A recent report from Bloomberg has attracted a lot of headlines - the suggestion is that Sony is struggling to control construction costs for the PlayStation 5 and that the $399 pricing sweet spot enjoyed by both PS4 and PS4 Pro may be out of reach for the next-gen system. Bloomberg says that the current cost for building the machine is around $450, suggesting that a retail price of at least $470 is likely - but is the analysis accurate? Will the next generation consoles really be more expensive at launch than their current-gen equivalents?
Pricing of console hardware has been a topic we've raised in the past on Digital Foundry because the fact is that the economics surrounding consumer electronics have been growing increasingly challenging for console manufacturers in recent years, while the pace of technological innovation has also slowed down significantly.
I genuinely believe that next-gen console hardware from both Sony and Microsoft will be superb - but inevitably, delivering a proper generational leap comes at a price. Solid state storage is a game-changer, we'll have a desktop-class high performance CPU component this time and while it can't hope to match the 8x increase in capacity we saw last-gen, 16GB of GDDR6 looks likely. In terms of graphics power, the arrival of the enhanced consoles - PS4 Pro and Xbox One X - has muddied the waters somewhat in terms of defining what a generational leap actually is, but I'd still suggest we'll get there.
Let's put it this way. In 2013, PlayStation 4 was the most powerful console on the market, with 1.84 teraflops of GPU compute power. If the December 2019 AMD testing leak proves accurate, PS5's 9.2TF amounts to a 5x increase in performance, even before we factor in the Navi architecture's inherent improvements. It's a proper generational leap, but with Pro and X on the scene, it's from there that users now expect that huge leap in performance. And it's not helped by the fact that the standard 1080p resolution from 2013 has now increased four-fold to 4K in 2020.
The cost pressures on next-gen come from all directions. First of all, the main processor - the system on chip - is fabricated on a 7nm process, and while the crucial 'cost per transistor' is much lower, cost in terms of actual silicon area is a lot higher, based on this information at least, which suggests that a processor with the same area costs 66 per cent more than the 16nmFF process used by PS4 Pro and Xbox One X. We can only guess at the size of the PS5's chip, but Microsoft is showing off an Xbox Series X chip that looks bigger than anything else it's ever made before - significantly so, if the measurements of myself and others are accurate. However, you'd hope to see volume manufacturing on the scale Sony and Microsoft enjoy to bring about better pricing.
The Bloomberg report points to memory and storage costs being higher than expected too. Nothing official has been revealed as of yet, but I would expect both Series X and PS5 to ship with 16GB of GDDR6, higher than any consumer-level PC graphics card on the market (bar the out-of-production Radeon 7, yours for a cool $700). Even before any apparent shortage in production, this would have already been high up on the bill of materials. In terms of the availability of NAND memory, I'd also expect to see both next-gen machines ship with 1TB. Prices have been falling here, thankfully, and to be honest, I would have expected both platform holders to have secured supply way ahead of production. Compared to an old mechanical drive though, it's still a considerable investment.
Bloomberg also talks about an unusually expensive cooling solution and rather puzzlingly comes up with a one dollar cost for a standard cooler. I'd expect it to be a whole lot more than that, certainly with Xbox One X which features an innovative vapour chamber heat sink. However, the point that cooling will be a pricier item for next-gen build costs certainly stands - if the AMD test leak turns out to be true, PS5 has a GPU that hits 2.0GHz, a sustained clock speed that's extremely difficult to achieve on today's PC Navi parts. It's easy to believe that both Sony and Microsoft will want to get the highest performance possible out of 7nm silicon that's so expensive to produce - and running it faster, at higher clocks, is the way forward.
So, to achieve a next generation leap in performance, we should be prepared for a higher cost. A 7nm system on chip of equivalent size to the PS4 Pro or Xbox One X will be more expensive to produce, solid state storage is pricier than mechanical drives while doubling memory allocation only adds further pressure, and that's before we factor in innovative cooling. Stack that up against what Microsoft delivered with Xbox One X back in 2017 - a 360mm2 slice of silicon, 12 gigs of GDDR5 and a mechanical hard drive. Back then, Phil Spencer revealed that Microsoft made no money from the ticket price of the machine with a loss on hardware recouped from elsewhere within the Xbox ecosystem.
With that in mind, the concept of either Series X or PS5 hitting the $399 sweet spot seems unlikely. Technology and pricing has shifted from 2017, but a look at the PC GPU market and costs there suggests that it hasn't moved on that much. An AMD RX 5600 XT with 6GB of GDDR6 and a 251mm2 die size still costs $280, while a 5700, based on a higher grade version of the same processor with two gigs more VRAM is well north of $300. Again, console manufacturing volume should give Sony and Microsoft a better deal but as a baseline, it's still concerning.
However, while the console pricing sweet spot may seem unattainable, my contention is that even if the PS5 has a $450 build cost as Bloomberg claims, a $399 console is still very much doable if Sony is so inclined - and the reason why is very straightforward: supplementary revenues within the ecosystem but outside of the actual bill of materials are much higher than they were back in 2013.
Bloomberg's idea that retail price will be a minimum of $470 is based on costs from 2013, but these days, the way that games are sold should give the platform holders much better terms. Sales from digital stores can be anything up to 60 per cent of a game's overall sell-through, something unheard of back in 2013. With retail stores frozen out, the platform holders should be making more money from each game sold - and at a console's launch, attach ratio of software to hardware will be relatively high. Secondly, back in 2013, Sony was still building PlayStation Plus and it wasn't as essential to the console experience as it is today. In pricing a launch console, likely to be bought by committed gamers, Sony can rely upon yearly subscription income to offset some of the loss.
Put simply, while build costs may be significant - perhaps even higher than the $450 Bloomberg suggests - there is a fair amount of latitude available to the platform holders in setting a retail price, arguably more so than they've ever had in the past. However, despite this, there is some pretty compelling evidence that prices are going to rise - specifically, the existence of Microsoft's Lockhart machine.
Phil Spencer hinted at the development of more than one Xbox back at E3 2018, and it seems like Lockhart's design target is about trading game rendering resolution - and only resolution - in exchange for a console that's more affordable. Nothing official has been revealed and Microsoft isn't talking about anything other than Series X, but it's not difficult to envisage how the Lockhart concept works. By targeting a lower rendering resolution, the machine requires a much less capable GPU (4TF if the leaks are true) meaning a much cheaper processor, while we are likely to see a drop in memory too - leaks have suggested 12GB instead of 16GB. Whether Microsoft will also choose to cut back the SSD to save further costs remains to be seen but I would hope not.
I've been modelling four teraflop Navi performance with PC parts using console-level workloads in the last week, and while there's going to be a lot of scepticism surrounding the idea of the machine, I think it could work - and work surprisingly well, if games target 4K or dynamic 4K on a prospective 9.2 teraflop PlayStation 5. Scalability across resolutions looks promising, but I think the key point is this: if there were no concerns about next-gen build costs or retail price for consumers, Lockhart would never have made its way so far into Project Scarlett's development cycle.
As for PS5 and Series X pricing, I think it's going to be challenging, but I honestly think that at this point, it's not so much issues with DRAM or NAND supply that are going to define the price-point. Console manufacturers mitigate costs across the entire generation and in context of the whole ecosystem. Not only that, but both Sony and Microsoft will be looking forward to the next generation with far more confidence than they did back in 2013. Back then, everyone was predicting the death of the console and indeed the PC in favour of the smartphone and tablet. Seven years on, the console market is still healthy and the platform holders need not design so conservatively.
Ultimately, I think the biggest takeaway from the Bloomberg article is the suggestion that pricing may not be set until Sony has a better idea of what Microsoft's plans are. Of course, it's equally likely that Microsoft itself may well be operating with the same strategy, meaning that it's not supply or cost issues that are the real issue here - rather that we're seeing an elaborate game of corporate chicken play out before our eyes. Who is going to blink first? Ultimately, competition should ensure that we'll get a good deal and certainly, everything I know about PS5 and Xbox Series X is telling me that the wait will be worth it.