Intel details 18A process technology — takes on TSMC 2nm with 30% density gain and 25% faster generational performance

MEMBER EXCLUSIVE
Intel
(Image credit: Intel)

Intel has published a paper about its 18A (1.8nm-class) fabrication process at the VLSI 2025 symposium, consolidating all its information about the manufacturing technology into a single document. The new 18A production node is expected to deliver significant improvements in power, performance, and area over its predecessor, increasing density by 30% while enhancing performance by 25% or reducing power consumption by 36%.

But, perhaps more importantly, 18A will be Intel's first process technology in years that will compete head-to-head with TSMC's leading-edge technology when both enter mass production in the second half of this year.

PPA advantages

Swipe to scroll horizontally
How 18A stacks up against previous nodes
Row 0 - Cell 0

Intel 3 vs Intel 4

18A vs Intel 3

Power

?

36% (at 1.1V) - 38% (at 0.75V)

Performance

18% (?)

18% (at 0.75V) - 25% (1.1V)

Density

-

1.3X

SRAM Cell Size

0.024 µm²

0.021 µm²

Transistor

FinFET

RibbonFET GAA

Power Delivery

Front-side

PowerVia BSPDN

HVM

mid-2024

H2 2025

Latest Videos From
Swipe to scroll horizontally

Tom's Hardware

Intel 7

Intel 4

Intel 3

Intel 18A

Contacted Poly Pitch

54nm/60nm

50 nm

50 nm

50 nm

Fin Pitch

34 nm

30 nm

30 nm

?

M0 Pitch

40 nm

30 nm

30 nm

32 nm

High Performance Library Height

408 nm

240 nm

240 nm

180 nm

High Density Library Height

-

-

210 nm

160 nm

HP Library Height x CPP

24.4K nm²

12K nm²

12K nm²

9K nm²

HD Library Height x CPP

-

-

10.5K nm²

8K nm²

Anton Shilov
Contributing Writer

Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends.

  • TerryLaze
    Admin said:
    Intel has fully detailed its 18A (1.8nm-class) process at VLSI 2025, highlighting major performance, power, and density improvements enabled by RibbonFET transistors and PowerVia backside power delivery.

    Intel details 18A process technology — boosts performance by 25% or lowers power consumption by 36% : Read more
    They made an arm core on intel 3?!?!
    How is that not the news?
    So could they use intel 3 to make arm, and maybe other, cores for customers?!
    It manages to achieve this without increasing voltage or circuit complexity when running a typical Arm core sub-block, implemented using a 180CH HD library at 1.1. When operating at the same clocks and 1.1V voltage, it also cuts power usage by 36% compared to the same design on Intel 3. At a reduced voltage of 0.75V, 18A offers an 18% speed increase and uses 38% less energy. Furthermore, designs fabricated on 18A occupy roughly 28% less area than those built with Intel 3.
    Reply
  • S58_is_the_goat
    Let's all hope it's not delayed like 10mm was... we don't need any more 14nm++++++++ jokes.
    Reply
  • dalek1234
    Comparing 18A to Intel 3 node doesn't say much. Intel 3 is a slightly improved Intel 4 which used to be called 7nm which is probably the wort node Intel ever came up with.

    And if you pick the power-saving option in 18A vs the performance option, then you get less power usage and zero performance gain. Given that intel3/4 CPU's suck way to much power, sucking a 1/3 less (which is an improvement but not a good enough one) while not improving performance, is not looking good for Intel.
    Reply
  • EzzyB
    TerryLaze said:
    They made an arm core on intel 3?!?!
    How is that not the news?
    So could they use intel 3 to make arm, and maybe other, cores for customers?!
    Here's all I could find with a quick search. Apparently Intel was using ARM a LONG time ago (early 2000's). Doesn't prove much except that Intel has manufactured ARM chips in the past.

    https://en.wikipedia.org/wiki/XScale
    Reply
  • TerryLaze
    dalek1234 said:
    Comparing 18A to Intel 3 node doesn't say much. Intel 3 is a slightly improved Intel 4 which used to be called 7nm which is probably the wort node Intel ever came up with.

    And if you pick the power-saving option in 18A vs the performance option, then you get less power usage and zero performance gain. Given that intel3/4 CPU's suck way to much power, sucking a 1/3 less (which is an improvement but not a good enough one) while not improving performance, is not looking good for Intel.
    Draw a vertical line at 125w which is the base power of intel, the 14900k is at about 1700 points so 25% above that would put it above the 285k at that power.

    Being able to push stupid amounts of power into a cpu without it blowing up is extremely different from the CPU using that much power.

    https://www.computerbase.de/artikel/prozessoren/intel-core-ultra-200s-285k-265k-245k-test.90019/seite-5https://i.imgur.com/00HJj5p.jpg
    Reply
  • DS426
    SRAM cell density isn't even close to TSMC's 2nm. That would be a non-starter in some applications, would it not?

    Gains look decent comparing Apples to Apples, but what happens when customers compare Apples to Oranges (Intel to TSMC) and pricing is also a main consideration?
    Reply
  • thestryker
    TerryLaze said:
    They made an arm core on intel 3?!?!
    How is that not the news?
    So could they use intel 3 to make arm, and maybe other, cores for customers?!
    How is any of this news to you?

    Anything can be manufactured on any node so long as the chip is designed for it. In this case it's likely a very basic controller type test chip they use for testing nodes. I'd be surprised if this hadn't been a standard for many years given how Arm scales down so well.
    Reply
  • thestryker
    dalek1234 said:
    Given that intel3/4 CPU's suck way to much power,
    Care to back this up with any evidence?
    dalek1234 said:
    which is probably the wort node Intel ever came up with.
    Based on what exactly? Your feelings?
    Reply
  • thestryker
    Part of me wonders if the lack of higher voltage support is why there aren't any desktop CPUs planned with PTL. If this is the case then perhaps that's where the usage of N2 comes in for NVL. This may also be the performance part they're referring to when talking about 18A-P.

    The only client part currently being manufactured with Intel 3 is MTL and it's a good improvement in clocks over Intel 4 at the same power. I imagine the clocks on PTL probably aren't going to be a lot higher so I'd bet on higher base clocks and an emphasis on maintaining clock speeds.

    165U v 265U (12W min/15W base/57W turbo):
    Max Turbo Frequency
    4.9 GHz v 5.3 GHz
    Performance-core Max Turbo Frequency
    4.9 GHz v 5.3 GHz
    Efficient-core Max Turbo Frequency
    3.8 GHz v 4.2 GHz
    Performance-core Base Frequency
    1.7 GHz v 2.1 GHz
    Efficient-core Base Frequency
    1.2 GHz v 1.7 GHz
    Reply
  • Mr Majestyk
    thestryker said:
    Part of me wonders if the lack of higher voltage support is why there aren't any desktop CPUs planned with PTL. If this is the case then perhaps that's where the usage of N2 comes in for NVL. This may also be the performance part they're referring to when talking about 18A-P.

    The only client part currently being manufactured with Intel 3 is MTL and it's a good improvement in clocks over Intel 4 at the same power. I imagine the clocks on PTL probably aren't going to be a lot higher so I'd bet on higher base clocks and an emphasis on maintaining clock speeds.

    165U v 265U (12W min/15W base/57W turbo):
    Max Turbo Frequency
    4.9 GHz v 5.3 GHz
    Performance-core Max Turbo Frequency
    4.9 GHz v 5.3 GHz
    Efficient-core Max Turbo Frequency
    3.8 GHz v 4.2 GHz
    Performance-core Base Frequency
    1.7 GHz v 2.1 GHz
    Efficient-core Base Frequency
    1.2 GHz v 1.7 GHz
    Well ,Panther Lake is going to have higher core counts than Lunar Lake and is more the spiritual successor to that rather than Meteor Lake IMO. Top tier PL is 4P+8E+4LP-E +12Xe3 vs Lunar Lake 4P+4LP-E+8Xe2. We get to compare 18A to N3B. I will bet PL has higher boost clocks than LL.
    Reply