AMD Reports First Quarter 2026 Financial Results
Posted by DeeJayDelicious@reddit | hardware | View on Reddit | 55 comments
Posted by DeeJayDelicious@reddit | hardware | View on Reddit | 55 comments
Sevastous-of-Caria@reddit
Its like beating the dead horse atp but if you guys keep asking the "why gpu release pricing" "why arc dgpus cancelled" heres why.
Amd didnt actually miss any opportunities. They are too busy making bank on ai and server demand. And allocating tsmc fabs beforehand according to that. Why waste it on low margin brand perceptes consumer gpus when you can đź’µMAKE IT RAINđź’µ on AI. Its good business these days. Dont expect a agressively priced rdna5. A token effort and low stock is once again where my money is at.
kingwhocares@reddit
If only AMD actually used that money on R&D. They are far behind Nvidia in GPU and even behind Intel.
SirActionhaHAA@reddit
"Operating expenses were $3.1 billion, an increase of 42% year-over-year as we continue to invest in R&D"
Amd's r&d has been growing at >30% each yr. What are ya talking about?
TechTechTerrible@reddit
The efficiency of AMD GPUs is still lacking compared to nvidia, DC or otherwise.
Geddagod@reddit
On the client side, they might be behind Nvidia, but behind Intel too? Granted it's been a while since I checked, but I highly doubt it.
For DC, AMD from a strictly hardware perspective appears to be extremely competitive. And even if they aren't overall very competitive against Nvidia for DC AI GPUs, Intel doesn't even have DC AI GPUs at all, with constant cancellation and delays on their roadmap for said product.
The only thing they have resembling that so far is Gaudi, which seems immensely unpopular, and Ponte Vecchio, which is its own shit show.
TechTechTerrible@reddit
Behind Intel? Absolutely not. Should have made that part clear. I was more commenting on the R&D expenditure not being enough to catch nvidia.
kingwhocares@reddit
That is their whole R&D expenditure. Includes CPUs and others.
hackenclaw@reddit
it still a miss opportunity. Have you seen Nvidia's consumer GPU profit? Even half of that is still billions.
AMD can have both.
SirActionhaHAA@reddit
Their gaming gpu revenue's around $4.5-5bil each quarter, half of that is $2.5bil. That's like $10bil every year. Have ya seen amd's dc gpu revenue? It went from almost $0 to $8bil in 3yrs. That's equal to capturing 80% of the gaming gpu share in 1.5 generations, and it's expected to more than double in 2026 to over $16bil
Anyone with commonsense would know where to put their resources.
Pimpmuckl@reddit
Don't forget that gamers are also incredibly easy to keep within a brand.
Normies wanting to buy a GPU ask "which xx70 or xx60 should I buy". Not which brand, not which options make sense. It's simply Nvidia because that's been the buy for 20 years now.
In the Datacenter you have people that really care about TCO. That's a lot easier to make a case for as AMD than your average xXxSuppaKillaxXx
varateshh@reddit
Tbh with DLSS going xx60/xx70 going Nvidia is a no-brainer unless AMD offers a much deal on fps/$. I don't give a fuck about ray tracing (with current technology) but DLSS is good enough on its own. AMD is still playing catch-up here.
Pimpmuckl@reddit
That's reasoning and whatever your reasons are, one or the other, that's an informed decision.
That's not what 99% of all gamers do though.
If the situation was flipped, AMD had an insanely good upscaler and Nvidia had just shitty cards, people would still buy Nvidia.
That's my point: It doesn't even matter who has the better value card.
What matters, to 99% of the gaming market, is the sticker on the box.
Which means AMD giving about zero fucks to that consumer group makes perfect sense from a pure economy play here.
varateshh@reddit
People say that but AMD has not provided scale since they transitioned to TSMC. If they do not offer scale then OEMs will not do many builds with AMD in it. AMD cards will also have a token amount of cards available to the DIY market at elevated prices.
Vega 56/64 was the last time I saw competitive pricing on AMD cards anywhere near launch. RX5000 offered good deals as well but iirc it took ages for them to be properly available. After that the offers were dogshit. I have attempted multiple times to look at AMD as an option and every time they have disappointed. AMD has done paper launches and limited supply for ages now.
Pimpmuckl@reddit
That heavily depends where you are and in which market.
In Europe, you could buy each and every flagship card for MSRP or lower at launch and after launch, the only exception was the crypto-drought.
AMD consumer supply is utter trash in some parts of the world, notably Asia, but that has literally nothing to do with TSMC or GlobalFoundries making their chips.
SirActionhaHAA@reddit
Yea amd prefers customers who actually want their stuff and evaluate based on numbers, not brand loyalty.
randomkidlol@reddit
billions in revenue or profit? i guarantee the profit margins are much lower on consumer parts than datacentre parts.
trackdaybruh@reddit
Yup, I remember reading the profit margin on Nvidia’s datacenter GPU was around 80% and their datacenter gpu revenue for the year was around $200 billion
EmergencyCucumber905@reddit
Higher price, higher profit margin, and way way more volume.
ResponsibleJudge3172@reddit
Those roes are 5 years old and are not credible anymore
KoldPurchase@reddit
At some point, they will need a commercial GPU for these AI server farms.
It might not be bad business to derive consumer GPUs from there.
996forever@reddit
They already do have a commercial GPU and they always have. It's called the Radeon Pro. Formerly Fire Pro. And yes they are the same architecture and dies as gaming gpus (unlike datacentre GPUs without video output) with more memory and priced much higher.
Do some of you exclusively read about gaming gpu and nothing else?
KoldPurchase@reddit
I am aware of all that.
What I'm saying is they need to continue developping these products.
And they go in tandem, gaming and pro gpus. Might as well put a littke budget toward gaming gpus, that will give them visibility everywhere.
996forever@reddit
They probably figured even the Radeon Pro gpus are poor margin now because they’re memory focused. R9700 is 1300 with 32GB ram.Â
AMD pro cards are always priced much lower than equivalent Nvidia cards. A much bigger gulf than between their gaming gpus even.
Deeppurp@reddit
Heres where you kind of fall apart:
You want the consumer mind share before they enter the professional work space. So the low margin market matters because you want the CEO/CFO's children and potential new recruits all talking about how good AMD is, reinforcing their own sales team but for free.
The "low margin" market is literally advertising for AMD by hobbyists and for people actually starting to get serious about tech. The mind share matters because the interest matters.
It stops mattering once they start getting educated for those that enter the profession, but getting the attention however.
0101010001001011@reddit
Honestly this is a reasonable take, while I think RDNA5 architecture is going to be amazing, actual supply is competing against server and console demands. Unfortunately it may be high price low stock...
SirActionhaHAA@reddit
Diy gpus have some of the lowest margins out of all products at amd, probably as bad a consumer mobile (not enterprise). The total dgpu + console is just $700mil this quarter. Dgpu's probably like <5% revenue at amd.
railven@reddit
Gamers think NV is abandoning gaming!
Gamers have a hard time applying reality to what their favorite streamer feeds them.
varateshh@reddit
This was true before AI as well!
AMD had higher margins on CPUs (enterprise, server and consumer) and enterprise GPUs. They have been limited by foundry capacity ever since they ditched GlobalFoundries in 2018-2019. Ryzen 3000 was inferior to Intel and that gave them more spare capacity to RX5000 Navi, but with Ryzen 5000 they would rather sell consumer CPUs than more Radeons as the demand was solid.
railven@reddit
Exactly!
If anyone is surprised by any of this - get of Reddit, because you clearly aren't learning anything reading these topics.
I'm brought back to that one sad poster posting he was surprised RDNA4 wasn't on Steam Charts on it's first month of release due to the glazing HUB gave.
By the 3rd month of no RDNA4, I got the feeling these people finally connected all the dots but nope HUB put out a new tweet (last month) that AMD is crushing NV. Some where. Some how.
Geddagod@reddit
Maybe this makes sense today with the AI boom, but this question has been being asked for generations now, and AMD didn't have that excuse back then.
This reasoning does not cut it for why arc dgpus are allegedly canned.
Sevastous-of-Caria@reddit
Conveniently next generation of arc graphics architecture are still coming to... Enterprise server gpus.....
Geddagod@reddit
On a node that they massively overbuilt, and with rumors pointing to 18A capacity being pushed back or reduced even further than what was originally planned (DMR launch delay, 4+8 NVL compute tile going from 18A to N2). Don't forget that Wildcat Lake 18A shipments are also likely going to be lower than originally projected too because of increasing costs at the low end of client hurting sales.
They might have canned client stuff because the products themselves were too low margin, but Intel really should have no problem with capacity internally on 18A for their own internal stuff. If they can stomach ramping a node aggressively that has relatively low yields and for stuff that has relatively low margins that is... which it sounds like they don't want to do. Also unlike AMD, where they likely realistically can't get more wafer allocation even if they wanted to.
Minced-Juice@reddit
What a bunch of crap.
"Rumors pointing to 18A being pushed back" - multiple statements on 18A ramping throughout 2026 to support new products in H2.
"4+8 NVL compute tile going from 18A to N2" - lol.
"Don't forget that Wildcat Lake 18A shipments are also likely going to be lower than originally projected too because of increasing costs at the low end of client hurting sales." - likely == vibes.
Pimpmuckl@reddit
And multiple job postings at sites with planned sites pointing to an acceleration of their schedule, not a delay.
And they bought out the JV in the fab in Ireland.
That's the one thing Pat really got right, the all-in on the nodes is finally looking a lot better now.
gorion@reddit
Wow, I expected that servers are about 8x bigger than Gaming, but i didn't realized that even Embedded Segment is also bigger.
R-ten-K@reddit
This sub tends to significantly overestimate both the scale and the impact of gaming within the broader tech industry (esp revenue wise).
Whirblewind@reddit
I very much doubt that they do, because I almost never see that happen and very frequently see the wolf cry posts like your own.
996forever@reddit
They don’t need to explicitly state the sentence, their self importance in everything they post and the constantly asking of “why company no focus on gaming only server” speak ten times louder than any explicit wordings ever could.Â
ProfessionalPrincipa@reddit
Maybe so but the guy making that assertion is quite the pompous poster themselves when making their opinions known.
R-ten-K@reddit
...but enough about yourself.
SirActionhaHAA@reddit
Gamers just don't got the $ to compete with ai, that's all it is.
Seanspeed@reddit
Neither does OpenAI, really. They're basically just beggars and dupe investors into giving them tens of billions.
SirActionhaHAA@reddit
Not the only customer of amd. Epyc is the #1 choice in dc rn, everyone is goin for it. Mi450 is getting huge momentum according to the q&a today
Pimpmuckl@reddit
You could have written this exact sentence two years ago and replaced openai with Intel Foundry.
And we see it with anthropic today already how much compute they need because of large scale deployments of Claude code and similar products.
OpenAI has absurd user growth for their codex product, they must be making bank on inference with the token costs they charge for 5.5 right now.
Not saying these companies aren't burning cash, but any industry where you need a ton of infrastructure to make things work well have upfront capex requirements.
And the use case got a lot more positive for them in the last few months given just how many users are being added to their coding tools
ResponsibleJudge3172@reddit
Waiting for the monthly XX company doesn't care about gaming that people love to make videos about
kiziadamonll@reddit
Does embedded include FPGA? If so, no shit xilinx is bigger than gaming. 1k gpus for consumers isn't going to compete with 50k FPGAs for the military industrial complex and scientific researcherers
b84ui@reddit
Embedded is mostly Xilinx. AMD should be worried if that was lower than Gaming after they bought Xilinx out for $50 bil a few years back, especially since Xilinx was still by far the largest FPGA company. Game consoles and consumer GPUs are small time compared to how much FPGAs are used across many industries that can't justify designing an ASIC.
Geddagod@reddit
Sounds like AMD is going to continue to gain share in the DC CPU space, despite the CPU shortage helping out Intel here.
Also interesting that AMD and Intel both deny raising ASPs significantly, just in relation to increasing core counts and offsetting cost increases.
AMD also seems to have made a pretty strong gain in client revenue share YoY. Not too surprising ig given Intel's position in desktop, and it seems like PTL ramp is still pretty slow. Intel moving wafers away from client to DC couldn't have helped either.
Mercury research is going to release their market share estimates in a week or so, looking forward to seeing that.
imaginary_num6er@reddit
PTL ramp is still pretty slow? No, it's just significantly more expensive than prior gen AMD CPUs.
996forever@reddit
Do we have a concrete source on this?Â
The framework 13 pro offers both Panther and Strix point concurrently and the Strix point variant is more expensive. You can’t compare prices of old AMD model vs brand new Panther model. You have to compare concurrent models.
1800lampshade@reddit
AMD is slaughtering in the datacenter CPU space. It's hard to find a peer of mine still buying Intel's for the datacenter.
996forever@reddit
But their revenue share is still at below 50% and unit share below 30%. They’re gaining fast but by numerical fact they’re not the majority yet.Â
SirActionhaHAA@reddit
"In server, we delivered our fourth consecutive quarter of record server CPU revenue. Revenue increased more than 50% year over year, with sales to both cloud and enterprise customers each growing more than 50%. Share gains accelerated year over year, reflecting the ramp of fifth gen Epyc turned CPUs and continued strength of fourth gen Epyc processors across a wide range of workloads.
As a result, we now expect server CPU revenue to grow by more than 70% year over year in the second quarter, with robust growth continuing through the second half of 2026 and into 2027."
Dc cpu grew >50% this quarter, expected to grow >70% next quarter.
ProZoid_10@reddit
Nooticing teletubers aren’t making “amd doesn’t care about gaming content when dc revenue is high vs gaming”
dabocx@reddit
The earnings call was incredibly bullish from Lisa, she is usually pretty conservative on saying anything but it really sounds like Helios and Datacenter in general is going to be insane sales wise.