Over the years, chip makers have moved to smaller and smaller processor nodes in an effort to fit more transistors into a smaller space to improve performance and efficiency. But different manufacturers use different technologies, so what one chip maker calls a 7nm processor may not have the same characteristics as a competitor’s.

Now Intel has announced plans to move away from using nanometers to describe its next-generation chips. The company’s upcoming Alder Lake chips will still be manufactured using on a 10nm node, but when they arrive later this year, they’ll be branded as Intel 7. And in 2022 when Intel “Meteor Lake” chips launch, they’ll be built on a 7nm node, but they’ll be known as Intel 4.

This move makes sense since the number is almost as meaningless as megahertz these days. But it also means that Intel, which is still making 10nm chips, doesn’t look as far behind AMD (7nm) and Apple (5nm).

Intel says some of the architecture changes coming in the next few years will bring higher levels of improvements than the move to a new processor node. For example, after Intel 4 comes Intel 3, which the company says will bring an 18% improvement in performance-per-watt compared with its predecessor, along with other changes including a denser, high performance library and increased use of EUV (Extreme ultraviolet lithography) technology.

Next up is will be Intel 20A, which is expected to come in 2024. It will introduce an entirely new transistor architecture that Intel called RibbonFet (replacing FinFET) and a new power system called PowerVia that uses the back of the chip for power routing, reducing noise in power delivery and freeing up resources on the front of the chip.

Intel is also already working on Intel 18A which is scheduled to launch in 2025, but the company isn’t providing many details about that yet.

But that’s all long-term stuff. In the short term, Intel says that the first Alder Lake/Intel 7 chips for mainstream computers will ship later this year, bringing a 10% to 15% improvement in performance-per-watt over 11th-gen “Tiger Lake” chips thanks to the use of Intel’s new “Enhanced SuperFin” manufacturing.

In addition to designing and building its own chips, though, Intel’s long-term plans also include ramping up its contract manufacturing, and the company will make its new technologies available to rival chip designers.

Qualcomm has already announced plans to manufacture chips using Intel’s 20A technology. Amazon is also planning to use Intel Foundry Services to package upcoming chips. And Intel says it’s in talks with other semiconductor companies as well.

For a more in-depth look at Intel’s roadmap and what it could mean, I highly recommend checking out AnandTech’s deep dive.

via Intel (1)(2)(3)

Support Liliputing

Liliputing's primary sources of revenue are advertising and affiliate links (if you click the "Shop" button at the top of the page and buy something on Amazon, for example, we'll get a small commission).

But there are several ways you can support the site directly even if you're using an ad blocker* and hate online shopping.

Contribute to our Patreon campaign

or...

Contribute via PayPal

* If you are using an ad blocker like uBlock Origin and seeing a pop-up message at the bottom of the screen, we have a guide that may help you disable it.

Subscribe to Liliputing via Email

Enter your email address to subscribe to this blog and receive notifications of new posts by email.

Join 9,545 other subscribers

6 replies on “Intel’s 2025 roadmap: New architectures, new technologies, and the end of nanometer naming”

  1. To be honest, Intel is making the right moves here:

    1- swallow your pride, and use TSMC for your CPUs
    2- sell your older wafers to your competitors Qualcomm
    3- put your image/marketing inline with the competitors
    4- set up ambitious yet realistic goals for yourself

    Intel was the first to 16nm and they’ve heavily relied on it for 6 YEARS, that’s a lifetime in the tech industry. Granted there’s a substantial difference between the earliest (2014) 16nm, to the midlife (2017) node, and to the final (2020) wafer. The reason I bring this up is because the same thing applies to their 10nm nodes. Overall, they’re 2-years behind the competitor and its hurting their products, image, and business. Now to add that Intel’s wafers have traditionally been ahead of their competitors, so from a marketing viewpoint, they’re doing a disservice to the product. In general, the “nm” rating is most accurate (or best) from Intel, then Samsung, TSMC, Micron, GlobalFoundries, and lastly is SMIC.

    Example:
    TSMC 28nm < Intel 32nm
    TSMC 20nm < Intel 22nm
    TSMC 16nm < Intel 14nm-early ~ TSMC 12nm
    < Intel +14nm ~ TSMC 10nm < Intel 14nm+++
    < Intel 10nm ~ TSMC 7nm ~ TSMC +7nm
    < Intel +10nm/7nm < TSMC 5nm ~ TSMC 4nm
    < Intel 5nm < TSMC 3nm < ????

  2. So is an Intel 7 better or worse than an Intel 3?

    Unbelievable.

    btw I’m typing this on a MacBook M1. It’s incredible how much faster this thing is, battery lasts forever, and no fan. I run large statistical models all day long. I now use this tiny computer when I need the output quickly, rather than a large rig.

  3. tl;dr: Marketing seight of hand, Intel’s still behind by 1-2 process nodes.

  4. Intel is attempting to make things more confusing to hide the real numbers (deficits). It’s just smoke and mirrors. That’s all you need to know about Intel.

  5. Haha, this is hilarious, Intel thanks for the laugh 🙂
    If it is not obvious how far behind competition you are falling.
    (This reminds me of Nokia).
    You are making marketing naming even more confusing.
    Now every Intel chip has dozen different code names, families, series, generations, relations and combinations of them, and of course all meaningless.

    1. Intel has totally lost their marbles. Btw, maybe SMIC too can copy this “marketing remake”. Who needs nanometers right? Its gonna take a heck of a lot to sell consumers on this.

Comments are closed.