Friday, 1 November 2013

Counting PC Gamers

There has never been a better time to work out how many PC gamers are out there buying new hardware each year.

Coming up for a decade ago we knew that around 90% of motherboards shipped with integrated graphics, back before CPUs gave up an area of their die for an iGPU.  But they were also quite crappy a lot of the time and lacked the ports and performance, even for 2D sometimes, to be useful.  Just as motherboard audio was often looked down on as no good, even if you didn't play games then a discrete GPU wasn't the worst of ideas.

But something changed.  Just as with motherboard audio, combined in that case with the Vista driver changes that killed hardware audio acceleration, the floor kept rising and soon there was not a lot of reason to buy your own discrete AIB (add-in board).  We are now at a point where the integrated graphics, now on the CPU rather than the motherboard chipset, are everywhere and good enough.  Intel and AMD are even starting to push the claim that if you enjoy some 3D games then you'll still be able to just buy a single clip from them so before their claims hold too much water we are given an interesting opportunity to get some rough idea of how many new or refurbished gaming PCs are being purchased every year.

Luckily Intel and AMD have helped us out, because if you're not gaming then you don't need an AIB (discrete card GPU) or a laptop with an nVidia chip in (AMD also bundle their CPUs in the numbers so I'm ignoring them - nVidia numbers come with a percentage of discrete laptop GPU market indicator from which you can extract the real AMD numbers).  There isn't even really a super-low end, not for discrete GPUs, because the rising performance of iGPUs that come with every CPU you buy has killed the market.  There are not 115m AIB sales like there were in 1999 (seems to be the peak, also about the time more chipsets started integrating a graphics option on some motherboards) but almost all the ones that are left are for gaming systems.

We should have a rough idea of where the other major market is.  Consoles, both home and portable.  The market leading console can sometimes sell ~20m a years, which is how we get one or two consoles that break well over 100m global sales over their (previously somewhat limited for time in the Sun as the primary device) lifetime.  Everyone else (at least in recent times, where there has been no clear loser) is getting closer to 10m sales a year (but you give sales from 2005-2013 and you get your 75-80m units that press releases from Sony and MS point to).

How does the PC platform stack up against that, in what we think is the last couple of years of some pretty good pickings for extracting only the 'gaming was a feature request' sales?  You see about 35-40m nVidia desktop GPUs sold each year.  AMD is more like 25m.  That's still getting on for as many dedicated gaming cards in a year as there have been 360s sold since sometime in 2006.  But if you consider the 50 million nVidia mobile GPUs and 20m AMD discrete units then we're not even playing in the same ballpark.  There are almost certainly more gaming PCs sold every year (consistent for at least a while, much harder to get an impression of sales when I can't be assured people didn't need to buy a GPU to make a complete PC and so it becomes real murky to estimate gaming PCs more than a few years ago) than there have been console devices of a single platform in play at any one time.  Only the lifetime PS2 and DS sales even poke their head above the ~100m ceiling to possibly eclipse PC dedicated GPU sales for a single year and with the hardware revisions, and expected lifetime, and known element failure rates (which don't really count for GPUs that are all within year 1 of the warranty) then how many of those were actually in any state to be used to play games by the end when the sales total reaches that high?

Lost, broken, sitting on the shelf at a used store, in the cupboard.  That doesn't sound like the fate of many GPUs purchased in that year but a 6+ year old console (one of many iterations and colours) may have a significant hit to how many are really in the wild.  I'd be pretty confident that annual gaming PC sales is significantly larger than the combined console market and that probably has been true for a long time.  And unlike mobile sales, where you don't really know how many people want a phone and how many want a gaming platform with benefits, the current market segmentation means you're throwing money away for an identical product for your needs if you don't buy that discrete GPU because you at least know you'll want the option of gaming.  Maybe you don't realise your 10 year old games will run just fine on the CPU's iGPU today but I'm still going to count that as a gamer worth counting.  They're buying the gaming hardware.  Just like a cinephile might have purchased a PS3 for the cheap access to a good HD movie player, or in today's market maybe a crazy person purchased a Netflix box rather than buying a $50 Android or similar device for the same purpose.

I hadn't seen it actually laid out, with a reasonably coherent argument for how the current market actually makes it pretty easy to count likely PC gamers.  The numbers are more analyst report averages (GPU makers don't seem to publish chip volumes, only financials) but do roughly tie into where I was expecting and a vague idea of revenues (and the comparative volumes sold at each price tier).  I don't think they're going to be off by enough for it to matter (even if you halve the GPU numbers you're still looking at quite a gap to the next nearest platform adding new units).  When we talk about how well PC gaming does (is it dead, is it reborn, can it never truly die, is it going mobile tomorrow?), the only real question is how many gamers we can convert to the insatiable console appetite for a high attach ratio (several games per year habit, and paying for them at retail for the tracking to work) to make the software side explode into being more important than consoles.  In terms of hardware then the consoles are the amusing iOS minnow to the Android shark, lots of noise and software sales but very few units out there globally when you look at the competition.  And when I started writing this post I had no idea I was going to end on that apt comparison to another closed platform that can make far more PR and software sales via the mandated single store but not nearly the hardware sales volume of the open platform it competes with.