AMD Closes in on Intel in Latest Steam Hardware Survey (tomshardware.com) 58
AMD's share of processors among Steam users climbed to 47.27% in December 2025, a 4.66% jump in a single month that continues the company's steady encroachment on Intel's once-dominant position in the gaming CPU market. Intel held roughly 77% of the Steam Hardware Survey five years ago, and that lead has eroded considerably as AMD broke the 40% threshold in the third quarter of 2025 and kept climbing.
The gains came despite an ongoing memory shortage that has pushed DDR5 prices to record highs -- AMD's AM5 platform requires DDR5 exclusively, while Intel's Raptor Lake Refresh chips support both DDR4 and DDR5. Many gamers are turning to older AMD Zen 3 processors like the Ryzen 5 5800X, which topped Amazon's bestseller lists during the holiday period and work on DDR4-compatible platforms. Meanwhile, the proportion of Steam users running 32GB of RAM rose to 39.07%, nearly matching the 40.14% still on 16GB, as gamers likely rushed to upgrade before prices climbed further amid AI's demand for memory.
The gains came despite an ongoing memory shortage that has pushed DDR5 prices to record highs -- AMD's AM5 platform requires DDR5 exclusively, while Intel's Raptor Lake Refresh chips support both DDR4 and DDR5. Many gamers are turning to older AMD Zen 3 processors like the Ryzen 5 5800X, which topped Amazon's bestseller lists during the holiday period and work on DDR4-compatible platforms. Meanwhile, the proportion of Steam users running 32GB of RAM rose to 39.07%, nearly matching the 40.14% still on 16GB, as gamers likely rushed to upgrade before prices climbed further amid AI's demand for memory.
Latest survey looks sus (Score:2)
Intel? (Score:2)
Intel, what the heck happened to you? In the past I would have never considered an AMD processor and/or GPU. Now it looks like the only way forward. How the mighty have fallen.
Re: Intel? (Score:2)
easy (Score:1)
Re: (Score:2)
Athlon was great in P3-P4 era. Bulldozer sucked monkey balls and got utterly crushed by intel's offerings at the time. It wasn't until ryzens that AMD recovered.
Re: (Score:1)
Bulldozer sucked monkey balls...
Except this wasn't actually even true unless you were running software built with the Intel compiler which exclusively sandbagged AMD chips - a fact they eventually got caught and sued for, and lost, twice! Linux users (who almost exclusively used GCC-compiled software at the time) had already realized what was up by then, but somehow Intel still managed to keep their crimes secret from the bulk of their Windows-using fanboys and the stock market as a whole for years.
Nothing stays hidden forever though, and
Re: Intel? (Score:2)
I had several FX CPUs. There were significant problems with overheating. I had to give up using one in an HTPC. It required too many small fans and was too loud for home theater use. The performance was fine.
Re: (Score:1)
I just keep mine in another room with massive aftermarket fans on them, and stream to a Steam Link over the LAN. (Yes, I'm still using some, the performance is just fine for the high-end ones still.)
Re: (Score:2)
I am very sensitive to noise, and there is no room I want to have loud fans in. There were other technical constraints as well, such as HDMI cable length limits, already borderline from projector to receiver - it may not have worked with a remote HTPC source. Right now, that HTPC is using an Intel 13400, passively cooled. No fans in the case at all.
Re: (Score:2)
It will never cease to be funny to me how fanboys are just blind to reality.
Bulldozer era was a disaster for AMD. They got utterly crushed in CPU market because of how awful their offerings were at the time. All while Intel made what was probably their best CPUs ever on Sandy Bridge. 2500k alone is utterly legendary in how long it remained viable due to its hilariously efficient architecture.
You have to be utterly blind to reality to even pretend otherwise. Bulldozer to AMD was what P4+RDRAM was for intel.
Re: (Score:1)
You're projecting pretty hard here, and your account of the situation clearly isn't based on first-hand experience. Sales were suppressed by marketing lies; you can't just use those as your benchmark for which chips were better.
Re: (Score:2)
Sales were suppressed in Athlon-P4 era, not Bulldozer. That was intel's "let's do architecture from ground up" and they made a retarded single thread focused thing that required that narrow bus super fast super expensive RAM. That was so awful that they had to force resellers to limit AMD's Athlon sales or lose access to intel's preferential pricing. Iirc Intel got convicted for this.
Bulldozer was the AMD's turn to do "retarded from the scratch redesign architecture era", where AMD went with their "clustere
Re: (Score:1)
Seriously? You're still trying to get the last word in here? You're still spouting stupid talking points based on Intel fanboy dogma. I'm not the blind fanboy here. Everything your argument is based on is either Windows specific or second hand lies. You're completely delusional, and it's pathetic how hard you need to latch your pride on to some stupid failing megacorp.
Re: (Score:2)
Son, you can find this shit on AMD brochure at the time. They went all in to advertise that "new way to do multithreading". They really through it would catch on, just like inter thought P4's idea of "just fucking single thread maximum megahurtz" would work.
Both companies had to reset and go back to basics after those stupid misadventures where they tried to reinvent their CPUs. And both succeeded.
And I have zero company loyalty. I've owned CPUs from both AMD and intel. I've owned GPUs from ATI/AMD and Nvid
Re: (Score:1)
Dude, you don't run Linux so you have no idea how irrelevant this shit all is. All you've got is what other people who were already invested in Intel winning and desperate to force the matter with words said. The chips were faster and cheaper for practical uses, period. You can't even use all the cores in those Intel chips at once without taking a performance hit. It's useful idiots like you, parroting these stupid talking points and bullshit corner-case benchmarks with Intel's cheater compiler, that kept I
Re: (Score:2)
I'll just remind you that corporate loyalty as a customer is one of the dumbest, most self-damaging things you can perform.
Re: (Score:1)
But that's what you're doing here, dude. I literally just told you that. I actually started out as an Intel fanboy too, back when I was a kid and didn't know there were alternatives, but eventually I got smart enough to do my own research and testing, and include the price to performance ratios in the data. Look at yourself, you're here on this stupid site in the middle of the night, trying to get the last word in on an argument you've lost 3 times now but can't admit it to yourself. The problem with you In
Re: (Score:2)
"I bought slot A athlon, it was great, way better than P3"
"That makes you an intel fanboy!"
Your emotional state has reduced your IQ to this level. Think of your brain cells. They don't want to live in this enslaved state.
Re: (Score:1)
Are you kidding me? You're still replying to get the last word in and you think you're not obviously trying to astroturf here? How about this, prove me wrong by not replying to me this time. I fucking dare you; see if you have the willpower. Somehow I doubt it. Somehow I think it'll drive you absolutely insane not to get the very last word in on this conversation thread, because you're such a hardcore, blinded, brainwashed Intel fanboy, and that's proof in itself that your view here isn't based on rationali
Re: (Score:2)
Ryzen happened. Then the Intel 13/14th gen happened that had stability issues. Then Intel's latest gen CPUs bombed because they couldn't beat Ryzen in any metric. The pandemic really helped.
However, Intel 13th/14th gen are making a comeback thanks to AI - because they can run DDR4 memory. AMD's AM5 platform is DDR5 only, and if you've
Re: (Score:3)
In the past I would have never considered an AMD processor and/or GPU.
Never? That would make you a fanboi. These companies leapfrog each other periodically. There were several times in the past where it made far more sense to buy an AMD CPU for gaming than Intel, including back in the Netburst architecture days.
Re: (Score:3)
I was a PC tech for local computer shops back in the day. AMD processors came back too often with issues. Intel CPUs never did. Not to mention that stupid P rating AMD used back then. For example, they marketed the CPUs to end users as being 133MHz but the BIOS reports the true speed as 120MHz. Try explaining that to a customer that thinks you ripped them off. I used AMD in the 386 era (loved my AMD 386DX-40, my first DOOM machine) but the CPU failure issues among other things that came later made me person
Re: (Score:2)
For example, they marketed the CPUs to end users as being 133MHz but the BIOS reports the true speed as 120MHz.
That was the type of thing that happened when the ram was either rated slower than required OR the BIOS defaulted to the lower speed for the RAM. Use/Set the memory correctly and you got the full speed. You could set it higher and overclock the CPU.
Re: (Score:2)
This was back in the 486 / early Pentium days, 1994 ~ 1999. BIOS' were a bit different back then. Attempting to set a jumper to a higher frequency usually ended in frequent and random lockups to no joy coming from the system at all. Adjusting CPU voltages with jumpers didn't really become a thing until the Pentium MMX and AMD K6 days and even then changing voltage for the sake of trying to overclock usually didn't work as hoped.
Re: (Score:2)
On one hand this is partly true. For example the early processors with multipliers were locked, there was no way to change them. On the other hand there were still things you could do, and one of them was adjusting the FSB frequency.
Re: (Score:2)
Sure, but it was not as straightforward or supported as it is today. You took a chance when overclocking and hoped your CPU/motherboard/RAM combination could handle and accept the altered settings.
Re: (Score:2)
That's still true. Unless you're just turning on PBO, there are still literally all of the same risks with overclocking.
Re: (Score:2)
It shouldn't be too difficult to explain to a customer if you have basic communication skills (and by that I don't mean to insult you but rather have an ability to distil something complex into something more simple). The P rating was an equivalent rating. You could have educated someone with information they were going to very shortly need as the entire industry moved away from MHz a few years later.
As for issues, I was also a tech back in the day. The leapfrog principle also existed. Both companies produc
Re: (Score:2)
Why not? I have used both Intel and AMD CPUs in the past decades (and back in the 1990s Cyrix and Nexgen). Same for graphic cards, have tried out all kinds of stuff back in the day when hardware wasn't boring yet. They all had their benefits.
Re: (Score:2)
They all had their benefits.
Cyrix was always far slower than their benchmarks suggested in the real world, and almost everyone else's chips as well. The only company more worthless was IDT. This is weird to me because Cyrix had a lot of fancy features, but they just never paid off. At least a K6 was fast if you built software for it.
Re: (Score:2)
They were good for a cheap second computer that is rarely used.
Re: (Score:2)
They were good for a cheap second computer that is rarely used.
Ordinary operations became tedious and compatibility was poor, requiring compatibility patches for many programs. That may have been largely intel's fault (with their icc compiler deliberately providing inferior performance on compatibles) but that was no comfort to the user who was affected and none of that is good. This was a problem for the K6 as well, but at least by the time of the K6/2 they had fixed the architectural deficiencies (notably with the fpu's precision) and it was just down to intel's fuck
Re: (Score:2)
I had zero problems with compatibility. Everything worked as it should. Honestly, the only hardware problems I had back in the day wwre caused by bad quality EDO RAM - so much crap memory has been sold back then. My first test for that used to be the OS/2 Warp installer floppy, the installer always crashed if the RAM was bad.
Re: (Score:2)
Re: (Score:2)
Arrogance, stupidity, incompetence, and then they lost the only advantage (besides criminal business practices) they ever had: The superior semiconductor manufacturing process. They are done for.
Steam survey lagging (Score:1)
The last truly dominant desktop CPU Intel produced was the 9900k. The last competitive gaming CPU they produced was the 12900k. The 13900k and 14900k destroy themselves, and the 285k was too little/too late. Gotta wonder why Intel still has so much representation on the Steam Survey?
Re: Steam survey lagging (Score:2)
Re: Steam survey lagging (Score:2)
I still have a i5-9400F that does everything I need. Won't be replacing it anytime soon unless something breaks (the i5-9400F in fact was a purchase forced on me by a motherboard failure in January of 2020)
Re: (Score:2)
Maybe, but it really shouldn't have taken this long.
Re: (Score:2)
Yep, I have used an old sandy bridge xeon for gaming until maybe two years ago when it finally went back into a server.
Re: (Score:3)
Re: (Score:2)
And? You don't just buy AMD for a "big powerful gaming rig". An old 5800X3D is the best deal in old CPU tech going if you can find one (which is difficult to do). And AMD sold the hell out of the 3600/3600X.
Re: (Score:2)
Re: (Score:2)
I occasionally play on an old W510 laptop, Also Intel, the OG Core i7 laptop CPU.
Re: (Score:2)
Laptops remain the domain of intel.
Also you don't need to upgrade CPU for many years now.
Re: (Score:2)
In terms of ubiquity, yes. Intel did stuff the channel with a lot of downmarket 10nm/Intel 7 junk. Hard to say if that's what's really affecting Steam survey results though.
Re: (Score:2)
x64 is the important one at this point.
Re: (Score:2)
x64 is the important one at this point.
That is pure ignorance, x64 is not at all important. You can't have an x64 processor without x86 patents. The 64bit part is just a few minor extensions on what is predominantly an x86 processor.
Re: (Score:2)
The way Intel and AMD had their treaty, Intel was allowed by AMD to use the 64-bit extensions of x86 that the latter invented, while AMD was now allowed carte blanche to make 32-bit CPUs that Intel had previously prevented them from making. So if a company wants to make a 64-bit only x86 CPU, they would just need to license it from AMD. If they wanted to make a 32-bit only, they'd just have to license it from Intel
All this assumes that the patents on those things haven't yet expired
Re: (Score:2)
Re: (Score:2)
Depends on what we are talking about. If the CPU maker in question wants to build Wintel boxes for the low end of the market, or whether they are into embedded stuff. If it's the former, they could license it from Intel or AMD, depending on whether they want 32-bit or 64-bit CPUs. If it's the latter, then as you said, they could either use RISC-V, or maybe look at licensing Arm
Only thing w/ RISC-V: it's still pretty new, and there's not too much embedded software for it, the way there is w/ ARM, in the
Re: (Score:2)
they could license it from Intel or AMD, depending on whether they want 32-bit or 64-bit CPUs
32 bit is over outside of embedded. (Even there I'm seeing mostly 64 bit cores now in the higher end.) Neither of these companies is making any 32 bit PC CPUs any more, intel announced they would stop two years ago. The x86 compatibility in modern processors is a minuscule part of the whole thing now.
Re: (Score:2)
Only the base patent. If you wanted to make a processor which performed like a Pentium II then you're in luck, no need to pay anyone royalties. If you want anything faster then you better pony up, the entire SSE extension set is still covered under patents.
Re: (Score:2)