It took a few years longer than expected for Intel to make the move from 14nm to 10nm processors… and the company still hasn’t completed that transition. Only some 10th-gen Intel Core chips are manufactured using a 10nm process, while others are 14nm chips.
But according to a presentation slide from one of Intel’s partners shows the chip maker’s plans for the next decade, the company hopes to make the jump to a smaller node every other year for the next 10 years… eventually resulting in 1.4nm chips by 2029.
Update: Intel tells AnandTech that the slide is an altered version of one provided by the chip maker, which means the process nodes are more of an educated guess than an Intel timeline.
Here’s Intel’s timeline:
- 7nm in 2021
- 5nm in 2023
- 3nm in 2025
- 2nm in 2027
- 1.4nm in 2029
Generally speaking, the move to a smaller process node results in improved efficiency and performance.
Starting with the launch of its first 65nm chips in 2005, Intel was following a “tick tock” cadence, where the company would release a new family of processors using the latest fabrication process, followed by a next-gen family or processors that introduced a new architecture.
So every other year, the company moved to a smaller node… until it got stuck at 14nm.
Between 2014 and 2019, Intel released 5th-gen, 6th-gen, 7th-gen. and 8th-gen chips all manufactured on a 14nm node. Even Intel’s 10th-gen Comet Lake chips are 14nm processors. So far the only mass produced 10nm chips are the 10th-gen Intel Ice Lake processors, and there aren’t a lot of computers using them yet.
All of which is to say that while it’s interesting to see Intel’s plans for the next decade, it’s probably a good idea to take them with a grain of salt — it’s possible that Intel might not be able to meet its ambitious goals.
That said, the image above does show that there are also “backport” opportunities, where a processor that’s planned to launch as a 5nm chip, for example, could instead be released as a 7nm++ chip if it turns out that the company isn’t ready to ramp up production of 5nm processors yet.
So it looks like Intel has a contingency plan for any unanticipated delays.
Meanwhile, while Intel is just starting to ramp up production of its 10nm chips, AMD and many ARM-based chip makers are already producing 7nm processors. I wonder what they’re roadmaps for the next 10 years look like.
Costs and complexity per new node are growing exponentially. Considering the troubles Intel has had with 10nm, there’s no chance Intel will be able to stick to a 2 year cadence. lol
Microsoft broke the endless cycle of WinTel PC upgrade suffering with Windows 8 and Vista. Lots of users switched to alternative non-WinTel devices after those fiascos. I just don’t see the demand for upgraded PC hardware these days.
IMO that’s more because things have reached a “good enough” state for most people, rather than because WinV/8 had issues. We’re seeing a similar reception to 5G too, with a lot of people saying they don’t feel a need for faster mobile data.
We as humans haven’t really changed over the years, after all. There is an analog limit to how much stuff we can consume with our Mk.1 Eyeballs and Brain V1.0.
@Intel good luck with that
There are no consumer 10 nm CPU from Intel to buy. And I don’t see them for another year.
1.4nm, dream on, they probably missed the millennium and meant 3029.
….. So by about 2033 they should be down to 0nm (to the nearest nm!)
Interesting, I’m not sure anyone has solved the quantum tunnelling issue that’s expected to show up on real 5nm nodes, seems like the currently announced 5nm chips will be marketing speak rather than any specific measure of die geometry according to wikichips.
Of course. Dangle nanometers is the game. So easy to say, so hard to actually deliver.
Considering that they are reviving haswel pentium processors of 2013 , I highly doubt that they can do this.
Well, maybe they can outsource to TSMC? They plan 5nm for Q2 2020 while intel hopes to achieve that by 2021. Ryzen 4 will be already 5nm while Intel prepares to get to 7nm, and that’s their optimistic plan.
TSMC will not want to sell to Intel. So not only is it highly unlikely, if they did the price would be too high where they would be losing money on per chip. That’s not even considering the CCP banning USA products, or trade-war they are having. And even if they did, TSMC has their own limitations on productions for their flagship wafers (7nm, 5nm, etc etc) so supply would be too small to make any meaningful impact (paper launch?).
And lastly, Intel is a semiconductor foundry first, and an architecture designer second. So TSMC isn’t the matter, it could be Samsung, Toshiba, SK Hynix, Micron etc etc. It still wouldn’t make any sense. It’s like a bakery which has an oven that’s not heating as well, with the baker going across the street and buying bread from the other baker. He’s got all the water, flour, grains, salt, and mixers stocked.
Not to mention the shot it would have to their reputation, with investors pretty much flocking away from the company at a hat’s drop.
They count nm differently though, intel and TSMC
Comments are closed.