Friday, January 23rd 2026

Intel Core Ultra Series 4 "Nova Lake" Processors Launch in Late-2026

Intel is preparing to launch its Core Ultra Series 4 "Nova Lake" processor generation toward the end of 2026, CEO Lip-Bu Tan confirmed. This processor generation would span various form-factors, including importantly, desktop. Intel is building its next-generation "Nova Lake-S" desktop processor around the new Socket LGA1954 platform that replaces the current LGA1851. The company is planning to release Core Ultra Series 3 "Arrow Lake Refresh" for LGA1851 in the run-up to "Nova Lake-S" by yearend.

With "Nova Lake," Intel is expected to make fundamental changes to its processor design to help the company claw back performance leadership over AMD in the desktop segment. These changes include a doubling in CPU core counts going up to 16P+32E, a new iGPU based on the Xe3 "Celestial" graphics architecture, a new NPU that meets Copilot+ requirements, and the introduction of special SKUs with BLLC (big last-level cache), Intel's answer to AMD's 3D V-Cache. Intel will update the CPU complex with new "Coyote Cove" P-cores, and "Arctic Wolf" E-cores, both of which are expected to improve IPC over the current "Lion Cove" and "Skymont" combo. "Coyote Cove" does not feature HyperThreading, something the Intel CEO wants to see restored on the company's P-cores, but "Nova Lake" will make up for it with a numerical increase in P-core counts.
Source: VideoCardz
Add your own comment

62 Comments on Intel Core Ultra Series 4 "Nova Lake" Processors Launch in Late-2026

#1
SL2
I'm curious what these will be capable of. Let's hope there are other components available to pair them with at launch, like GPU and RAM.. :rolleyes:
Posted on Reply
#2
Upgrayedd
32 e-cores?? Give me like 8e, delete the iGPU entirely from the design and use the rest of the space for more P-cores or cache. Why are they even playing around with the market segments AMD has taken from them?

Who at Intel is going. "yeah normies need 64 threads and half of them need to be weaker than the other half". Anyone who actually needs that is just going TR or Xeon. I understand this is a cheaper alternative for some people than TR or Xeon, but the mass majority that will get something like this will NEVER need 32e cores.
Posted on Reply
#3
rattlehead99
Upgrayedd32 e-cores?? Give me like 8e, delete the iGPU entirely from the design and use the rest of the space for more P-cores or cache. Why are they even playing around with the market segments AMD has taken from them?

Who at Intel is going. "yeah normies need 64 threads and half of them need to be weaker than the other half". Anyone who actually needs that is just going TR or Xeon. I understand this is a cheaper alternative for some people than TR or Xeon, but the mass majority that will get something like this will NEVER need 32e cores.
What benefit do you get from having more than 8P cores that is not a productive/professional workload?
Also their new E-Cores are on par with the last gen P-Cores in terms of IPC, and the clocks seem to be going up as well and now E-Cores will have the same instructions as the P-Cores.
If anything the P-Cores seem to be there just to top benchmark charts.
Posted on Reply
#4
dgianstefani
TPU Proofreader
Upgrayedd32 e-cores?? Give me like 8e, delete the iGPU entirely from the design and use the rest of the space for more P-cores or cache. Why are they even playing around with the market segments AMD has taken from them?

Who at Intel is going. "yeah normies need 64 threads and half of them need to be weaker than the other half". Anyone who actually needs that is just going TR or Xeon. I understand this is a cheaper alternative for some people than TR or Xeon, but the mass majority that will get something like this will NEVER need 32e cores.
16+32 = 48, not 64.

The IGPU of modern Intel CPUs is also incredibly useful, and the E cores add more MT more efficiently than a quarter of the number of P cores would.
Posted on Reply
#5
SL2
rattlehead99Also their new E-Cores are on par with the last gen P-Cores in terms of IPC
You're right, they're not Economy Cores™ anymore AFAIK.

I just think TPU members prefer big cores, for various reasons based on past experiences from Raptor or older, and not the latest E-cores.
dgianstefani16+32 = 48, not 64.
Also, four low power cores on top of that.
Posted on Reply
#6
dgianstefani
TPU Proofreader
SL2You're right, they're not Economy Cores™ anymore AFAIK.

I just think TPU members prefer big cores, for various reasons based on past experiences from Raptor or older, and not the latest E-cores.


Also, four low power cores on top of that.
Yeah I was wondering if the 32 included those or not. Wouldn't have been surprised to hear the 32 die was binned down to 28.

I hope they revert the course of adding HT back tbh. At these core counts it doesn't seem necessary, and I'd take a more secure P core with 5% more throughput and 5% lower power draw than a second thread only adding 30% of the MT best case scenario.
Posted on Reply
#7
ThomasK



Intel is essentially Amd from the past. The only difference is Amd didn't get a bailout from government.

But bigger numbers do look better on PR material.
Posted on Reply
#8
dgianstefani
TPU Proofreader
ThomasK


Intel is essentially Amd from the past. The only difference is Amd didn't get a bailout from government.

But bigger numbers do look better on PR material.
FWIW AMD is also supposedly increasing core counts by 50% this time around. 8-12 CCD.
Posted on Reply
#10
rattlehead99
SL2You're right, they're not Economy Cores™ anymore AFAIK.

I just think TPU members prefer big cores, for various reasons based on past experiences from Raptor or older, and not the latest E-cores.


Also, four low power cores on top of that.
Yes, Skymont(the E-Cores in Arrow Lake and Lunar Lake) have the same IPC as the P-Cores in RaptorLake, but lacked AVX, Darkmont(300 series E-Cores) now has complete feature parity with Lion Cove(300 Series P-Cores), while also having higher clock speeds than its predecessor, much closer to the P-Cores and the IPC will likely be on par or higher than Arrow-Lake P-Cores as intel have stated that the goal of modern E-Cores is to have IPC/PPC parity with the last gen P-Cores or higher.

Also the modern E-Cores are also OoOE like the P-Cores, while the old ones iirc weren't.

The LPE Cores are the old-school E-Cores.
Posted on Reply
#11
melkor_unlimited
dgianstefani16+32 = 48, not 64.
Its not 64
And its not 16+32
Its (8+16) + (8+16) + 4
rattlehead99What benefit do you get from having more than 8P cores that is not a productive/professional workload?
With (8+16) + (8+16) + 4 you will not see big benefit where latency kicks in.
If 12 Zen cores in one chiplet will give any benefit over 8 ... we have to wait for it to come.
I wont bet on Intel.
Posted on Reply
#12
Simkin
ThomasK


Intel is essentially Amd from the past. The only difference is Amd didn't get a bailout from government.

But bigger numbers do look better on PR material.
Instead of this fanboy nonsense, competition is good, and Intel back on track is what we need.

You know, lower prices etc.
Posted on Reply
#13
xSneak
isn't this thing going to be kneecapped by dual channel ddr5 bandwidth?
Posted on Reply
#14
ThomasK
dgianstefaniFWIW AMD is also supposedly increasing core counts by 50% this time around. 8-12 CCD.
IKR.

At least Amd will give me +4 full "p-cores", which will fit into my existing motherboard. Same can't be said about LGA1851 and Nova Lake.
Posted on Reply
#15
rattlehead99
melkor_unlimitedIts not 64
And its not 16+32
Its (8+16) + (8+16) + 4


With (8+16) + (8+16) + 4 you will not see big benefit where latency kicks in.
If 12 Zen cores in one chiplet will give any benefit over 8 ... we have to wait for it to come.
I wont bet on Intel.
Outside of cache even a Zen4/5 6 core is as good as Zen4/5 8 core in games, both on a single module. For productivity, you have no downside of having the hybrid cores.
Posted on Reply
#16
Wirko
xSneakisn't this thing going to be kneecapped by dual channel ddr5 bandwidth?
Yes, the bottlenecks will just move elsewhere. The number of cores is following the megapixel race dark pattern.

Meanwhile, people will be buying DDR5-5600 memory because it will be a little less horribly priced than 6000.
Posted on Reply
#17
Thunderslat
rattlehead99Yes, Skymont(the E-Cores in Arrow Lake and Lunar Lake) have the same IPC as the P-Cores in RaptorLake, but lacked AVX, Darkmont(300 series E-Cores) now has complete feature parity with Lion Cove(300 Series P-Cores), while also having higher clock speeds than its predecessor, much closer to the P-Cores and the IPC will likely be on par or higher than Arrow-Lake P-Cores as intel have stated that the goal of modern E-Cores is to have IPC/PPC parity with the last gen P-Cores or higher.

Also the modern E-Cores are also OoOE like the P-Cores, while the old ones iirc weren't.

The LPE Cores are the old-school E-Cores.
Lol what? All E cores have been OoO. You would have to back to the original Atom to find something in order. I also don't see the new E cores matching an Arrow Lake P core but we shall see.
Posted on Reply
#18
rattlehead99
ThunderslatLol what? All E cores have been OoO. You would have to back to the original Atom to find something in order. I also don't see the new E cores matching an Arrow Lake P core but we shall see.
At the same clock speeds it will. Of course it will be clocked lower. Skymont at the same clock speeds in almost all tasks, considering that it also had less cache and didn't have all instructions, was pretty much around Raptor Lake P core performance.
Posted on Reply
#19
FeelinFroggy
ThomasKIntel is essentially Amd from the past. The only difference is Amd didn't get a bailout from government.

But bigger numbers do look better on PR material.
While I understand the sentiment, Intel did not get government money for making CPUs. They got it because of their fabs, something that AMD does not have and outsources to TSMC.

That's why AMD did not get any money while Intel and TSMC did. Because there are not enough semiconductor manufacturing sites in the US. I'm sure if AMD had fabs, they would have gotten paid too.
Posted on Reply
#20
ThomasK
FeelinFroggyI'm sure if AMD had fabs, they would have gotten paid too.
Not as if you cared enough to know or read about it, but Amd did have fabs, up until 2009, when it was spun off, to what now is GlobalFoundries. And back then, it was already bleeding cash and way behind Intel.

What saved it was getting rid of its fabs and a 1bn dollar settlement from Intel, for the x86 licences and anticompetitive practices lawsuit, not government/taxpayer money.

By the way, the consumer/taxpayer just got the middle finger from Intel:

www.techpowerup.com/345535/intel-reallocates-pc-production-capacity-to-server-cpus-amid-tight-wafer-supply
Posted on Reply
#21
LastDudeALive
dgianstefaniI hope they revert the course of adding HT back tbh. At these core counts it doesn't seem necessary, and I'd take a more secure P core with 5% more throughput and 5% lower power draw than a second thread only adding 30% of the MT best case scenario.
Agreed. When even the lowest tier CPUs and your lightest mobile CPUs get 8 cores, and you can put 52 cores on your flagship desktop CPUs, there's no need for multi-threading anymore.
xSneakisn't this thing going to be kneecapped by dual channel ddr5 bandwidth?
That's what CUDIMM is for.
Posted on Reply
#22
melkor_unlimited
rattlehead99Outside of cache even a Zen4/5 6 core is as good as Zen4/5 8 core in games, both on a single module. For productivity, you have no downside of having the hybrid cores.
I did not divide apps to games and non-games. Mentioned Latency.
Games are not only thing suffering more with bigger latency.

6 cores as good as 8 cores?
You may be right with old games made to use 1~4 cores.
But there are games utilising more than 6 cores.
And not of them are simulators like Factorio.
www.techspot.com/articles-info/3075/bench/HZD.png
www.techspot.com/articles-info/3075/bench/SM2.png
www.techspot.com/articles-info/3075/bench/ACC.png


Slow RAMs, 16GB RAM or less, 8GB VRAM or less, slow GPU ... and every other bottleneck makes L3 cache and number of cores less and less relevant.
Posted on Reply
#23
LastDudeALive
melkor_unlimitedI did not divide apps to games and non-games. Mentioned Latency.
Games are not only thing suffering more with bigger latency.

6 cores as good as 8 cores?
You may be right with old games made to use 1~4 cores.
But there are games utilising more than 6 cores.
And not of them are simulators like Factorio.
www.techspot.com/articles-info/3075/bench/HZD.png
www.techspot.com/articles-info/3075/bench/SM2.png
www.techspot.com/articles-info/3075/bench/ACC.png


Slow RAMs, 16GB RAM or less, 8GB VRAM or less, slow GPU ... and every other bottleneck makes L3 cache and number of cores less and less relevant.
Games just need to be hybrid-core aware. This CPU design has been around for 5 years already, and both vendors have it. If your game scales to a large number of cores, you need to be able to divide the workload between performance and efficient cores, or if the work can't be divided, only run it on the performance cores.
Posted on Reply
#24
igormp
rattlehead99Yes, Skymont(the E-Cores in Arrow Lake and Lunar Lake) have the same IPC as the P-Cores in RaptorLake, but lacked AVX
No, they had proper AVX2 support, and have had since Gracemont (Alder Lake). They lacked AVX-512, which is still not present in any current E-core offering. Lion cove did have AVX-512 at first but Intel disabled it due to the disparity with the E-cores.
rattlehead99Also the modern E-Cores are also OoOE like the P-Cores, while the old ones iirc weren't.
They have always had OOE since Silvermont in the Atom era.
rattlehead99The LPE Cores are the old-school E-Cores.
Not really. Since Meteor Lake the LPE and E-cores have the exact same µarch (Crestmont). The difference was the node they're fabbed at, with the E-cores usually on the compute tile along with P-cores on a better node while the LPE-cores get shunned to the SoC tile with worse node and slower clocks, and the lack of L3 on the LPE cores. The only exception to that was Arrow Lake mobile, which re-used the SoC from MTL, so the E-cores were Skymont but the LPE-cores were Crestmont
However, Panther Lake will share the µarch, with both E and LPE-cores being Darkmont and both living in the compute tile. The difference is just that the LPE ones will clock lower and won't have an L3.
I assume Nova Lake will be similar to MTL, using Arctic wolf cores for both E and LPE, but with the LPE-cores being shunned to the SoC tile with lower clocks and no L3.
Posted on Reply
#25
dyonoctis
ThomasKNot as if you cared enough to know or read about it, but Amd did have fabs, up until 2009, when it was spun off, to what now is GlobalFoundries. And back then, it was already bleeding cash and way behind Intel.

What saved it was getting rid of its fabs and a 1bn dollar settlement from Intel, for the x86 licences and anticompetitive practices lawsuit, not government/taxpayer money.

By the way, the consumer/taxpayer just got the middle finger from Intel:

www.techpowerup.com/345535/intel-reallocates-pc-production-capacity-to-server-cpus-amid-tight-wafer-supply
AMD's old spun-off foundries are irrelevant; they don't qualify them for getting money from the chip act, whose main purpose is to boost manufacturing sovereignty. Both the US and the EU made that act after seeing how bad relying on other faraway countries could be during the pandemic
Posted on Reply
Add your own comment
Nov 11th, 2026 17:20 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts