Following the interview conducted a couple of days ago with James Ramey from Codeweavers, we have learnt that the typical port using Codeweavers’ Crossover/WINE requires about 15 000 USD. With this kind of information, we can actually derive some rough estimations as to how many sales are actually needed to make profits with a Windows to Linux port. The following will only consider the case of WINE ports, since I have no reliable data source for Feral/Aspyr kind of costs - which are probably on a different scale. If you are still with me, let’s play with assumptions for a while.

In this hypothesis, let’s consider that a game developer or company has not thought at all about doing a Linux port from the get go, and only used Windows-friendly technologies. Yeah, that sucks, but that’s very common, for the reasons James Ramey mentioned.

So, those game developers suddenly get requests for a Linux port, and have no idea whether or not it’s going to be worth the cost/time to spend developing that additional version. If they are lazy and/or inexperienced, they may get in touch with Codeweavers to have them do a port using Crossover/WINE, which would take about a month if it is technically feasible (which may or may not be the case depending on your game). The question is, how many units of Windows games should they have sold so far to consider if they can break even with a Linux port ?

It can all be summed up in a pretty simple equation:

WindowsSalesReqforTuxProfit = WinePortCost / (Price * (1-ValveMargin) * TuxShare)

First, a warning. This is, on purpose, extremely simplified - there are additional factors to take in account (for example, additional maintenance costs of the port if you continue developing the game), but if you just want to know whether that is even a good idea to start with, that little calculation is kind of helpful to get a sense of where you are headed.

WinePortCost is one of the variables. The average cost is 15 000 USD as mentioned earlier, but actually it ranges from 5000 USD to 50 000 USD from what James Ramey told us. Let’s assume that indie games are going to cost around 15 000 USD, and AAA titles around 50 000 USD - that may not be 100% correct, but let’s work with these assumptions.

ValveMargin is unknown - as far as I know there is no public data shared by Valve on how much they “take” on your sales, but since most stores out there work with a margin at about 30%, let’s assume that factor will be 0.3.

Price is obviously the pricepoint of the game, in USD. For indie games we can assume something like 15 USD. For AAA games, somewhere like 45 USD is more likely.

TuxShare represents the percentage of the Linux market for gaming. There are wide variations between different categories of games - simulations/strategy games tend to do better, i believe, than action games. But let’s assume that it remains at about 1%, so 0.01.

Let’s consider a few different scenarii:

1. A not too complex indie game, sold at 15 USD on Steam.

Assuming the port is relatively classic and feasible, costing 15 000 USD, we find that the number of units that have to be sold on the Windows version in the first place to break even on Linux is about… 143 000 units (corresponding to 1430 units on Linux to be sold). A cheaper indie game, at 10 dollars instead of 15, would need to sell about 214 000 units on Windows before expecting a potential Linux port to hit profitability.

2. An AAA game, sold at 45 USD on Steam

This is interesting, because one would expect, by intuition, that AAA titles require more sales to be profitable. Even if we consider the port takes about 50 000 USD of costs (the upper range for a WINE port with Crossover), at 45 USD you would recoup your costs and break even at about 158 000 Windows units (i.e. 1600 Linux units). Which is actually quite similar to the former example of an Indie game at 15 USD. And if we assume the port is cheaper than expected, as in, 30 000 USD instead of 50 000 USD, 95 000 units (950 Linux units!) would roughly suffice to break even.

But How big is the Linux Gamers market, really ?

Good question - and I don’t have a clear answer for you. This is vital, because this will tell us if the numbers we are dealing with are realistic or not. We kind of lack hard data on this (and that is also why I am running the Linux Gamers Survey recently on BoilingSteam, to hopefully derive some answers later on). Well, we do have a couple of indicators. We know that there are 125 millions of accounts on Steam, and that about 1% of them are owned by Linux users. Which may give us something like 1.25 million Linux gamers on Steam. By being a little bit conservative, you would probably consider than about half of these accounts are actually buying games on a good frequency, while the other half may be very casual / dormant / inactive.

That would still amount to something like 600 000 active Steam Linux accounts purchasing games once in a while. The figures I have mentioned above expect that an indie game sells at about 1430 units among Linux users, and a AAA games at about 1580 Linux users. Among a base size of 600 000 users, this is NOTHING! This means that as long as you manage to sell something like 2000 games out there to Linux users (or to 0.3% of Steam Linux users!), whether it’s an indie title or a AAA title (using WINE as a porting system, important detail), you would be already generating some profits ! This may explain why James Ramey mentioned that companies are often surprised to find a bigger market than they expected in the first place.

How many Games out there pass the 150 000 units mark on Steam?

Hey, we may not have super accurate hard data, but ArsTechnica has developed their own system (Steam Gauge) to evaluate with some degree of accuracy how many games were sold on Steam in 2014, and derived a lost of the 401 most popular ones, all released within 2014. I am going to assume that they are not too much off the mark. And well, among these 401 games, 207 actually pass the 150 000 units mark.

Based on our calculations above, this means that any of these 207 games, as long as they could be portable using WINE/Crossover, could actually make sense financially on Linux, without much effort from the developers/publishers. That’s more than 50% of the most popular games in that list.

In these 401 games, the median in ownership is 159415 units - and if you look only at the group selling more than 150 000, you will find again that the median among such games is 287000 units. Which means that for about 25% of the games in that list, once could easily expect something like 3000 games sold for Linux Gamers, at the very least. For an indie game at 15 USD, that would mean about 16 000 USD in profits, without being involved in the porting process (since you would ask a company like Codeweavers to do it for you). For an AAA title as per our earlier assumptions, you’d be able to pocket something like 45 ooo USD in profits. For AAA games publishers, this may not be a lot of money to care about in the first place - so while Linux would be profitable, it may not impact their bottomline that much for them to care. For an indie dev, however, an additional 16 000 USD may be a relatively attractive proposition.

And games in the top of list, like Goat Simulator, can enjoy even larger benefits: 1.5 million games sold, with probably about 15000 units for Linux, i.e. represents about 55 000 USD in profits for the Linux side (assuming they had to spend about 50 000 USD internally in time/resources when they did the port by themselves). Of course the Linux Share may be lower than 1% (See GamingonLinux articles on developer’s revenues shares per OS, Linux shares lower than 1% are occuring in several cases) but again we are using 1% as a general rule here.

There are many games at the top, above 200 000 units, that could definitely generate profits with a Linux port (and that’s not even talking about a Mac port!). A few examples:

• Dark Souls II with almost a million of sales on Steam for Windows.
• Divinity Original Sin (we are still waiting): 800 000 units
• Wolfenstein, the New Order: 600 000 units
• Watch Dogs: 420 000 units
• Street Fighter IV : 450 000 units
• Metal Gear Rising: Revengeance: 450 000 units
• Sniper Elite 3: 380 000 units
• Metal Slug 3 : 440 000 units
• Broforce: 230 000 units (the port is already planned actually)
• Assetto Corsa: 214 000 units

You may be wondering what to do with the games at the bottom of the list ? Well, to be fair, while those games did not reach 150 000 units in ownership, a lot of them had only shipped in the middle or later part of 2014, some in early access and therefore not finished, with probably good margins for further sales down the road. Certain of those games actually already have Linux versions, too. Hatoful Boyfriend, The Talos Principle, OlliOlli, Ultimate Generals: Gettysburg, Ziggurat, Dreamfall Chapters, just to name a few - while it still remains a minority in that group. There’s always the “long tail” effect as well on Video Games, where games sell most of their units in the first few months, but the ongoing sales over years are not negligible either and accumulate over time.

The Bigger Picture

You are probably aware that most ports made to run for Linux and basically pretty close to run on Mac as well, and vice-versa. And since the Mac gaming market is several times bigger than Linux, you get a synergistic effect: for not too much more, you can get two different versions on the market and grab better (Linux + Mac) returns, so it makes even more sense. On top of that, one can expect that the Linux share will grow a little once the Steam Machines start shipping. Maybe not much at first, but somewhere like 2% is not unthinkable, and this would further make it easy to reach the number of units to break even.

WINE will soon bring DX11 to the table and make it easier to port more Windows games. Codeweavers has also mentioned that they hope to reduce the time needed for a Windows to Linux port down to three weeks since they keep improving their toolsets, which should be further bringing the costs down and commoditize the porting market. I have no data on eON and their prices, but needless to say, once there are several solutions on the market to port DX11 games to Mac/Linux, there will certainly be more cost competition happening. For games made with engines like CryEngine, Unreal and Unity, porting to Linux should become easier and more painless over time.

So, at some point, we may see the following happening:

• Specific porting to Linux/Mac will become cheap enough not to care about it anymore and get 80%/90% there in performance.
• Most Engines will support native Linux builds by default, reducing the friction to release titles on all platforms.
• SteamOS will provide a clear standard target to build for in the Linux world.

Now, I am wondering what will be the place of companies like Feral and Aspyr in this changing environment. Right now they are the right places to go to when you want to get ports with high performance and good quality - but as CrossOver/WINE/eON improve and make things “good enough” for many ports out there at reduced costs, and multiplatform engines reduce the need for ports in the first place, are they not going to be stuck between two trends? I don’t think they have to worry about anything now, there is still a large catalog of games that are yet to be ported, but what to expect for them, let’s say, 5 years down the road ? Interesting times.