Intel has lost all of its dedicated GPU market share

Daniel Sims

Daniel Sims

Posts: 1,878   +49
Staff
In a nutshell: The two main players in the dedicated desktop graphics card market, Nvidia and AMD, enjoyed a positive quarter. Unfortunately, Intel failed to latch onto the upswing as it still struggles to break into the sector. The lack of movement in Intel GPUs certainly isn't helping as the company faces troubles elsewhere, but the next round of the AIB race is set to begin late this year or early next year.

According to a report from Jon Peddie Research, dedicated desktop GPU sales rose significantly in the second quarter of 2024 for everyone except Intel. Although the market is up overall, share didn't change from last quarter.

Shipments of AIBs reached approximately 9.5 million units, an increase of 9.4 percent from Q1 2024 and 48 percent from Q2 2023. The quarterly jump is well above average, indicating a strong recovery.

Market king Nvidia benefited the most, seeing a 9.7 percent quarterly rise and a 61.9 percent annual uplift. Furthermore, the company maintained its 88 percent market share from last quarter.

Meanwhile, although AMD's shipments only increased by three percent year-over-year, its quarterly improvement was similar to Nvidia's. Team Red also held onto its 12 percent market share from Q1.

Intel, though, has effectively been shut out, with sales remaining flat compared to last quarter. After entering the AIB market with its Arc Alchemist series in 2022, the company briefly achieved a four percent market share toward the end of that year, but it had evaporated by early 2024.

Desktop CPU-to-AIB attachment rate since 2016

Alchemist was supposed to provide an alternative to competing mid-range graphics cards. However, shipment delays put it behind the price-to-performance curve compared to Nvidia Ampere, Ada Lovelace, and AMD's RDNA 3 chips. The initial drivers also dragged performance significantly downward, though they have since improved dramatically.

All three AIB manufacturers are preparing new generations of GPUs. Nvidia's Blackwell architecture, likely to be named RTX 5000, is expected to debut at CES in January 2025 with high-end and enthusiast products likely costing over $1,000. AMD might introduce RDNA 4 (Radeon RX 8000) around the same time while focusing on much cheaper parts that could significantly improve on the latest available mid-range GPUs.

However, the latest reports regarding Intel's upcoming Arc Battlemage lineup indicate that the company is still targeting the 2024 holiday season. It remains unclear whether Team Blue plans to step into the enthusiast tier where Nvidia currently has no competition, or launch another stack of mainstream cards. Either way, if Intel can avoid Alchemist's delays, it might have some of the best GPUs on the market for a few weeks.

Permalink to story:

Intel has lost all of its dedicated GPU market share

 
It remains unclear whether Team Blue plans to step into the enthusiast tier…I’d say it’s pretty clear they can’t. Maybe in five years, unless they’re not bought by Qualcomm in the meantime.
 
But, but, but, Intel offers cards in the low and mid price prackets! And EVERYONE wants affordable cards! That's where the market is!

Almost like not having high end cards in the lineup tanks a brand's marketability. Hmmmm.....
KorvaxChris said:
Likely a big reason why Qualcomm is attempting to swoop in with a buyout. Buy low.
Click to expand...
I bought it at $18.90 during the dip. give it a few years I'll bet it prints a mint.
 
  • Like
Reactions: KorvaxChris and emmzo
It’s going to take a long time for Intel to get to 10% marketshare at this rate. If they want to speed that up they’ll have to cut prices even more. Even being 20-30% better for the money just won’t cut it in a market as mature as this. If you want a solid GPU ecosystem, marketshare is everything. Just look at CUDA.
 
Theinsanegamer said:
But, but, but, Intel offers cards in the low and mid price prackets! And EVERYONE wants affordable cards! That's where the market is!

Almost like not having high end cards in the lineup tanks a brand's marketability. Hmmmm.....
I bought it at $18.90 during the dip. give it a few years I'll bet it prints a mint.
Click to expand...
Yep, AMD is also toasted.
 
Unlike the other commenters I don't think a lack of a high end product kills the brand.
If you have to spend a good chunk of R&D on making a card that competes with NVIDIA (I don't see AMD or Intel exceeding NVIDIA's performance by a large margin anytime soon) you lose a lot of money on a product no one buys. AMD couldn't do it when they had the better products ages ago, and nowadays people seem to be all about that DLSS and FrameGen
(both techniques that are méh to me personally at best).

Just be the peoples champion, with a solid card targeting up to 2560x1600 and you got almost 90% of the market in your pocket (89.37% going by the latest Steam Survey). AMD is claiming that they're going after this market but let's face it, they'll release something that's about -5%-+5% worse or better than what NVIDIA has for about 10-15% less money and their sales will amount to almost nothing.

Previous results showed AMD customers are much more likely to try an Intel card than NVIDIA buyers. Undercut AMD by 25% and they might just sell enough to get some decent market share, this time they don't need to be as hamstrung by terrible drivers either. To Intel's credit they have done far more with the drivers than I ever thought they would and actually stuck to it. If they can sell enough cards with the next gen they might just be able to convince the higher ups that there's place for them in the GPU market.

It's kinda funny Qualcomm (and by extension their Adreno graphics branch) is eying up Intel. Adreno was (possibly still is in parts) based on licensed Radeon technology.
 
Last edited:
opckieran said:
It’s going to take a long time for Intel to get to 10% marketshare at this rate. If they want to speed that up they’ll have to cut prices even more. Even being 20-30% better for the money just won’t cut it in a market as mature as this. If you want a solid GPU ecosystem, marketshare is everything. Just look at CUDA.
Click to expand...
I think they need to compete at the high-end. If people see them being competitive with 7900 or 4080 etc then they will assume they are also good in the mid-range and low-end. It's analogous to car companies spending a fortune on racing like Formula 1 just so they can sell some consumer vehicles. Right now, Intel is not compelling enough for people to risk jumping to their GPUs.
 
I really think that this card (which performs similarly to the 1080Ti) would sell like hotcakes today if it just doubled the GDDR6 VRAM to 36GB. It would only cost an extra $30 (that was the price of 16GB of GDDR6 a year ago), which is a small price to pay for the ability to interfere with 70B llms at ~4t/s at 3q. I guess the question on everyone's mind is why Intel made this rookie mistake?

It's a bit of a shame, but it looks like they will make again the same mistake with the Battlemage. Instead of going for at least 256GB of vram, they're probably going for 48GB. I'm sure they think that's more than enough for graphics the other cards have lower, but these aren't just graphics cards anymore. They're now AI interference cards. You can play any game at 60Hz at 1080p with just a 3060. Why to buy Intel card instead of Nvidia which has and cuda? What’s the competitive advantage?
 
  • Like
Reactions: Knot Schure
unoficialoficial said:
I really think that this card (which performs similarly to the 1080Ti) would sell like hotcakes today if it just doubled the GDDR6 VRAM to 36GB. It would only cost an extra $30 (that was the price of 16GB of GDDR6 a year ago), which is a small price to pay for the ability to interfere with 70B llms at ~4t/s at 3q. I guess the question on everyone's mind is why Intel made this rookie mistake?

It's a bit of a shame, but it looks like they will make again the same mistake with the Battlemage. Instead of going for at least 256GB of vram, they're probably going for 48GB. I'm sure they think that's more than enough for graphics the other cards have lower, but these aren't just graphics cards anymore. They're now AI interference cards. You can play any game at 60Hz at 1080p with just a 3060. Why to buy Intel card instead of Nvidia which has and cuda? What’s the competitive advantage?
Click to expand...
^ THIS THIS THIS....
I was chatting to a well-known chatbot recently and posed the question of why Intel with their 'sub standard' GPU architecture didn't load one up with a ton of VRAM? It would fly off the shelf.

And disrupt the market. "I was really onto something here..." the chatbot told me.

Clearly Intel's marketing department must only use their keyboards for sleeping on.
 
Axeia said:
Unlike the other commenters I don't think a lack of a high end product kills the brand.
If you have to spend a good chunk of R&D on making a card that competes (I don't see anything actually exceeding by a big margin) you lose a lot of money on that and no one buys the actual product (AMD couldn't do it when they had the better products ages ago, and nowadays people seem to be all about that DLSS and FrameGen (both techniques that are méh to me personally at best).

Just be the peoples champion, with a solid card targeting up to 2560x1600 and you got almost 90% of the market in your pocket (89.37% going by the latest Steam Survey). AMD is claiming that they're going after this market but let's face it, they'll release something that's about -5%-+5% worse or better than what NVIDIA has for about 10-15% less and still almost nothing.
Previous results showed AMD customers are much more likely to try an Intel card than NVIDIA buyers. Undercut AMD by 25% and they might just sell enough to get some decent marketshare, this time they don't need to be as hamstrung by terrible drivers either. To Intel's credit they have done far more with the drivers than I ever thought they would and actually stuck to it. If they can sell enough cards with the next gen they might just be able to convince the higher ups that there's place for them in the GPU market.

It's kinda funny Qualcomm (and by extension their Adreno graphics branch) is eying up Intel. Adreno was (possibly still is in parts) based on licensed Radeon technology.
Click to expand...
Yeah, except.....intel had that when the A770 launched. And it didnt light the world on fire. AMD constantly lost marketshare once they stopped making high end cards then, right as things began to turn around, announced they were going to stop again.

There's a pattern here.
 
I was really hoping Intel would establish a foothold in the GPU market and create much needed competition, but their product offerings aren't up to par for the needs of the standard consumer when for a few dollars more you can buy from AMD or NVIDIA that outperforms in every area.

I wish I could feel sorry for Intel, but they've done it to themselves and milked the CPU market for everything it was worth while on top.
 
Theinsanegamer said:
Yeah, except.....intel had that when the A770 launched. And it didnt light the world on fire. AMD constantly lost marketshare once they stopped making high end cards then, right as things began to turn around, announced they were going to stop again.

There's a pattern here.
Click to expand...
Intel did not have that at launch. What they had were cards that had a far higher power usage and incredibly flaky performance. People don't tend to purchase a card because it works great in a few games and is unplayable in the majority. It was (and is) barely if at all cheaper than AMD's RX 6600 XT but with far worse drivers.
Even now after massive driver improvements they're still far behind on AMD/NVIDIA.

AMD imo started losing market share due to people that kept claiming their drivers were bad* and NVIDIA playing dirty (NVIDIA Gameworks tech leading to insane amount of Tesselation in Crysis or Hairworks in The Witcher 3 absolutely tanking performance). Now they're down in the ditch because they don't have CUDA or DLSS and their counterparts fall short. They have had periods where they had the better card and still lost in sales big time (The Radeon 7970 was far superior to the GT 580 - and NVIDIA still outsold them by at least three to one).

*I've had cards from both sides for many years now and despite reading for decades by now that AMD (Radeons) drivers are supposedly bad they've always been about the same for me. I've had problems with both brands over the years not more with one than the other.
Even right now with a NVIDIA RTX 3060 12GB and Radeon 6700 XT both PCs have smalll annoyances. The NVIDIA PC has videos stuttering whilst gaming if Instant Replay is enabled, the AMD PC has high idle power draw if the dual monitor setup runs at different refreshrates.

Heck if anything, I much prefer AMDs drivers nowadays because Adrenalin gives a lot to fiddle with to undervolt/overclock and no forced login is a huge bonus. (NVIDIAs Beta App is looking promising, that might level the playing field for RTX 5000 vs AMD Radeon RX 8000 and AMD will lose one of its very few perks)

--

My simple take on things for AMD/Intel to gain marketshare:

AMD needs to avoid the mistake THEY ALWAYS make - launching with high prices that quickly drop. That completely ruins launch reviews for them, they're judged by the launch price and when it's nearly the same as NVIDIA's nearly indentically performing product with more features. The review conclusions aren't positive (enough) and they lose out on sales. Launch at the prices they expect to sell for 2-3 months after launch and preferably even lower. LTT, Gamers Nexus, Hardware Unboxed and various techsites could finally give AMD a "recommended buy" launch for the first time in ages if they have solid performance up to 2560x1600 performance @ ~$/€/£300 (350 max), a solid 1080p 120Hz $200 card would be most welcome as well)
Marketshare is more important than fat profit margins since they've been reduced to such a small part of the market. Developers will stop caring to test on their cards if their marketshare falls further.

Intel needs to sell for prices as low as they can and if they can manage strike a deal with some popular game to launch with. If they can show they're the best bang-a-buck for a modern freshly launched game it will help restore faith in them. They need marketshare for a different reason - they're in no position to try and make much of a profit and if they try to chase fat margins they won't sell and Intel in its financial distress will definitely pull the plug.
 
Axeia said:
Intel did not have that at launch. What they had were cards that had a far higher power usage and incredibly flaky performance. People don't tend to purchase a card because it works great in a few games and is unplayable in the majority. It was (and is) barely if at all cheaper than AMD's RX 6600 XT but with far worse drivers.
Even now after massive driver improvements they're still far behind on AMD/NVIDIA.

AMD imo started losing market share due to people that kept claiming their drivers were bad* and NVIDIA playing dirty (NVIDIA Gameworks tech leading to insane amount of Tesselation in Crysis or Hairworks in The Witcher 3 absolutely tanking performance). Now they're down in the ditch because they don't have CUDA or DLSS and their counterparts fall short. They have had periods where they had the better card and still lost in sales big time (The Radeon 7970 was far superior to the GT 580 - and NVIDIA still outsold them by at least three to one).

*I've had cards from both sides for many years now and despite reading for decades by now that AMD (Radeons) drivers are supposedly bad they've always been about the same for me. I've had problems with both brands over the years not more with one than the other.
Even right now with a NVIDIA RTX 3060 12GB and Radeon 6700 XT both PCs have smalll annoyances. The NVIDIA PC has videos stuttering whilst gaming if Instant Replay is enabled, the AMD PC has high idle power draw if the dual monitor setup runs at different refreshrates.

Heck if anything, I much prefer AMDs drivers nowadays because Adrenalin gives a lot to fiddle with to undervolt/overclock and no forced login is a huge bonus. (NVIDIAs Beta App is looking promising, that might level the playing field for RTX 5000 vs AMD Radeon RX 8000 and AMD will lose one of its very few perks)

--

My simple take on things for AMD/Intel to gain marketshare:

AMD needs to avoid the mistake THEY ALWAYS make - launching with high prices that quickly drop. That completely ruins launch reviews for them, they're judged by the launch price and when it's nearly the same as NVIDIA's nearly indentically performing product with more features. The review conclusions aren't positive (enough) and they lose out on sales. Launch at the prices they expect to sell for 2-3 months after launch and preferably even lower. LTT, Gamers Nexus, Hardware Unboxed and various techsites could finally give AMD a "recommended buy" launch for the first time in ages if they have solid performance up to 2560x1600 performance @ ~$/€/£300 (350 max), a solid 1080p 120Hz $200 card would be most welcome as well)
Marketshare is more important than fat profit margins since they've been reduced to such a small part of the market. Developers will stop caring to test on their cards if their marketshare falls further.

Intel needs to sell for prices as low as they can and if they can manage strike a deal with some popular game to launch with. If they can show they're the best bang-a-buck for a modern freshly launched game it will help restore faith in them. They need marketshare for a different reason - they're in no position to try and make much of a profit and if they try to chase fat margins they won't sell and Intel in its financial distress will definitely pull the plug.
Click to expand...

I think AMD's problem is not about launching new products at high prices. I think that is common practice because you will want to quickly cover as much of the R&D cost in the shortest amount of time. And its new, so people may be willing to pay a premium for it. You can't run business by starting off with the lowest possible price everytime. You can sell, but you won't make money, and so kind of moot.

To me, AMD's biggest problem (which likely also applies to every big company) is they tend to slack off quickly and fall behind competition.
 
  • Like
Reactions: Bamda
Don't worry Battlemage is coming and can match a 4060 and is a bit faster than the 7600XT. That'll ignite sales.
 
Watzupken said:
I think AMD's problem is not about launching new products at high prices. I think that is common practice because you will want to quickly cover as much of the R&D cost in the shortest amount of time. And its new, so people may be willing to pay a premium for it. You can't run business by starting off with the lowest possible price everytime. You can sell, but you won't make money, and so kind of moot.

To me, AMD's biggest problem (which likely also applies to every big company) is they tend to slack off quickly and fall behind competition.
Click to expand...
Pulling the numbers out of of my behind, but it feels like AMD every single time does this:
Ask for 20-30% to much margin and get bad initial sales due to the bad reviews this leads to. 2 weeks to three months later (of disappointing sales) lower the prices. During this short initial period they sell probably like five percent of what they'll sell in total.
However if they had launched at lower prices they would have made up the price difference at the start already from better reviews and sales. The better reviews lead to a better image and sales and their total sales might be 50% higher.

Of course this only works if they're not supply constrained (it might be smart to consider Samsung for production instead of the more coveted TSMC nodes)
 
  • Like
Reactions: guillejp
I'm not surprised that nVidia has 88% of the GPU market share, what I find surprising is that there are 12% of the market choosing the less capable cards that AMD puts out.
 
  • Like
Reactions: guillejp
Axeia said:
Intel did not have that at launch. What they had were cards that had a far higher power usage and incredibly flaky performance. People don't tend to purchase a card because it works great in a few games and is unplayable in the majority. It was (and is) barely if at all cheaper than AMD's RX 6600 XT but with far worse drivers.
Even now after massive driver improvements they're still far behind on AMD/NVIDIA.

AMD imo started losing market share due to people that kept claiming their drivers were bad* and NVIDIA playing dirty (NVIDIA Gameworks tech leading to insane amount of Tesselation in Crysis or Hairworks in The Witcher 3 absolutely tanking performance). Now they're down in the ditch because they don't have CUDA or DLSS and their counterparts fall short. They have had periods where they had the better card and still lost in sales big time (The Radeon 7970 was far superior to the GT 580 - and NVIDIA still outsold them by at least three to one).

*I've had cards from both sides for many years now and despite reading for decades by now that AMD (Radeons) drivers are supposedly bad they've always been about the same for me. I've had problems with both brands over the years not more with one than the other.
Even right now with a NVIDIA RTX 3060 12GB and Radeon 6700 XT both PCs have smalll annoyances. The NVIDIA PC has videos stuttering whilst gaming if Instant Replay is enabled, the AMD PC has high idle power draw if the dual monitor setup runs at different refreshrates.

Heck if anything, I much prefer AMDs drivers nowadays because Adrenalin gives a lot to fiddle with to undervolt/overclock and no forced login is a huge bonus. (NVIDIAs Beta App is looking promising, that might level the playing field for RTX 5000 vs AMD Radeon RX 8000 and AMD will lose one of its very few perks)

--

My simple take on things for AMD/Intel to gain marketshare:

AMD needs to avoid the mistake THEY ALWAYS make - launching with high prices that quickly drop. That completely ruins launch reviews for them, they're judged by the launch price and when it's nearly the same as NVIDIA's nearly indentically performing product with more features. The review conclusions aren't positive (enough) and they lose out on sales. Launch at the prices they expect to sell for 2-3 months after launch and preferably even lower. LTT, Gamers Nexus, Hardware Unboxed and various techsites could finally give AMD a "recommended buy" launch for the first time in ages if they have solid performance up to 2560x1600 performance @ ~$/€/£300 (350 max), a solid 1080p 120Hz $200 card would be most welcome as well)
Marketshare is more important than fat profit margins since they've been reduced to such a small part of the market. Developers will stop caring to test on their cards if their marketshare falls further.

Intel needs to sell for prices as low as they can and if they can manage strike a deal with some popular game to launch with. If they can show they're the best bang-a-buck for a modern freshly launched game it will help restore faith in them. They need marketshare for a different reason - they're in no position to try and make much of a profit and if they try to chase fat margins they won't sell and Intel in its financial distress will definitely pull the plug.
Click to expand...

No, AMD is hurt by their abysmal software. Their drivers are just as bad as Intel's and have NEVER not been that way. They absolutely ruined ATI when they bought it, and they've been limping it along ever since. I work in IT and have for over 20 years. 99.9% of the time, the issue is AMD's software/firmware. It is not stable. Even if they somehow perform better from time to time, you can't rely on their stability. Any of my customers that use AMD get switched over or they get dropped as a customer. I refuse to deal with AMD's incompetence.
 
Axeia said:
Unlike the other commenters I don't think a lack of a high end product kills the brand.
Click to expand...

I agree with you. Polaris sold well for AMD. The problem with Intel's cards isn't positioning but that they offer worse value for money than the competition. Hardware Unboxed's tests of God of War and Warhammer show the A770 to be 25% slower than a 6600, which is abysmal performance. I imagine it's lack of driver optimisation, but that's the problem with Intel cards, drivers still aren't as good as other brands. With Intel not having a performance/price advantage, there's no reason to buy it.

Intel margins are already low enough that it probably can't lower prices more without losing money. It's a large chip (twice the size of Navi 33 on the same process, and with lower performance). I'm hoping for Intel that Battlemage is more competitive.
 
  • Like
Reactions: Dreadcthulhu
No wonder, it has not been a focus at all. They have been improving drivers and hardware.

Intel actually did very well for a 1st gen gaming GPU tho. Performance gains with software alone has been huge.

Xe2 looks very good.

Battlemage is going to compete in low to mid-end market just fine, if price is right. AMD should be worried, because this is their prime segment now. They left high-end.

Nvidia owns high-end but also dominate in mid-end. Nvidia cares very little about low-end.
 
You must log in or register to reply here.

Similar threads

Daniel Sims
Global server market revenue nearly doubled in 2024 with Nvidia's continued dominance
Replies
1
Views
135
daffy duck
D
S
Google researchers uncover critical security flaw in all AMD Zen processors
Replies
6
Views
144
Joeg1484
J
S
Consumers race to Apple stores as tariff fears spark iPhone rush
Replies
3
Views
131
scoffer
S

Latest posts

Back