Transcription
So, how much is it and how powerful is it? Well, Steam Machine has finally broke cover and we finally have definitive answers to the former and benchmarks for the latter.
Delayed owing to the hardware component crisis, Valve's beautifully tiny console-like PC is finally here. And from a form factor perspective, it's kind of irresistible. Imagine a taller Nintendo Gamecube and you have some idea of how tiny this thing is. While performance falls into line quite nicely with what the Steam hardware survey defines as the quintessential mainstream gaming PC.
But man, the prices are going to hurt. It starts at £1,049 or $879 great British pounds for the basic 512 GB model, rising to an oddly specific $1,128 USD or £938 for the bundle pack with the Steam controller. Now those prices rise to $1349 USD or £1149 for the preferable two terabyte model and that rises once again to $1428 USD and £128 for the bundles with the Steam controller. Now the two TB model also comes with a brace of face plates, solid walnut and red fabric, but man, it's pricey. Very pricey.
It definitely makes you think about considering a pre-built PC or building your own. And I'm sure there'll be a range of spec suggestions from anyone that knows anything about PCs. Valve itself in its blog isn't shying away from the fact that the audience has options and that they may well prove compelling. Ultimately, if you're buying PC games on Steam, they'll be fine either way. Still, there's something about this pricing I think we should consider.
Um, there's the physical machine itself, which has a premium feel, amazing form factor, virtually no noise in operation, very power efficient. The replaceable front plates, love it, very easy to work with, no screws. The panel is attached via magnets, so it's literally just a case of pulling off the existing cover and putting on the replacement. The standard Steam machine is an otherwise anonymous looking black box. So, the opportunity to personalize it in such an easy manner is really nice. Plus, there's nothing to stop third parties making their own face plates. Combine all of these things together and we have a lovely bespoke mini PC.
And I think we also shouldn't lose focus on Steam OS itself, which remains an absolutely brilliant front end packed with options. Everything that worked on Deck works here. The difference being that the extra performance makes for a flowing, non-flawless interface. So there's the experience, right, which is pretty flawless in my opinion. But then there are the quantifiables, specs, and performance. And if that's your focus, yeah, obviously maybe you should be looking at the alternatives.
If you're looking to upgrade your display and enhance your gaming experience, the Innocent G27S1Q is a 1440p QD OLED monitor that delivers perfect blacks and vibrant colors for total immersion in modern titles. Built for pristine motion clarity, it features an impressive 280 Hz refresh rate and a near instantaneous 0.03 millisecond response time for flawless high frame rate gameplay with two Display Port 1.4, four, and two HDMI 2.1 inputs along with G-Sync compatibility. It's fully equipped for both high-end PC gaming and current generation consoles. To find out more, check out the link in the video description below.
Spec-wise, then let's kick off by looking at the CPU where we have a six core, 12-thread AMD Zen 4 CPU clocked at up to 4.8 GHz. And I've seen a couple of those cores hit 4.88 GHz in gaming. The rest running significantly slower, likely down to the makeup of the Hawk.2 architecture, which has two full-fat Zen 4 cores, the rest Zen 4C. 30W TDP there according to Valve.
Then there's the GPU, RDNA3, 28 compute units, rated clock speeds up to 2.45 GHz. I've seen it maxed around 2.3 in that sort of area. 110 watts TDP. I can confirm now that it's definitely a pared-back version of the RX7600 based on the Navi 33 processor. 8 GB of VRAM there with 16 GB of DDR5 memory. Despite the memory crisis, we do still get 16 GB of DDR5 system RAM. But it is just one 16 GB module in the Steam machine there, not two 8 GB modules in a dual channel configuration. Uh, we understand by speaking to Valve that the OG spec was dual channel, but lack of availability of 8 GB SODIMMs has ruled that out.
I didn't want to completely dismantle my review unit while I still needed to test it, and time was limited anyway, but Valve was kind enough to send me a couple of pictures of the main board. The board itself is tiny, almost Pico-ITX like according to Valve. And there you can see on the left image there the Navi 33 GPU, uh, surrounded by four 2 GB GDDR6 memory modules for 8 GB total. And on the right there you'll see the Hawk point2 CPU. The image on the right flips the board over showing the two SODIMM slots, one occupied. One module in place is also confirmed if you run CPU-X on the Linux desktop. So you can upgrade components in Steam Machine, albeit in a limited sense.
Upgrading the SSD is quite simple. First of all, flip the machine over and remove the four screws there. Unfortunately, Torx screws are used there, but Torx 8 or Torx 9 screwdrivers will do the job of removing the screws and the feet they're mounted into. Then on the rear, you've got another couple of screws to unfasten. There are clips at the bottom on the rear there that you depress to pull off the casing. And there you'll find the SSD. A 2230 SSD is found as standard, presumably so Valve can share sourcing with Steam Deck, but there's plenty of space for using a standard 2280 drive instead. Uh, one thing you'll notice is that the vast amount of the device internally is taken up by the heatsink. Valve has been really serious about keeping everything cool here. The heatsink is quite a unit, but yeah, gaining access to the main boards to swap memory looks a touch more involved. However, Valve says it's easier than accessing Steam Deck and will be producing repair and upgrade guides.
So, lovely machine then. The Gabe Cube is a thing. Barring Switch 2, a completely different type of console, Steam Machine is far, far smaller than a PS5, a PS5 Slim, or Xbox Series X, and can produce broadly equivalent gaming experiences. You can build your own PC, of course, but pretty much the best you can hope for in terms of form factor will be some kind of Mini-ITX based construction. Uh, prices on the Steam Machine are high, but prices are also high for building a small form factor PC, right? Uh, PS5 and Series X are cheaper, of course, but that's not much help if you're looking to play your library of PC games, right?
And I also think we need to talk about how going for a small form factor box gives console-level efficiency even when a single system on chip isn't being used. Typically not something that excites people, but it makes the Steam machine's tiny form factor possible and puts system performance into perspective. So at the front end, at idle, you're typically in 15 watts to 17 watts territory, while interactions with the menus can see that spike into the 50 W range. But it's gameplay that's more interesting. Crimson Desert here at optimized settings in the Bug Hill area sees CPU and GPU close to being fully tapped out. Both of the main power drivers here are really drawing as much power as they need. Power draw measured from the wall varies between the 180 to 190 W area. I should say that this is the worst case I could find in my limited testing. A game where we're typically GPU limited with a lower level of required CPU performance takes that down into the sort of 170 to 180 watt range. And obviously if it's a more basic game, it'll probably drop still further.
So in PC terms, what kind of performance does Steam Machine offer compared to building a PC yourself? The first thing I did was to actually build one a while back to satisfy my curiosity about ballpark Steam Machine performance. So, the Ryzen 5 8500G has a very similar CPU architecture to the Steam Machine's Hawk.2 derived processor. However, I quickly discovered the chip only supports PCIe 4.0 across four lanes connected to the GPU, not eight, limiting bandwidth with an inevitable impact to graphics performance. I can confirm that there are eight lanes connecting the CPU to the GPU within Steam Machine. And with that 8500G, I benchmarked it in CPU-limited scenarios, finding that the Ryzen 5 5600X was a fairly close match in gaming performance. And that 5600X also happens to be the entry-level CPU of choice for the AM4 DDR4-based platform. A solid budget option in the current market. So that's what I went with for most of my Steam Machine versus PC comparisons. I also use 3200 megatransfers per second DDR4. You can get faster, but this remains relatively cheap with the uh emphasis on relative obviously in the current market.
GPU then, uh, in a desktop space, there can only be one: the Radeon RX 7600, the exact same Navi 33 processor as the Steam machine. However, with the 7600, wattages are higher and all 32 compute units are active as opposed to Steam Machine's 28. So we should expect more performance then. But of course, the RX7600 is effectively old technology now. RDNA3, and it's been replaced by the Radeon RX960 XT. Yeah, RDNA4 costs more, but you get a ton more performance. Rounding off the group test, I went for the Radeon RX6600. For years, you could buy this for under $200, and it also represents the GPU I'd recommend as the lowest spec for still getting a decent experience from AAA Gaming. Uh, DirectX12 Ultimate feature set and the value it offered back in the day was just tremendous. Um, it's an interesting comparison point for Steam Machine as it has similar GPU clock speeds, the same 28 compute units. Memory bandwidth is higher on Steam Machine and obviously the 6600 is using the RDNA2 architecture whereas the machine's using RDNA3.
But before we let the benchmarks roll, at the time of testing, these numbers come from a GPU driver that hasn't yet received ray tracing optimizations. However, neither had the GPU driver from our competing GPUs. And what we're really looking at here aren't the basic numbers as such, but the performance differentials. Um, but uh, we'll talk about ray tracing a bit later. And yes, everything here is tested under Linux, of course. We're not maxing everything out. We're choosing optimized settings. And optimized settings where specifically we've tailored them for 8 GB GPUs. Uh, because this is a much more realistic choice for this class of hardware on modern games.
And kicking off with Black Myth: Wukong here. The numbers may look low, but that's mostly down to a weird issue with the game where the display resolution of the device connected can impact the numbers even when you specify lower resolutions. So, I'm selecting 1440p here on optimized settings, but my screen is 4K. So, for some reason, the numbers are lower than they would be in an actual gaming scenario. Even so, as I said, it is about the percentage differentials between the GPUs tested, which all have the same issue. Oh, and one more thing to note before we crack on. Benchmarks are all tested with VSYNC disabled for full performance. However, Steam OS itself defaults to VSYNC on. The reference videos in the background of the benchmarks here were captured from Steam Machine at the same settings with VSYNC active. So the benchmark numbers represent full performance. The video represents what you'd actually see out of the box with Steam Machine.
So with all of that in mind here, the full-fat RX7600 in Black Myth: Wukong only manages to deliver an 11 percentage point uplift over the Steam Machine on matched settings. I expected more and you will see more elsewhere, but even so, the margin is thin on this specific test. Meanwhile, the Steam Machine itself is 11% faster than the RX6600. So, the Valve machine sits at a kind of midway point between 6600 and 7600, which is kind of like what I expected after first seeing the specs. The 960 XT represents AMD's latest more budget-orientated offering and delivers a 58 percentage point performance lead. Uh, compared to 7600, you pay more, you get a lot more. For those building a Steam machine of their own on a budget, there are cheaper options. But yeah, the 960 XT is an excellent contender, but you would have to pay a lot of money for the preferable 16 GB version, which is the one I tested here.
Um, let's move on to Cyberpunk 2077, the Phantom Liberty now, specifically in the black market at the beginning of the expansion. This is a properly heavy GPU workout. We're using the equivalent to PlayStation 5's performance mode settings here, and we're running at native 1440p. And as expected, the RX 7600 moves ahead. I'm clocking a 17 percentage point increase for the desktop GPU over Valve's integrated alternative. Meanwhile, the RX6600 offers about 90% of Steam Machine GPU performance. 9060 XT is a good 40 percentage points ahead in this test. Now, remember the reference video here is indeed the game running on Steam Machine. The overall average is 64 FPS at native 1440p on the equivalent of PS5's performance mode settings. Uh, not bad for a heavy scene. Great for VRR.
Now, if you think the 9060 XT is underperforming here, that isn't down to the GPU. It's down to the CPU unable to keep it fully fed. And that's something else to bear in mind when dealing with balancing mainstream-orientated PC hardware. Your bottlenecks can come from anywhere. The 1960 XT offers an arguably worse experience when looking at frame times there. And a reminder that it's always best to run GPU limited or at the refresh rate limit of your display or to a set frame rate target. Personally, I'd be increasing settings with a 9060 XT and using a frame rate cap.
And you can kind of see the same thing happening here in Resident Evil 4 Remake. The 9060 XT is actually 78% faster than the Steam Machine GPU here. But again, the spiky frame times result in some unattractive gameplay. So yeah, higher settings, frame rate caps, that's the ticket. The RX7600 is 16% faster than Steam Machine here. 55 FPS on Valve hardware versus 64 FPS on AMD's discrete GPU. And it's a 9-point advantage for the Steam Machine over RX6600.
Forza Horizon 5 now, 1440p on lower-end optimized settings basically aimed at 8 GB GPUs without ray tracing in play here. The RX7600 is 12 points clear of Steam Machine with the 1960 XT 70% ahead. Meanwhile, the RX6600 is only a couple of percent away from Steam Machine. But something's wrong here. You may notice that the frame times are pretty awful on all cards and that often the game seems to be moving in slow motion. This isn't how Forza Horizon 5 is supposed to run. And perhaps a reminder that we aren't running under Windows, and levels of incompatibility may be a thing occasionally on Linux. Oh, and one more thing. As I said, we're using the 16 GB version of the 1960 XT in these tests, and it exhibits the same issue. So, it's definitely not a memory limitation.
Doom: The Dark Ages is next. A game that uses ray tracing as standard. Optimized settings here tailored for 8 GB GPUs. Remember, there are a bunch of different benchmarks to choose from, and I'm using the Abyssal Forest stage as it's the most demanding in the bunch. RX7600 here is only 9% faster than Steam Machine. 9060 XT though, a gigantic 76% ahead. RDNA4 really does have very, very decent ray tracing capabilities. Of course, the RDNA3 architecture had RT improvements over RDNA2. So, here the results skew in favor of Steam Machine. A typical 10 to 11% lead transforms into a 17-point advantage for the Valve hardware.
Finally, Crimson Desert, another ray tracing title, though its use of actual hardware RT is more limited. Optimized settings are demanding clearly with just a 27 frames per second on Steam Machine at 1440p. But do remember this is native resolution and upscaling options are available and the Valve hardware will get FSR4 support at some point. Here I'm seeing a 16 percentage point lead for the RX7600 combined with a 73 point lead for the 960 XT. The RX6600 has about 93% of Steam Machine performance, meaning a 7.5 percentage point lead in favor of the Valve box.
So towards the end of the review period, Valve did release a new Steam OS with upgraded ray tracing performance. So I had the chance to compare before with after. It didn't make any difference for me compared to the prior driver in Doom: The Dark Ages on our selected settings. Similarly, Crimson Desert only saw an improvement within margin of error, like sub 1%. On a couple of other benchmarks, I added ray tracing where there was none before. So, in Forza Horizon 5, I kept the same settings and added medium RTGi and used FSR3 quality mode to improve the depressed frame rate level. End result is a 6 percentage point gain, though the problems in general performance still remain clearly. Uh, finally with Cyberpunk 2077, I kept the PS5 performance mode settings, added FSR3 performance mode, and turned on RT reflections and RT lighting at the medium setting. These are what I would describe as our optimized ray tracing selection for this game. Um, this delivered a 9.7% frame rate boost, which is certainly significant, though we are still under 30 FPS at times on this one. Yes, it's a demanding scene, but we are at FSR3 performance mode there. Internal 720p. So, yeah, still just a touch too demanding for the Steam Machine there.
Another takeaway here is that the improvements are context-sensitive. Some areas of the bench show no improvements at all. Others show dramatic improvements, demonstrating that comparing frame rate averages, perhaps only shows part of the picture. Ultimately, the improvements to ray tracing performance are obviously welcome, but I guess RDNA3 at the architectural level isn't really a ray tracing powerhouse, and Navi 33 is a processor. You know, it isn't a top-end offering. One more thing to factor in is that ray tracing does add a burden to VRAM and CPU resources, and those may also be at a premium on Steam machine depending on the game.
Now, let's talk about the CPU. And here's where things get a little tricky. In most games, you'll be GPU limited most of the time, but when you do hit CPU limits, it's keenly felt, often with very spiky frame times. It's not great. And the CPU in Steam Machine is based on the Hawk.2 architecture, which is relatively weak, relatively, as in compared to what many consider to be good entry-point gaming CPUs. It's further compounded by something I discovered in my testing and confirmed uh with Valve when I spoke to them. Typically CPUs are paired with two memory modules to give dual channel bandwidth and the more bandwidth the CPU has the better it performs in gaming. Certainly in CPU limited scenarios. Valve couldn't source 8 GB SODIMMs. So instead of two 8 GB modules in dual channel, you get one 16 GB SODIMM in single channel. And there will be performance implications to that. Um, to give you a broad idea of what this means, check out this test of Crimson Desert's now infamous Bug Hill tested on Steam Machine. Ryzen 5 8500G in single and dual channel configurations and the Ryzen 5 5600X in dual channel. Outside of Steam Machine testing, we're using the RX7600 here as it's the closest equivalent GPU to the Valve offering.
First of all, note the crappy frame times on all configurations here. The most obvious manifestation of being completely CPU limited. 1080p PS5 performance mode settings, it's a viable real-world test. The 8500G doesn't seem to be PCI Express limited here. And I noted frame rate differentials of between 15 to 25% across the length of the bench comparing single channel to dual channel. So bandwidth does matter, but the extent of how much it matters can vary even within the context of a single benchmark. Like for like single channel versus single channel. The 5600G is 18% faster than the Steam Machine CPU, while the dual channel 5600G is 35% faster than Steam Machine. And I think it could go higher if it had more GPU headroom. We're still CPU limited on the 5600X, I should add, and it's 31% faster than the Steam machine. Valve were basically forced by market conditions to use single channel, which is a shame as CPU performance could and maybe should be better. But I guess that does leave the door open for a user upgrade further on down the road. Remember the Steam Machine board itself does support dual channel memory configurations and has two slots.
Okay, so that's just one CPU test, but I think it lays out the expectation level you should have uh for this component in the Valve machine and how it compares to desktop parts. So, the conclusion is pretty clear. Steam Machine delivers what I'd call ballpark entry-level PC gaming performance, or at least entry-level based on, you know, a decent mainstream experience. The CPU should provide better results than a console running the same workloads, while the GPU sits at a kind of midpoint between the RX6600 and 7600 as predicted. It'll run older games just fine, but no problems at all. While more modern fare from this generation and will require special attention to settings depending on the title. So yeah, there's a reason why we chose optimized settings and optimized low presets for the benchmarks. Just like a console, you'll need to tweak settings to taste for good performance.
Now, I'm wondering the extent to which developers will use Valve's APIs and deliver ready-made, dial-in settings out of the box. Digital Foundry specializes in optimized settings, lower-end optimized settings for 8 GB cards. And judged by that criteria, perfectly credible experiences are possible.
So, bearing in mind the form factor, the use case, how does Steam Machine compare to a console? We got a wealth of information on PC titles running at PS5 equivalent settings. So, let's get straight into it and we'll start by revisiting the Crimson Desert CPU test on matched settings. And that includes using VSYNC as PlayStation 5 does. Not that this adds any particular cap to performance here on Steam Machine. We're well under 60 FPS. Returning to Bug Hill, Steam Machine performs significantly better when CPU limited. A 34.8 FPS average across the run on PS5 is improved to 41.2 FPS average on Steam Machine. So, the Valve hardware has more legs clearly. We're also CPU limited in the city of Bouldersgate in Baldur's Gate 3. And here's where things get weird. Using the default native Linux build, which is what will install if you download it, we're only 11% faster on Steam Machine versus PlayStation 5. However, force the use of Proton instead and use the original Windows code working through the translation layer, and we get a 24% improvement in performance against PlayStation 5. Larian did update Baldur's Gate 3 for Steam Deck, significantly improving CPU-bound performance and stability, but I'm wondering if those optimizations were Deck-specific to the detriment of other PC hardware. Put simply, if you're playing Baldur's Gate 3 on Linux on Steam OS, or on Steam Machine, I'd recommend switching to Proton here. It's an automatic 11% performance boost in this test versus the native Linux app.
Graphics now. And in a world where the RX7600 is broadly comparable to the PS5 GPU, we should expect the Steam machine to fall a little short. But I didn't see anything that drastically concerned me to be honest. Um, let's kick off with Black Myth: Wukong. Um, it's a good start because it's not a particularly great PS5 port, so it should favor Steam Machine. The widest delta sees the PS5 running 17 to 18% ahead of Steam Machine, while other areas can see a smaller lead for the Valve machine over PS5, while elsewhere we're looking at absolute par. It's fascinating to see it actually. On average, it's a 3% lead for PlayStation 5. But this doesn't reflect the reality that on a moment-by-moment basis, um, it's much more variable.
Next up, the classic Alan Wake 2 Chapter 2 stress test. And some interesting results here in performance mode. We're looking at a 9% lead for the PS5 version. Uh, 55.3 FPS plays 50.7 FPS. Not exactly a gaping chasm, right? Alan Wake 2 isn't available on Steam, of course. I got it running by using the Heroic Launcher installed in Steam Machine's desktop mode. Yes, it is a Steam machine, but we got to remember that it's not just a console-like experience. Under the hood there, it is a full PC with full desktop access. Uh, we can also do like-for-like testing in Crimson Desert based on performance mode. And the GPU performance differential is more like 17% here in favor of the PS5. And that makes sense to me bearing in mind how often we saw this number when comparing the Steam Machine to RX7600. And yes, the 7600 does have broadly similar GPU performance to PlayStation 5.
Forza Horizon 5 next. Uh, we can't do apples to apples here precisely as console could use dynamic resolution scaling topping out at native 4K with 4x MSAA and some settings like reflections themselves are dynamic which isn't an option on PC. Uh, what I've done here is to show Steam Machine running the same sequence at native 4K on the same settings and then at 1620p which represents 75% of 4K resolution on both axes. We're way short of PS5 performance at a lot 4K, but almost there at 1620p. I think bandwidth is the issue here. Bandwidth is um quite lavish on PlayStation 5 in comparison to Steam Machine. And even at 1620p, bandwidth heavy effects can cause small frame rate drops, but you know, 1620p still looks fine. And VRR basically would make this pretty much a non-issue. And actually at this point, I'd like to show something else I quite like about Steam Machine and AMD GPUs in general. There's support for a wide range of output resolutions, 1800p, 1620p, just to name a couple. As we're demonstrating here, we save a ton of performance, and on a 4K screen, especially in the living room at couch distances, you don't lose that much. In fact, the static HUD elements are more noticeably scaled.
One final Steam Machine versus PlayStation 5 game test. Hitman 2. We have console equivalent settings direct from IO Interactive on this one. But we have got a bit of an issue. The PlayStation 5 version has dynamic resolution scaling from 720p to 1080p using FSR 3.1.5 to upscale to 4K. The PC version does not have DRS and neither will it let you use FSR upscaling from 720p to 4K. You'd need an ultra-performance mode there that is not an option in the game. So yeah, the stresses that we're putting on the GPU here will vary even though the content is effectively matched. In the explosive airfield segment, we're kicking off a major chain reaction of destruction. It's one of the only GPU limited parts of the game where I've seen drops on PlayStation 5. So presumably we're at minimum 720p DRS at this point. FSR3 performance mode on Steam Machine has the same internal resolution, but far less FSRG GPU demands since we're only upscaling to 1440p, not 4K. And there are small dips here both on PlayStation 5 and on Steam Machine, running internally at 720p on FSR3 performance mode. Meanwhile, FSR3 balance mode has a higher internal resolution and drops to between 48 to 50 FPS. It's frustrating that PC doesn't get dynamic resolution scaling to match PlayStation 5 spec, but at the same time, PC does have full VRR support, and so does Steam Machine.
Here's another test where we're testing streaming. A PS5 can drop a few frames here and there, which does kind of suggest a CPU bottleneck. It's a clean 60fps with FSR3 performance mode on Steam Machine, and what must be the occasional GPU limit on balance mode. Either way, both Steam Machine variations here do run smoother than the PS5 build, and it's likely that the extra CPU performance is making itself known. But with that said, in our PC performance review, we noted some issues with crowds on the PC version that don't manifest on PlayStation 5. It's slight, but it's borne out in this shot. Locked on PlayStation 5, dropping frames on Steam Machine, and it doesn't really matter what resolution or upscaling factor you choose, you'll still get the same drops. That does indicate a CPU limited scenario. So, while end results can be matched, it's not always 100% clear what's going on behind the scenes.
Ultimately though, we've got console-class performance here from Steam Machine in a unit that's a fraction of the size. It'll run modern games well, assuming you don't go crazy with ultra settings, especially on Unreal Engine 5 titles. And that you keep in mind that Valve only supplies 8 GB of memory for the Steam Machine GPU. Yes, there are optimizations there. And as work is ongoing there at Valve, I decided to test this in depth at another time. But reasonably, uh, we can expect these mitigations to go so far uh, compared to a GPU with more RAM. Of course, RDNA4 and some kind of 9060 XT level GPU with 12 gigs of uh, VRAM would have been better, but this is a machine built to a budget, a budget that was then torpedoed by the storage and memory crisis. For as long as games are built to look good and run well on PlayStation 5, I'd say that Steam Machine should be capable of producing comparable results with flexibility in resolution, upscaling, and yes, frame generation options, delivering opportunities for Steam Machine users that PlayStation 5 gamers won't have.
So, let's try to bring everything together. When Steam Machine was first revealed, the question was all about price and performance. Based on specs, we've arrived with a piece of hardware that pretty much runs as expected, and that's fine. Technology-wise, it doesn't shift the needle, but I love the package. The form factor is beautiful, and it's wonderfully put together. Virtually silent, basically. To see it is to want it.
Now, back at the reveal, there were hopes that Valve was going to attack the market with an aggressive price point, $500 even, uh, bringing a PC-style PlayStation 5 level device to market with an appealing form factor that would work in the living room. But that was October 2025. In the meantime, Valve's made it clear it's not in the business of subsidizing hardware in the way that a console manufacturer might, and it clearly doesn't have the sourcing power of Microsoft or Sony. So, in June 2026, we've got a machine that starts at $1,500. And it's going to be up to you to decide based on everything you've seen across this review and others whether that's the machine that you want to buy. I mean, based on the recent Steam Deck restock, chances are it's going to sell out immediately anyway. But the point is that there are alternatives. Not having a Steam machine won't lock you out of PC gaming. After all, the PC remains the ultimate open platform. And even in these difficult times, there is still choice out there in the hardware space.
Still, that's all from me for now. So, the usual stuff applies about liking, subscribing, sharing, bell ringing, and whatnot. Do you have a supporter program? Well, joining us, supporting the team, it all helps. For example, all those PlayStation 5 comparisons, optimized settings, they're built from a body of work we've amassed over time, and that's only been possible uh thanks to our Patreon support. If you think we're useful and want to support our mission, patreon.com/digitalfoundry, you'll get updates from the team, early access opportunities, a brilliant community, ad-free high-quality video downloads, ad-free website. It's good stuff. But that's all from me for now. Thanks for watching and supporting Digital Foundry.