Intel Puts 1nm Process (10A) on the Roadmap For 2027 (tomshardware.com) 35
Intel's previously-unannounced Intel 10A (analogous to 1nm) will enter production/development in late 2027, marking the arrival of the company's first 1nm node, and its 14A (1.4nm) node will enter production in 2026. The company is also working to create fully autonomous AI-powered fabs in the future. Tom's Hardware: Intel's Keyvan Esfarjani, the company's EVP and GM and Foundry Manufacturing and Supply, held a very insightful session that covered the company's latest developments and showed how the roadmap unfolds over the coming years. Here, we can see two charts, with the first outlining the company's K-WSPW (thousands of wafer starts per week) capacity for Intel's various process nodes. Notably, capacity typically indicates how many wafers can be started, but not the total output -- output varies based on yields. You'll notice there isn't a label for the Y-axis, which would give us a direct read on Intel's production volumes. However, this does give us a solid idea of the proportionality of Intel's planned node production over the next several years.
Intel did not specify the arrival date of its coming 14A node in its previous announcements, but here, the company indicates it will begin production of the Intel 14A node in 2026. Even more importantly, Intel will begin production/development of its as-yet-unannounced 10A node in late 2027, filling out its roster of nodes produced with EUV technology. Intel's 'A' suffix in its node naming convention represents Angstroms, and 10 Angstroms converts to 1nm, meaning this is the company's first 1nm-class node. Intel hasn't shared any details about the 10A/1nm node but has told us that it classifies a new node as at least having a double-digit power/performance improvement. Intel CEO Pat Gelsinger has told us the cutoff for a new node is around a 14% to 15% improvement, so we can expect that 10A will have at least that level of improvement over the 14A node. (For example, the difference between Intel 7 and Intel 4 was a 15% improvement.)
Intel did not specify the arrival date of its coming 14A node in its previous announcements, but here, the company indicates it will begin production of the Intel 14A node in 2026. Even more importantly, Intel will begin production/development of its as-yet-unannounced 10A node in late 2027, filling out its roster of nodes produced with EUV technology. Intel's 'A' suffix in its node naming convention represents Angstroms, and 10 Angstroms converts to 1nm, meaning this is the company's first 1nm-class node. Intel hasn't shared any details about the 10A/1nm node but has told us that it classifies a new node as at least having a double-digit power/performance improvement. Intel CEO Pat Gelsinger has told us the cutoff for a new node is around a 14% to 15% improvement, so we can expect that 10A will have at least that level of improvement over the 14A node. (For example, the difference between Intel 7 and Intel 4 was a 15% improvement.)
Process != actual size (Score:5, Informative)
Just remember everyone, the "1nm process" has nothing to do with size.
Chip fabs just stole this unit of measurement to use as a branding noun.
The "1nm process" will represent transistors with gates the physical size of 42 nanometers of atoms in width, and a metal pitch 16 nanometers thick.
That is roughly 300 silicon atoms wide and 100 silicon atoms thick.
We have not by any measure reached the limits of the atomic scale despite what their marketing BS names would lead you to believe.
Re:Process != actual size (Score:4, Funny)
No, I totally believe Intel really makes 1nm processors and the next one coming is 0nm using metaphysical quantum gates at infinite clock speed.
Re: (Score:2)
Even better will be the -1nm process.
Re: (Score:2)
Indeed, -1nm will one of the most significant nodes, because Intel expects those processors to be capable of producing results before performing the computation.
Re: (Score:2)
Re:Process != actual size (Score:5, Funny)
Re: (Score:3)
When did this start? I've done circuit layout, but on older processes (130nm, 200nm, etc.) and those processes definitely had a gate length that equaled the name in the size (e.g. 200nm process had 200nm gate length). How did we get from 130=130 to 1=42?
Re: Process != actual size (Score:3)
Re: (Score:3)
Re: (Score:2)
The node name was supposed to roughly indicate the density of components that could be produced, which the roadmap wanted to double every 18 months. For a while that was roughly the same as the size of the minimum feature, so that length was used. Density doesn't scale linearly with length though, so they were always going to diverge. They diverged even faster when we had to give up just shrinking MOSFETs and started doing things like making FinFETs, and now we mostly attempt to achieve Moore's law (the opt
Re: (Score:1)
The node name many years ago used to match the polysilicon gate length. Each subsequent node was 0.7 times the prior one. I.e., the shrink factor is 0.7. Note that 0.7 * 0.7 = 0.49, or roughly 0.5 — so transistors on the new node take up 1/2 the space of the prior. Put another way, you can put twice as many transistors in the same chip area. This is the key — the transistor budget doubles each generation. Other benefits, such as increased transistor drive strength. However, the area availabl
Re: (Score:2)
Dennard scaling has been dead for about 20 years.
Re: (Score:2)
When it comes to Intel we need to wait for their CPUs to hit retail. If they are mini-furnaces again then we can judge their process appropriately.
16nm (Score:5, Informative)
This marketing horseshit is so tiresome.
They're 42nm gates on 16nm traces.
There is actually some really interesting engineering happening with these but they haven't solved quantum tunneling and the only 10 angstrom feature exists in the fictional world of a marketing department.
Some tech details on here:
https://en.m.wikipedia.org/wik... [wikipedia.org]
Re: (Score:2)
We're at a point we could ditch this measure, and start to use something like component height, or "number of functions per element".
Re:16nm (Score:4, Informative)
Re: (Score:3)
Where's the "astronaut gun" meme because what that article didn't have was who fired the first shot in this marketing buzzword war.
Keep clicking back through nodes and get to 22nm [wikipedia.org] and Since at least 1997, "process nodes" have been named purely on a marketing basis, and have no relation to the dimensions on the integrated circuit;[1] neither gate length, metal pitch or gate pitch on a "22nm" device is twenty-two nanometers.
AI powered fabs (Score:2)
Re: (Score:2)
Re: (Score:2)
Pretty much. Intel has a lot of experience lying to its customers and many of them eat it up.
Re: (Score:3)
On that one, TSMC and Samsung upped Intel when it came to lies.
Intel's 14nm was about equivalent to the others 10nm. 10nm to 7nm, etc...
Of course, Intel couldn't have the worse number, so they followed what the others were doing. In the end, I think manufacturers officially dropped the "nm" thing, so now "5nm" is just "5", and the A is just an A. The abbreviation for Angstrom has that little circle on top of the A (it is a Swedish letter), it doesn't appear in official Intel communication, nor do they say t
Re: (Score:2)
Re: (Score:2)
> I think Intel is serious about getting into fab services.
What gave it away? When they built a 100 billion dollar fab?
Re: (Score:2)
Re: (Score:2)
Well, crappy hardware maker makes hardware for even crappier software maker. Somehow I am unimpressed.
Re: (Score:2)
Intel has been using their older fabs to make 3rd party designs for a while now, maybe over a decade. It doesn't get much press because it mostly isn't a very interesting business segment. It is about as interesting as "hollywood blockbuster from the 1980s still sells a few new DVDs each year" or "popular 1980s movies still generate income on streaming services".
Intel has been making n
Re: (Score:2)
Worse than that, the performance improvement is only at best 15% .. Going from 2nm to 1nm implies a 100% improvement but it's only a 15% improvement. Each node hop has for the last few years been false advertising.
Putting things on a roadmap is easy (Score:2)
Actually delivering is hard. As especially Intel should know by now.
If only Slashdot supported Unicode... (Score:2)
We could use the actual Angstrom symbol rather than just a capital "A".
https://en.wikipedia.org/wiki/... [wikipedia.org]
A roadmap is to be postponed (Score:3)
If you believe they can actually deliver what a roadmap promised on the date given, you are still too young.
Sub atomic (Score:2)
In a few years they'll start claiming they are sub-atomic. Of course the actual device will be a micron in size, but it doesn't matter. They'll call it sub-atomic, same way you told your girl 5 inches was 6 inches.
Re: (Score:1)
In a few years they'll start claiming they are sub-atomic. Of course the actual device will be a micron in size, but it doesn't matter. They'll call it sub-atomic, same way you told your girl 5 inches was 6 inches.
Next time, insert your big toe. She'll think it's a foot.
Idiocy (Score:2)