[18] Dr. Ann Kelleher, Intel Process Nodes
ฝัง
- เผยแพร่เมื่อ 18 มิ.ย. 2024
- It’s somewhat of an understatement to say that Intel’s future roadmap on its process node development is one of the most aggressive in the history of semiconductor design. The company is promising to pump out process nodes quicker than we’ve ever seen, despite having gone through a recent development struggle. Even with CEO Pat Gelsinger promising more than ever before, it’s up to Intel’s Technology Development (TD) team to pick up the ball and run with it in innovative ways to make that happen. In charge of it all is Dr. Ann Kelleher, EVP and GM of Intel’s Technology Development, and on the back of some strong announcements last year we reached out for the chance to interview her regarding Intel’s strategy.
Written Version: www.anandtech.com/show/17243
[00:00] Cat Surprise
[00:24] Intro
[00:53] Q1: Journey Through Intel
[03:18] Q2: Transferable Skills
[06:42] Q3: TD vs R&D
[08:20] Q4: Research vs Pathfinding
[09:38] Q5: Intel Process Node Strategy Pivots
[14:27] Q5b: Gelsinger Era, or Kelleher Era?
[15:03] Q6: How Diverse Does Intel Need To Be
[16:26] Q7: Moving to Industry Standard IFS Tools
[19:27] Q8: Where to Install New Nodes
[21:05] Q9: Difficulties on 10nm
[23:29] Q10: Improvements on 10nm and Future Learnings
[25:04] Q11: Flexibility
[25:32] Q12: High-NA Development
[28:06] Q13: High-NA Installation
[28:52] Q14: Approach in developing long-life nodes
[30:23] Q15: Intel and DTCO
[31:22] Q16: Process Nodes and 'Being Open'
[32:38] Q17: Non-Intel Partners for New Nodes
[33:31] Q18: AI Acceleration in EDA Tools
[35:12] Q19: Is Intel's Investment Enough?
[35:50] Q20: Focus on Components Research
[36:29] Q21: Weirdest Element
[37:18] Cat Tax
Welcome to the TechTechPotato (c) Dr. Ian Cutress
Ramblings about things related to Technology from an analyst
Patreon helps support this channel: / techtechpotato
Join the TechTechPotato Discord! All we ask is you sign up for Patreon for as little as $1.50 a month and your account will be linked - it only has to happen once, and then you can cancel if you wish. Come chat and hang out!
Support the channel by hitting that subscribe button, or using our Amazon affiliate link. Please note that TTP may receive a commission if you purchase anything through these links
Intel Core i7-11700K (8C/16T): geni.us/SreZM
Intel Core i5-11600K (6C/12T): geni.us/T4WG
Intel Core i5-11400F (6C/12T): geni.us/6UISblx
AMD Ryzen 9 5950X (16C/32T): geni.us/5950X
AMD Ryzen 9 5900X (12C/24T): geni.us/5900X
AMD Ryzen 7 5800X (8C/16T): geni.us/5800X
AMD Ryzen 5 5600X (6C/12T): geni.us/5600X
Intel Core i9-10900K (10C/20T): geni.us/Bench2020-10900k
Intel Core i7-10700K (8C/16T): geni.us/Bench2020-10700K
AMD Ryzen 9 3950X (16C/32T): geni.us/Bench2020-3950x
AMD Ryzen 7 3700X (8C/16T): geni.us/Bench2020-3700x
AMD Ryzen 5 3600 (6C/12T): geni.us/d08ThJ4
Follow Ian on Twitter at / iancutress
Follow TechTechPotato on Twitter at / techtechpotato
#techtechpotato #intel #annkelleher - วิทยาศาสตร์และเทคโนโลยี
![[31] Dr. Ann Kelleher, EVP Technology Development, Intel](http://i.ytimg.com/vi/SOY0Yh8y-5Q/mqdefault.jpg)
![[31] Dr. Ann Kelleher, EVP Technology Development, Intel](/img/tr.png)
![[19] Raja Koduri: Intel SVP and GM of AXG](http://i.ytimg.com/vi/KdXjMfvf72E/mqdefault.jpg)
Fascinating. She has passion and pride in her job and it shows. Right woman for the job. Looking forward to the future lakes and rapids and forests from intel.
Dr. Kelleher might not be the most charismatic person in the field, but charisma doesnt move Intel to new nodes on time, but what does is a very determined and skilled engineer leading the team at Intel. Same applies to Pat. The leadership at Intel is now the best its been in 10 years. Within the next 3-4 years I bet you Intel is back on top as the industry leading foundry, and with Intel foundry services, that means they will be poaching tons of customers from Samsung fabs (like Qualcomm already) and TSMC customers will be next in line to move to Intel, heck even Nvidia allegedly has discussed a future of using Intel fabs. These are the most exciting times for Intel since Sandy Bridge in 2011.
Looking forward to watching this! She's a great mind and the perfect person to pose these questions to.
Purrfect
What an incredible career !
Thanks for the interview.
Superb thanks very much for asking about so many topics. She is right that there is insufficient awareness of the component and process challenges and yet its fantastically satisfying engineering work to be part of moving technologies forward.
I'd love to see you explore the "openness" of the tech xfer ecosystem and how they manage to share advances to mutual benefit. I had an amazing tutorial from the Toyota Purchasing Director about "Partnership Sourcing" and it seems that they want this philosophy more now.
As so much more technologies are part of the development, ML and periodic table referred to here, it will be fascinating to see how they evolve. But they clearly have a highly capable leader at TD and I hope Ann continues to share insights with you. Proper geeky stuff!!!
Sometimes it takes the pain from failure to breed the process innovation and paradigm shifts necessary to guarantee future success. With the failure of 10 nm to volume ramp in the time frame originally anticipated, Intel essentially fell on its sword and became a laughing stock. But they seem to have learned their lesson and to have added flexibility… nimble and quick to adjust. I wonder to what degree Jim Keller had influence on Intel’s paradigm shift?
But Intel needed to stumble, they needed to know that even the mighty intel could stumble, and that the market could and would move against them if they failed to deliver superior tech. AMD came back, Apple left and designed M1.
Intel seems hungry again, laser focused, and I am thrilled to see what new products they unleash to the ecosystem.
Dr. Cutress, thank you very much for this interview with Dr. Ann B. Kelleher. It is so impressive to see how many challenges are in front of Dr. Kelleher and her teams now. A lot of future success for Intel depends on her teams.
I am not sure what kind of questions are tolerated to ask when doing such interviews, the only one more answer I would be happy to hear from Dr. Kelleher is about collaboration with IBM on new processes and packaging aspects.
Thank you very much, Dr. Cutress for these amazing interviews.
Kind regards,
Vyacheslav
I've been told to expect 'more soon' on that. They've been saying it for a year though
I would be utterly amazed if Intel manages to stick to their new roadmap:
2021: Alder Lake
2022: Raptor Lake
2023: Meteor Lake
2024: Arrow Lake
2025: Luna Lake
2026: Nova Lake
Yep, we all will, it’s really aggressive, but quite a large portion of the compute on Meteor Lake and onwards will be fab’ed externally. I.e: TSMC.
With that context, Intel meeting their new roadmap seems somewhat more realistic, the risk is that they’re relying on their partners ability to deliver, also combined with Intels ability to package these products - EMIB (tiles) and then Foveros.
Intels new nodes will of course step in and takeover on compute die, which product/year exactly I can’t remember, 2025 I think, the risk there is will they be ready on time? Else the backup plan to continue outsourcing compute die.
The tiled approach of the industry means that external suppliers won’t be going away. Intel won’t be able to do it all themselves. At a high level, Should Intel start focus on compute die, then Fab capacity on other silicon needed is constrained, high volume but low profit die like motherboard chipsets will start to be outsourced for example.
Ian/Dr. Cutress: PLEASE Make a video and clarify what your just announced decision to leave Anandtech will mean for the channel!! And tell us where you are going specifically if you can! Please dont let this channel die now!
This channel will stay. Will make a video for next week if I can sort all the paperwork.
@@TechTechPotato Thanks Ian! Thats amazing! Thanks again for replying!
Great interview and learning. Appreciate Dr. Ian for such quality content. Would have been nice if there were questions around packaging as more questions were focused on process. Nevertheless, excellent content.
Great interview, keep up the good work!
Very good insights once again. You do a wonderful job of skirting the line between interesting technical details and verbose jargon.
Thanks for the interview, to both of you!
Fascinating interview, fun to see Intel being relatively so open about discussing this stuff, i guess they kinda have to be now that they pivoted a good chunk of their business as being a fab for others.
Very informative , thank you for this, I'd love an interview with Tom 'TAP' Petersen
I enjoyed listening to this.
Thanks for doing these interviews Ian and big thanks to Ann for coming on and having a chat.
This was a really enjoyable interview.
thanks a lot for the excellent interview Ian and Ann!
I was really hoping you'd dig a little deeper as to why Intel was so reluctant and frankly late on adopting ASML's EUV steppers? I really don't understand. Now they're trying to catch up but why would they let it come to this? I have yet to hear an official explanation for it.
While EUV wasn't ready when the 10nm Intel node was defined... no, but that's not an answer as to why it wasn't but on the roadmap the second ASML, TRUMPF and Zeiss got it working. Stating that nobody could have known is oversimplifying things IMHO. There are obviously great physicists and process engineers at INTEL, so they must have known that they needed to adopt it ASAP in order to stay competitive. How come this didn't translate into an immediate course correction - INTEL is big enough to pursue multiple routes and not adopting EUV immediately frankly seems like the biggest technological mishap possible. So from my perspective this either was a massive management failure OR their engineering/physicis-departments actually believed they could counter EUV with even more DUV exposures.
Maybe I am misinformed and INTEL did try to get some much earlier orders in with ASML and lost out - or was unwilling to pay the premium? "The timing wasn't right" sounds like the management didn't prepare for it becoming ready, basically.
Well... best of luck catching up. I just hope nobody tries to attack ASMLs de facto monopoly here - I'm really curious how long the americans will try and keep us europeans from selling the machines to china as well. I'm a bit tired we're being told what to do from the other side of the atlantic. That being said, I hope we wake up some more and build more fabs here as well.
Intel has lots of EUV machines, make no mistake. It was simply bad planning and sunk cost - no point doing a hard pivot when your next Q financials matter more to you than long term progress. BK had blinkers on and simply didn't care.
"part of the periodic table we use" - cards kept firmly close to the chest - great interview
This was a great video. Thank you
Fantastic interview 👏
Great interview, so much thought behind her answers, great insight thanks alot!
The right person for the job. Great interview.
@23:32 I got chills, lets go deeper! Love it!
Great interview, what a impressive person
Great interview.
What a talented, knowledgeable and well informed gal! Great interview!
She wouldn't be a Cork woman if she didn't use the word ..."like" at the start, middle and end of a sentence. You can take the person out of Cork but never the other way round. Shrewd and highly intelligent lady.
Do the headphones just look broken or are they broken?
They're a bit broken. Went through an airport scanner and got trapped in the rails.
Some acronyms could be rapidly defined as subtitles, I'm not that advanced in the terminology and it might help me the understand thoses key words in there contexts. although I'm might not be in your targeted audience. I hope you interview Lisa Su too.
Wait she was there in one of the event videos I think
Should have made her CEO. She sounds competent and not completely unlikable.
I'm probably butchering terms here, but, by flexibility with cutting in new technology, does that mean that a design might not need the new process everywhere on the chip, or does that mean the transistors themselves on a chip can be printed with the new tech like EUV?
Ann was brilliant, what an asset to have in your company all those years. I would love to have known what her thought were about Intel being forced to use TSMC, was it avoidable, will it help her in any way, when does she think they will overtake TSMC, etc?
It seems Pellicle issues and gap between TSMC and the rest of the industry was a factor, it would be great to know if this was a factor.
I have a hard time believing Intel will be able to stick to their node roadmap. If they do it will be an incredible achievement.
Thanks Ian and Ann.
So IMO the crap ton of TSMC 3nm is in case Intel 4 is delayed but doesn't it mean that they need to develop for 2 different processes or are they close enough that it's not that different?
TSMC N3 is going towards Meteor Lake (14th gen in Q2 2023) IGP's, and a few other things. It is NOT a replacement for Intel 4, which is what the main Meteor Lake CPU will be on. Intel originally struck that deal with TSMC long ago, because of 10nm issues, but has been able to keep Intel 4 on track for Q2 2023 launches.
Finally someone who is talking about a cat tax. Cats have been freeloading for far too long!
Great interview, I can only hope that all the new and older technology is being closely kept in security-related matters so that the CCP will not have access to their PPL .
Auto caption turned pat gelsinger to pat girl singer. Nice
You are gonna be adding some +s' to intro slide. But this women deserves a higher managment job for sure if she was not cut so great for her role she currently fills! Comforting to know intel is not hiring buck passing, bobble-headed, yes men.
Did we find out about the weirdest element used?
0:17
This always make me laugh.
Up Cork!
Awesome video! Also, first.
The real competitive benchmark vs amd is server processor, which provides highest profit per transistor.
Intel xeon currently stuck at 10nm ice lake and its performance is way below epyc zen3
This is a really interesting look into the the Intel hive mind!
Intel is great at trotting out people to repeat talking points over and over these days.
As if invoking "Mooore's Law" makes you beat the competition.
I guess I just have so much doubt I need to see results rather than hear about the To Do list.
23:00 I am pretty pleased to hear details like this. Makes me feel better about continuing to hold the stock.
oh yeah it sounds good then you realize no actual information was given
Plenty of information was given... :)
@@TechTechPotato You are right of course. I was referring, cynically, to the fact that long time corporate folks are good at saying a lot while committing to very little. Survival tactic I guess ;)
Yeah the whole thing is that can we within 12 months get competitive with the AMD and tsmc which is a Taiwan suppliers the low cost of this stock right now since we got a 6% drop makes it very tasty where is AMD is overvided right now and they have to give back shares just in order to make a profitable cuz it's not really profitable for the shareholder so at this point I haven't looked really at tsmc too much but I'm sure it's higher price than what Intel is so I think what the new CEO I heard that we have a good chance of being competitive with the other chipmakers if that's the case as a short-term stock even over the next year should
Great !
It sounds like the leadership, executive and management for this section of Intel is working well - at least if what they say turns out to become reality over the next 5-7 years.
There are parts of Intel that really fail in management and executive - and it impacts me personally with the Intel Arc graphics cards. Do you know who could give the proper answers here? It certainly sounded like the leadership roles got moved around a bit.
You definitely get the impression that Dr. Kelleher really knows her stuff.
Why is Intel kept secret how many transistors are there on each Chip
Cats FTW!
(Every video is better with cats)
Did she ever get back to you on the weirdest element question?
autogenerated subtitles are fun, so you will be running your asmr high na euv lithographie machine in a pub ;) 27:45
Yeah I'm wondering about the research and development and are you getting enough free cash and do you have enough investors or house stock sales and are you going to be able to have enough money for research and development in order to launch this program within the next
That's what I call timing see how lacquered this company was doing nothing and all of a sudden my ability to determine that this is a turning point so here you are going from slow. To possibly be in the number one competitor and that's really what timing is and then there's execution and then being there as you gradually progress up to the number one position to win the super bowl and this is where you go from 0 to 200 miles an hour and when NASCAR races so it's really how we see things the future midterm and short-term growth of a company so I know according to stock price and been watching Intel then it's bringing in about 99 cents a day per share so at that point say it's $50 a shared at 99 cents a day 32 days in a month or 30 days in a month you can estimate what we're bringing in over the month on average with this chip so you might bring in $15 a month because half the days in the market are bearish and only 15 days are bullish so you have to say you're going to have your up and down days so on average you have to determine whether a share is going to bring in 50 cents a day over a month. On average if the market's going upward if it does and the accumulate enough shares I can eliminate my liabilities and then go into complete
I wish I had a big brain...
i think intel is working with Atomera.
Atomera doesn't seem big enough?
@@TechTechPotato if intel plans to beat tsmc and samsung its new machine from asml will need to have atomeras mst recipe installed on the tool to make the new gate all around nano ribbon vfet. time will tell.
She looks like Ians Mom
I watched this because we need more woman in tech, I love being exposed to smart woman! Our industry needs more powerful woman!
while i clicked for a similar reason, its stupid to say we need more powerful women. if they are more interested/skilled than men that compete for positions, then yes. otherwise, no!
she's awesome. why isnt she running a space programme? intel is holding her back
What she's doing is probably more complicated than a space program, IMO. I mean, high energy photons vaporizing metals into higher-energy photon-emitting plasma to imprint a transistor pattern and then scaling that for products that produce several times more cash then NASA's annual budget....
Why do you assume she's so incapable of finding the career she prefers?
The answer that I hear from this interview is: "I prefer not to go into any detail"
Yeah, then why are we having this interview exactly?
My ears hurt. Why do your high profile interview objects not understand normal customers?
She does process node research. That's relevant to 'normal customers' how? She's speaking to her industry.
Tech managers are robots. Really boring process people.
I'm fucking fed up with murica poaching European talent.
Oh my god the corporate speak is nauseating. Doesn't sound like a person, she sounds like an AI generated buzzword storm.
Or ooor! hear me out here… maybe… your just dense?😂
Intel is a corporation so how are you expecting her to speak about their goals exactly?
it's unfortunate that business secrets etc. get in the way of some really interesting discussion sometimes, but that isn't anything new in tech unfortunately
@@defeqel6537 If you are expecting her to divulge company secrets then you're watching the wrong video. You will NEVER get corporate secrets from a TH-cam interview, if you do, don't expect that person to remain in post very much longer.
@@clansome I'm not expecting it, and it's why I point it out here. I'm just lamenting how some interesting discussion is prevented by that reality.
It is unfortunate that we are not able to learn more detailed technical information. However if you are smart, you can still glean insights. For instance, I have never attributed 10nm struggles to Intel planning 10nm before EUV was ready and then not being flexible enough to incorporate it.
Doesn't look very happy, the lady.
And when did she ever get to work because
of all the job changes?
I could be wrong, but that's what a
lost life looks like ...