)

What you notice is that Stadia is indeed a very good choice for non-gamers or for people who want to avoid the upfront costs of gaming. Even after the first couple of years, costs are actually quite similar, which could make Stadia’s conveniences a tie-breaker here.

However as we get to year 3 and further, Stadia starts to become more expensive for users, especially if you consider that you will always need outside hardware to run it, even if you do it on your TV or your old laptop.

The upside though is clear. Ubiquitous gaming that is not dependent on local hardware, requires no maintenance, upgrading or configuring and allows the ability to play your library from almost anywhere and any device. That is, of course, assuming you have that good and stable connection where you are.

In the end it all boils down to what kind of gamers are going to see value in it, which is not exactly clear at this point if you look at the main types of gamers.

Just some kinds of gamers out there…

For the sake of simplicity, lets consider the two main camps of gamers around: high-end gamers (aka hardcore gamers) and casual gamers.

First consider that high-end gamers, those who already own a gaming PC or one of the latest generation consoles, don't seem very likely to switch. Those players already have the sunk costs for a hardware that could easily last another 5 or more years (I own a 3 year old PC that still runs most games at max settings).

Also important to bear in mind that for this kind gamer, the risk of the image quality loss and potential latency issues that could ruin the experience, may not be acceptable. For them, Stadia would at most be an optional add-on that allows them to play while roaming, assuming you would have the connection for it wherever you are.

How about more casual gamers who either lack the time and/or hardware to play?

For them, a simpler service like Stadia could indeed make more sense by allowing on and off gaming at a low upfront cost.

However, we need to consider two issues here:
1 - Gamers who lack time to play might not see much value in committing to a monthly subscription just to play a few games in their spare time, especially blockbuster titles whose learning curve could demand hours of gaming.
2 - Casual gamers, those who usually play short sessions on their mobile devices, are used to playing mostly for free, which could be an issue if Stadia's model demands a monthly subscription fee + the cost of each game.

So in the end, who is Stadia for? Seems like casual gamers or wannabee-gamers who want to try higher-end games, but don't want the upfront costs or the hassle from traditional gaming hardware look like a more ideal target.

If Google is able to provide a good enough experience while also adding great content and interesting features, there might be a future for the service, just not as massive as one might think.

Next we focus on the content side of this analysis. Will Google and other streaming players really be able to attract high quality content while changing the industry business model?

Stay tuned… Part III is here.

Tech entrepreneur CEO of Nuuvem.com and Devise,vc, angel investor and board member at several web based companies. Also a crypto-enthusiast and an avid gamer!

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store