By David Manners 22nd April 2022
Assuming Intel’s 18A is an equivalent process to TSMC’s N2 – 1.8nm vs 2nm if the numbers mean anything – then Intel might be on track for upsetting TSMC’s reign as process technology leader.
At TSMC’s recent results call, CEO C.C. Wei said: “Our progress so far today for the N2 is on track. And all I want to say is, yes, at the end of 2024, you will enter the risk production. 2025, it will be in production, probably close to the second half or the — or the end of 2025. That’s our schedule.”
Intel recently announced that 18A is six months ahead of schedule with its introduction planned for H2 2024.
If this is how things work out, then Pat Gelsinger’s pledge that Intel would regain process technology leadership in 2025 will be fulfilled.
Wise words Peter, the proof of the pudding will be in the eating not the marketing fluff. If Intel’s datacentre and PC products give customers more than than products from AMD, Nvidia and the Arm camp then Intel will have won this battle. Comparisons of relative transistor density and performance-per-watt can come later.
But introducing them sequentially, Patrick, and certainly not “skipping” 4, 3, and 20A as you said.
Given that Intel have spectacularly failed with process technology at multiple nodes in the last 5 years or so and are still at least a year behind TSMC — who have succeeded with pretty much every node and half node in the same time period — you have to take Intel’s predictions of not just pulling level with TSMC but overtaking them in the next 3 years with a *very* large pinch of salt, regardless of what Pat Gelsinger says.
Talk and promises are cheap and easy, delivering on those promises is difficult and expensive…
Very true Ian, but TSMC have slipped on N2
True David, and TSMC N3 also slipped — which is why I said “pretty much every node” 😉
I think this simply shows that rolling out new nodes at these geometries with brand-new process features is simply getting harder and harder. I also think that given TSMC have far more experience than anyone else doing this recently (and with better success), so if TSMC slips (quite likely) then going by history I’d expect Samsung and Intel to slip even further…
Absolutely Ian. Going by the history, the one to bet on is TSMC.
Going by history I bet on TSMC over Intel’s LGBTQ rainbow coalition. From each according to his ability to each according to her need never worked out in practice. Exhibit A is the Red Army and a pack of needy Russian generals.
“Quality yields are incompatible with the new diversity equality culture” … man that is some good stuff you are smoking …
Also ASML will of course sell machines to Intel, why hold a grudge, that is not good business. Having more customers (diversity if you will) IS good business, especially if they are in different geographies and economies.
And whether you like it or not, putting all eggs in one basket (Taiwan), especially if that basket is heavily coveted by a large bully, is probably not the best idea. So Pat Gelsinger is right in making that statement and he is far from the only one making that statement loudly, hence the US and EU fab projects that have been discussed here numerous times on this blog.
ASML will sell tools to its long time loyal TSMC & Samsung customers before selling to competitor Intel. ASML bet the company on extreme UV tools. Intel refused to buy. Samsung and TSMC saved ASML. For sure Intel gets one tool for development. More than one tool depends upon whether TSMC and Samsung need all tools ASML can manufacture. TSMC has warned Intel not to invest in so many fabs because TSMC can supply the entire market [and intends to do so].
Pat Gelsinger damaged his prospects when he loudly declared that Taiwan was not a safe chip source because of hostile mainland China. He came away from that meeting with TSMC in Taiwan with no increase in allocation of chips from TSMC.
Gelsinger molded Intel into a woke LGBTQ affirmative action paradise. Quality yields are incompatible with the new diversity equality culture. Diversity and equality are the antipathy of meritocracy and excellence.
You’re about 2 years behind on current events Patrick. Intel indeed was one generation late to adopt EUV, but is now very focal to be the first mover in EUV High-NA (EXE platform). ASML confirmed Intel was first to order EXE machines, to be delivered in 2023.
Skipping three nodes to 1.8 nm is courting disaster and has never been done.
Please try and keep up Patrick, they are not skipping three nodes they are developing 4nm, 3nm, 20A and 18A sequentially. And who was the first company to give ASML $4bn to invest in EUV development back in 2012? Intel!
They are developing 4nm, 3nm, 20A and 18A in parallel.
Indeed. The explanation of what Intel are attempting seems to really be 2 “real nodes”, where 3 and 18A are incremental shrinks of 4 and 20A respectively. It may not be quite that simple, but I suspect it boils down to that in the end.
Am I alone in finding Intel’s marketing FUD and attempts to change the language here (nm to A, “RibbonFET”, Foveros, …) to create the impression that they have unique, leading edge technology when it’s likely equivalent to TSMC all rather tiresome.
I think they would do much better to play this straight and forget the marketing games.
At the end of the day, anyone who really wants to know how these processes (and libraries) stack up against one another simply completes layout on an ARM core and compares. You’ll never see the results published, but those who need to know know.
All that said, the Intel roadmap is certainly ambitious – but they do have some opportunity to close the gap to TSMC if TSMC do falter at 2nm and Intel execute perfectly. Is that a 20% chance ? 40% ? higher ? Really hard to say at this point.
Your email address will not be published. Required fields are marked *
Ruminations on the electronics industry from David Manners, of Electronics Weekly.
Get our news, blogs and comments straight to your inbox! Sign up for the Electronics Weekly newsletters: Mannerisms, Gadget Master and the Daily and Weekly roundups.
Read our special supplement celebrating 60 years of Electronics Weekly and looking ahead to the future of the industry.
Read the Electronics Weekly @ 60 supplement »
Read the first ever Electronics Weekly online: 7th September 1960. We've scanned the very first edition so you can enjoy it.
Read the very first edition »
Electronics Weekly teams up with RS Grass Roots to highlight the brightest young electronic engineers in the UK today.
Read our special supplement celebrating 60 years of Electronics Weekly and looking ahead to the future of the industry.
Read the Electronics Weekly @ 60 supplement »
Read the first ever Electronics Weekly online: 7th September 1960. We've scanned the very first edition so you can enjoy it.
Read the very first edition »
Tune into this Xilinx interview: Responding to platform-based embedded design
Tune into this podcast to hear from Chetan Khona (Director Industrial, Vision, Healthcare & Sciences at Xilinx) about how Xilinx and the semiconductor industry is responding to customer demands.
By using this website you are consenting to the use of cookies. Electronics Weekly is owned by Metropolis International Group Limited, a member of the Metropolis Group; you can view our privacy and cookies policy here.