• Guest, The rules for the P & N subforum have been updated to prohibit "ad hominem" or personal attacks against other posters. See the full details in the post "Politics and News Rules & Guidelines."

Speculation: Intel will become fabless

Page 16 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

With the loss of its manufacturing lead, will Intel become fabless?

  • Yes, Intel is a product designer at heart, and they will seek a more flexible fabless model.

    Votes: 20 13.0%
  • No, manufacturing is integral to Intel, and they will continue to invest to stay competitive.

    Votes: 134 87.0%

  • Total voters
    154

moinmoin

Platinum Member
Jun 1, 2017
2,369
2,941
106
TSMC, as of the end of 2019, represented just under 13% of the world's total wafer production capacity. This includes ALL of their nodes, including a bunch of older, non-competitive nodes.
As of 2020 TSMC with 13.1% of the world's total wafer production capacity had 2719k wafers/month. Intel is mentioned in the text having 884k wafers/month. That makes TSMC a little over 3 times Intel in terms of total wafer production capacity. Even discounting older nodes (should we exclude Intel's 14nm as non-competitive node as well?) there is a sizable difference there.

 

DisEnchantment

Senior member
Mar 3, 2017
757
1,801
136
As of 2020 TSMC with 13.1% of the world's total wafer production capacity had 2719k wafers/month. Intel is mentioned in the text having 884k wafers/month. That makes TSMC a little over 3 times Intel in terms of total wafer production capacity. Even discounting older nodes (should we exclude Intel's 14nm as non-competitive node as well?) there is a sizable difference there.

TSMC has lots of 200m wafers. Intel fully 300mm afaik.
EDIT:
OK they put 200mm wafer equivalent, but not shown for Intel.
 

LikeLinus

Lifer
Jul 25, 2001
11,089
444
126
As of 2020 TSMC with 13.1% of the world's total wafer production capacity had 2719k wafers/month. Intel is mentioned in the text having 884k wafers/month. That makes TSMC a little over 3 times Intel in terms of total wafer production capacity. Even discounting older nodes (should we exclude Intel's 14nm as non-competitive node as well?) there is a sizable difference there.

That includes memory and such. That isn't CPU wafers. Couple of us pointed that out just a few post ago.
 

Stuka87

Diamond Member
Dec 10, 2010
5,354
1,141
136
Curiosity, is the confusion because he is talking about wafers for CPU production and not memory? Your article says:

"It is not surprising that the 300mm ranking includes only DRAM and NAND flash memory suppliers like Samsung, Micron, SK Hynix, and Kioxia/WD; four of the world’s largest pure-play foundries TSMC, GlobalFoundries, UMC, and Powerchip (including Nexchip); and Intel, the industry’s biggest manufacturer of microprocessors.
Intel being the biggest manufacturer of microprocessors does not mean they are larger than all other microprocessor companies that use TSMC combined.

Also, Intel manufactures NAND at their fabs, as well as other types of ICs that arent x86 chips.

Intel's fab is definitely not larger than TSMC if we look at total capacity. And since we don't know how much of each fab is used for CPUs, we cannot say which has more capacity for x86 chips.

Which means your claim that 50% of all of TSMC's x86 capacity is less than 20% of Intel's capacity has no basis in fact. And since AMD and Intel don't release the exact number of CPUs shipped, there is no way for us to actually determine any of this.
 

LikeLinus

Lifer
Jul 25, 2001
11,089
444
126
Intel being the biggest manufacturer of microprocessors does not mean they are larger than all other microprocessor companies that use TSMC combined.

Also, Intel manufactures NAND at their fabs, as well as other types of ICs that arent x86 chips.

Intel's fab is definitely not larger than TSMC if we look at total capacity. And since we don't know how much of each fab is used for CPUs, we cannot say which has more capacity for x86 chips.

Which means your claim that 50% of all of TSMC's x86 capacity is less than 20% of Intel's capacity has no basis in fact. And since AMD and Intel don't release the exact number of CPUs shipped, there is no way for us to actually determine any of this.
FYI - I never made any claim. I simply pointed out that there may be some confusion between the two people quoting different stats.
 

ondma

Platinum Member
Mar 18, 2018
2,015
726
106
Intel being the biggest manufacturer of microprocessors does not mean they are larger than all other microprocessor companies that use TSMC combined.

Also, Intel manufactures NAND at their fabs, as well as other types of ICs that arent x86 chips.

Intel's fab is definitely not larger than TSMC if we look at total capacity. And since we don't know how much of each fab is used for CPUs, we cannot say which has more capacity for x86 chips.

Which means your claim that 50% of all of TSMC's x86 capacity is less than 20% of Intel's capacity has no basis in fact. And since AMD and Intel don't release the exact number of CPUs shipped, there is no way for us to actually determine any of this.
You seem to be switching the metrics in mid-post. First you are talking about microprocessor companies. This would include a huge number of Apple and ARM chips. Obviously, Intel is smaller than all these combined. However, Intel does own probably 70% (just guessing) of the cpu market. Unless TSMC has a huge amount of unused capacity (I dont think they do) Intel has more *available* fab capacity for cpus than TSMC.
 

Stuka87

Diamond Member
Dec 10, 2010
5,354
1,141
136
FYI - I never made any claim. I simply pointed out that there may be some confusion between the two people quoting different stats.
Ahh, I apologize. I mistook you for another user that did just that earlier.

Specifically, this user/post:
Even if AMD managed to claim 50% of all of TSMC'S two most advanced nodes production capacity, that wouldn't amount to even twenty percent of Intel's volume on 14 and 10nm.


You seem to be switching the metrics in mid-post. First you are talking about microprocessor companies. This would include a huge number of Apple and ARM chips. Obviously, Intel is smaller than all these combined. However, Intel does own probably 70% (just guessing) of the cpu market. Unless TSMC has a huge amount of unused capacity (I dont think they do) Intel has more *available* fab capacity for cpus than TSMC.
I would agree Intel has more available capacity, we all know TSMC is running as hard as they possible can. My comment was made specifically to the user that I quoted just above yours in this post. It just happened to be I then responded to the wrong person :(
 

LightningZ71

Senior member
Mar 10, 2017
715
647
136
Yes, your comment was aimed at me.

My claim is simply: on relevant, leading edge performance nodes (which, as we see in benchmarks, Intel's 14nm node is one), TSMC has substantially less capacity than Intel. TSMC has N7, it's various refinements on the same lines, and N5, in only a couple of fabs. Intel has 14nm, and 10nm in multiple variants, in nearly a dozen high volume fabs.

Does TSMC have more total, all-in capacity than Intel? Yes, I believe so. But I'm specifically focusing on leading edge processes. This excludes TSMC'S large capacity in their less than competitive 10nm node and larger. This is where AMD and Intel compete. AMD is hopelessly capacity constrained due to their situation. They can't push for more capacity without tanking profits because it results in a bidding war against MUCH deeper pockets.
 
  • Haha
Reactions: scineram

Vattila

Senior member
Oct 22, 2004
568
645
136
And since AMD and Intel don't release the exact number of CPUs shipped, there is no way for us to actually determine any of this.
Well, IDC and Mercury Research are two industry research companies that track and report on the x86 market quarterly. According to IDC, in 2021-Q4, about 92 million CPU units shipped in the PC market (workstations, desktops, notebooks, but excluding tablets and servers), up 26% year-on-year. For the full year, 303 million units shipped, up 31% year-on-year.

In Q4, AMD had 21.7% share of the x86 market overall, according to Mercury Research. So that's about 20 million units in Q4 in these segments, with Intel shipping 72 million.

In addition, AMD shipped ~8 million APUs into the new consoles (4.5 million PlayStation + ~3.4 million Xbox). And Intel probably shipped a few million CPUs into the tablet segment. And both shipped units into the embedded segment, but I'll leave the numbers as a research exercise.

As for servers, IDC reported 3 million units (servers) in Q4. Per 2019 data, x86 had ~87% share in the server space. Of the x86 server pie, AMD still had just 7.1% unit share in Q4 according to Mercury (if you count them all; AMD likes to only count the segments in which it currently competes, i.e. traditional 1S and 2S servers). In my estimate that is 185 to 371 thousand EPYC CPUs, depending on the balance between 1S and 2S servers in AMD's share.

(So the server segment doesn't move the needle much in terms of capacity requirement. Although the amount of silicon going into a server chip is much higher on average, of course, than in the PC segment, AMD is shipping 54 to 108 times as many CPUs into the PC segment. Make no mistake, AMD is in no way limited by capacity in the server market. AMD's server growth trajectory is solely determined by product features and performance, customer acceptance and design wins — as well as AMD's sales team performance, of course.)

As for the scale of wafer capacity at TSMC and Samsung: The number of smartphones sold in 2020 was 1389 million. Worldwide tablet sales stood at ~160 million units. There are now also X million ARM-based laptops sold. In addition, the leading-edge capacity at these foundries is used for GPU, FPGA, AI/ML, crypto and other accelerators from a multitude of vendors. And there are countless customers using older nodes. TSMC alone serves over 460 customers.

"Samsung's and TSMC's cumulative CapEx spending to total $55.5 billion in 2021."


I hope that gives some useful information and perspective.

PS. All numbers were found by simple Internet search and the use of a calculator. Correct any that are found to be wrong.
 

Vattila

Senior member
Oct 22, 2004
568
645
136
AMD is hopelessly capacity constrained due to their situation. They can't push for more capacity without tanking profits because it results in a bidding war against MUCH deeper pockets.
Although often claimed in forums, this is incorrect. Wafers are not paid for by retained earnings. Wafer cost must be covered by revenue. The deeper pockets you are referring to are companies that are addicted to very high gross margins (Apple, Intel), and they will not accept overpaying for wafers and thereby increase cost of sales, which would have detrimental effect on their coveted margins. I addressed it in this post.

And on AMD capacity constraints, you are exaggerating and expecting too much. AMD is growing way faster than planned, and they have great wafer supply visibility to support their accelerated plans for the foreseeable future. If they can achieve the forecasted 37% growth this year, with adequate supply from TSMC, and they can continue that for another two years, they will be at 1.37 ^ 3 = 2.57 times their current size in terms of revenue. If the x86 market stays about the same in terms of units, and AMD unit growth is proportional to their revenue growth, then we are looking at 21.7% * 2.57 = 55.8% x86 market share for AMD by the end of 2023. In my view, as a shareholder, that is nothing to complain about.

So I don't see much downside in their fab-less business model. On the contrary, it is going very well.
 
Last edited:

moinmoin

Platinum Member
Jun 1, 2017
2,369
2,941
106
on relevant, leading edge performance nodes (which, as we see in benchmarks, Intel's 14nm node is one)
Sorry, but ROFLMAO.

Intel's 14nm was leading before TSMC's N7. Since N7 it's merely competitive. Since N5 it's quickly falling behind in power efficiency. So no, Intel's 14nm no longer is a leading edge node anymore.

Unless you want to ignore the energy efficiency and insist on the performance part, in which case you get a second ROFLMAO.
 

Vattila

Senior member
Oct 22, 2004
568
645
136
On Intel's IDM 2.0 business plan and their fab options, this article from Scotten Jones at SemiWiki is a great read:


Scotten Jones obviously has tremendous industry experience and insight, and he highlights many of the issues and the pros and cons with regard to Intel's options.

That said, he sounds a bit like an Intel sentimentalist, as he concludes that Intel's best course of action is to not be distracted by their new foundry business, but instead double down on getting back to glory (2 year process development cadence), which implies that Intel maintains the market domination needed to be able to fill their fabs with Intel product, i.e. maintain "critical mass", as he calls it.

On this latter point is where I think he goes wrong. Without maintaining critical mass, Intel will not generate the revenue and profit needed to keep scaling up R&D and capex to stay at the increasingly expensive leading-edge of process development and fab investment forever.

Like I pointed out in my earlier post, if AMD achieves and continues 37% growth this year and for the next two years, they will be at over 2.5 the size today in terms of revenue, and possibly at over 50% unit share, by the end of 2023. Meanwhile, Intel will have transitioned to chiplet design at that point, and much of the silicon in their "disaggregated" chips may come from outside to stay competitive, as the chance of them regaining process leadership by then is slim. And like I pointed out before, without process leadership Intel will increasingly face fierce product competition from other directions as well, such as RISC-V, ARM, GPU, FPGA, AI/ML — all threatening their ability to maintain critical mass.

I'm starting to think that Intel Foundry Services will become the IDM exit strategy at Intel. As the prospect of getting back to dominance of past in all areas continues to fade, while fab expense and process R&D continue to increase relentlessly, they will look to spin off the fabs into IFS and likely partner with a foundry to get back to critical mass.
 
Last edited:

ondma

Platinum Member
Mar 18, 2018
2,015
726
106
On Intel's IDM 2.0 business plan and their fab options, this article from Scotten Jones at SemiWiki is a great read:


Scotten Jones obviously has tremendous industry experience and insight, and he highlights many of the issues and the pros and cons for Intel's options.

That said, he sounds a bit like an Intel sentimentalist, as he concludes that Intel's best course of action is to not be distracted by their new foundry business, but instead double down on getting back to glory (2 year process development cadence), which implies that Intel maintains the market domination needed to be able to fill their fabs with Intel product, i.e. maintain "critical mass", as he calls it.

On this latter point is where I think he goes wrong. Without maintaining critical mass, Intel will not generate the revenue and profit needed to keep scaling up R&D and capex to stay at the increasingly expensive leading-edge of process development and fab investment forever.

Like I pointed out in my earlier post, if AMD achieves and continues 37% growth this year and for the next two years, they will be at over 2.5 the size today in terms of revenue, and possibly at over 50% unit share, by the end of 2023. Meanwhile, Intel will have transitioned to chiplet design at that point, and much of the silicon in their "disaggregated" chips may come from outside to stay competitive, as the chance of them regaining process leadership by then is slim. And like I pointed out before, without process leadership Intel will increasingly face fierce product competition from other directions as well, such as RISC-V, ARM, GPU, FPGA, AI/ML — all threatening their ability to maintain critical mass.

I'm starting to think that Intel Foundry Services will become the IDM exit strategy at Intel. As the prospect of getting back to leadership in all areas continues to fade, while fab expense and process R&D continue to increase relentlessly, they will look to spin off the fabs into IFS and likely partner with a foundry to get back to critical mass.
Seems like a pretty big assumption that AMD can maintain 37% growth for 3 years. They certainly will continue to gain share, but a given percentage growth is easy when starting from a low number, but becomes increasingly difficult as the base number gets larger.

That said, Geslinger's plan sound great, but I seriously doubt Intel's ability to execute it. I dont really know how you are going to go from spending 5 years on one node to regaining leadership and getting back to a tic/tock strategy with such difficulty with each new node.
 
  • Like
Reactions: Vattila

moinmoin

Platinum Member
Jun 1, 2017
2,369
2,941
106
View attachment 41745

Intel’s New IDM 2.0 Strategy: $20b for Two Fabs, Meteor Lake 7nm Tiles, New Foundry Services, IBM Collaboration, Return of IDF (anandtech.com)

Wow! Gelsinger is pretty bold in his enthusiastic and aggressive move to double down on industry domination and the risk/money he is willing to take/spend to achieve it. It is pretty scary — for AMD and Intel shareholders alike.

In the long term, should it succeed, and enable Intel to retain its dominant position and market share, then it will clearly threaten AMD's success and growth path. If it fails, then Intel will be in even bigger problems than today, having invested even more in a currently failing IDM model.

In the short term, there is nothing new in the presented plan that will adversely affect AMD's road map for the next two years, while in the face of increasing competition, Intel shareholders may see declining gross margin and revenue ($72B projected for this year, while they have been above that in their trailing twelve months revenue for the last four quarters; see Macrotrends), while spending goes up dramatically (from $14B last year to $19-20B projected this year). Which means less dollars spent on propping up the share price with stock buybacks and dividend increases.

I'm a little bit baffled about the Intel Foundry Services business. As an independently run business, it looks like a serious attempt at a foundry play. But I guess it is just a safe-guard to sell off excess capacity, to justify expansion, and as an insurance, in case they cannot keep their fabs fully utilised with their own product success. If I understand correctly, the fabs will still be in the possession of the main company, which is doubling down on the IDM model. In short, I suspect there is not much foundry ambition within Gelsinger. If he was serious about it, he would have spun off the fabs properly and partnered with an existing foundry to create a truly independent entity. I get the impression he simply wants back to Intel glory, with market share sufficient to fill their fabs with Intel product.

As pointed out by many, with this aggressive domination plan from Intel, not only does TSMC have no incentive to be helpful to Intel — it has all the incentives to be helpful to AMD.
I don't subscribe to the viewpoint that a strong Intel requires a weak AMD or vice versa so I applaud any ambitious effort. Intel's and AMD's biggest competition are the solutions that make x86 expendable (mostly ARM, but also other efforts in HPC).

That said I don't find this announcement really that ambitious, it's actually overdue clarity if anything.

If Intel does want to keep the fabs (and as I wrote before everything else I consider insanity, nobody in the market can replace Intel in quantity right now) they need to double down. IFS then is a requirement if they want to keep up with pure play foundries, as only if Intel is competitive in the foundry market they have the incentive to keep up with the competition's cadence.

The IBM partnership will be the litmus test for the IFS effort. In the ideal case Intel creates what Samsung already does, completely separate departments for foundry and silicon design business that have no dependencies on each other. It's also a big loss for GloFo as it could have been IBM's partner but threw it away.
Edit: Scotten Jones claims IBM was involved in Samsung's 5nm, which only caught up with TSMC N7's energy efficiency at lower frequencies.

Another litmus test for the future of Intel's foundries will be Intel's handling of 7nm. 10nm has been and still is an unmitigated disaster, by all indications only kept alive due to the sunk cost fallacy. Now Intel announces that the two additional fabs are set to come online for 7nm production in 2024. You may remember with 10nm Intel early on reduced fab capacity again, both to increase 14nm capacity again and to have more fab space to work on 7nm development. The bigger focus on the use of chiplets/tiles as well as candid use of external foundries gives Intel some additional wiggle room, but the return to a workable node cadence has to be imperative for Intel going forward.
Edit: Scotten Jones claims Intel's capacity for 10nm is comparable to TSMC's N7. That seems... off? Or is that just wafers ignoring actual yield? Brute forcing capacity to get a decent amount of working chips? Insane if true.

Interesting is the mention of an Intel counterpart to AMD's semi custom business. That will be very interesting regarding how much flexibility it allows its customers and how the offers stack up to what AMD delivers close to a decade now. This service may well become a competitive advantage for IFS depending on the flexibility and cost involved.

Addendum to Scotten Jones' piece: Good read for historical context, like the several resets of Intel's node cadence etc.

I'm starting to think that Intel Foundry Services will become the IDM exit strategy at Intel.
No, IFS is a critical requirement to uphold the "critical mass" forcing the node cadence onward. On the play foundry market if a foundry stops developing nodes the its fabs become noncompetitive which again leads to decreasing capacity utilization which again means a lot of uncovered costs (see Micron's 3d xpoint fab). Intel instead just dropped its Intel Custom Foundry service and proceeded milking 14nm to this day, earning massive profit at the cost of losing all the lead it originally enjoyed both in process node and silicon design. So the lack of foundry service open to external customers ensured Intel as a whole stagnated. So it's in Intel's very interest especially as an IDM to not let that happen again.
 
  • Like
Reactions: Tlh97 and Vattila

Vattila

Senior member
Oct 22, 2004
568
645
136
Seems like a pretty big assumption that AMD can maintain 37% growth for 3 years. They certainly will continue to gain share, but a given percentage growth is easy when starting from a low number, but becomes increasingly difficult as the base number gets larger.
That is true and a very good point. I just put it in there to highlight the risk Intel is facing. AMD's financial model calls for 20% CAGR in revenue over the long term. If we use that model then the 2.5x growth I used in my calculation will take about 5 years (1.2 ^ 5 = 2.49). That said, AMD did achieve 45% growth last year, starting with a forecast of 25%, and this year they have the "greatest visibility ever" according to CEO Lisa Su. So, considering the strong markets and immense demand, it is not unlikely they will outperform the 37% growth forecast. In fact, some analysts expect over 50% growth this year. Since Intel's comeback isn't really hitting until 2023, AMD should be able to keep momentum through 2022 and into 2023, at least, provided they continue to execute well.

In any case, my main point in this thread is that there is an increasing threat to Intel when it comes to maintaining critical mass in terms of product success and market share, which in my view, will put their IDM business model under increasing pressure in the coming years.
 
Last edited:

Vattila

Senior member
Oct 22, 2004
568
645
136
I don't subscribe to the viewpoint that a strong Intel requires a weak AMD or vice versa so I applaud any ambitious effort.
Agree. A growing semiconductor market that lifts all boats would be great. 50/50 in the x86 market would be cool.

In the ideal case Intel creates what Samsung already does, completely separate departments for foundry and silicon design business that have no dependencies on each other.
Makes sense. In which case, I view that as practically abandoning the IDM model and adopting the fab-less model. I don't require fab-less to mean selling the fabs to someone else. I think a joint venture with a foundry makes most sense with regard to critical mass. I can see a clear path for IFS to get there; Intel transfers control and ownership of the fabs to IFS, then spins it off as a part-owned joint venture. There would then be a multi-year wafer supply agreement.

Scotten Jones claims Intel's capacity for 10nm is comparable to TSMC's N7. That seems... off? Or is that just wafers ignoring actual yield?
He is probably referring to wafer start capacity. Intel is indeed taking a noticeable hit to their gross margins because of the poor yields on 10nm.

PS. Great post!
 
Last edited:

Doug S

Senior member
Feb 8, 2020
525
716
96
My claim is simply: on relevant, leading edge performance nodes (which, as we see in benchmarks, Intel's 14nm node is one), TSMC has substantially less capacity than Intel. TSMC has N7, it's various refinements on the same lines, and N5, in only a couple of fabs. Intel has 14nm, and 10nm in multiple variants, in nearly a dozen high volume fabs.

So you consider Intel's 14nm to be a "leading edge performance node", then why not TSMC's 10nm? Seems like your definition is based on "whatever Intel is using is a performance node, once they stop using it then it and any equivalent foundry nodes are no longer considered performance nodes".
 

jpiniero

Diamond Member
Oct 1, 2010
9,047
1,743
126
So you consider Intel's 14nm to be a "leading edge performance node", then why not TSMC's 10nm? Seems like your definition is based on "whatever Intel is using is a performance node, once they stop using it then it and any equivalent foundry nodes are no longer considered performance nodes".
10FF is a short lived node and uses the same tooling at 7FF does.
 

moinmoin

Platinum Member
Jun 1, 2017
2,369
2,941
106
Makes sense. In which case, I view that as practically abandoning the IDM model and adopting the fab-less model. I don't require fab-less to mean selling the fabs to someone else. I think a joint venture with a foundry makes most sense with regard to critical mass. I can see a clear path for IFS to get there; Intel transfers control and ownership of the fabs to IFS, then spins it off as a part-owned joint venture. There would then be a multi-year wafer supply agreement.
Who would Intel make it a joint venture with? GloFo had big oil money backing 2 high end fabs instead Intel's 15 fabs, it still bled money to the point it left the bleeding edge node development. Imagine AMD being stuck with a non-existing GloFo 7nm and no TSMC N7 to evade to, that'd effectively be Intel's position in the worst case after becoming fabless.

In my opinion Intel will want full control like an IDM (or "as an IDM", it's a matter of definition, I'd call Samsung an IDM even though technically each part of it is fully independent), but freedom for both department, exactly because especially 10nm showcased how lack of progress in either process node and silicon design development hinders each other. Development needs to be able to go on unscathed regardless of failures. Failures must not lead to more than half a decade delay again like Intel's current messy sequence from 14nm over a pretty much abandoned 10nm to a by all indications also troubled 7nm process nodes. To be able to do that the cadence must be much finer grained. For a finer grained cadence to make financial sense you need scale to a point Intel previously was neither willing to spend for nor was Intel Custom Foundry competitive enough in the open market (due to lack of standardized tools and unpredictable development, both things promised to get fixed this round with Intel Foundry Service) to garner enough external customers (that at the time Intel didn't see as necessity to begin with).
 
  • Like
Reactions: Tlh97 and Vattila

Gideon

Golden Member
Nov 27, 2007
1,214
2,270
136
My claim is simply: on relevant, leading edge performance nodes (which, as we see in benchmarks, Intel's 14nm node is one), TSMC has substantially less capacity than Intel. TSMC has N7, it's various refinements on the same lines, and N5, in only a couple of fabs. Intel has 14nm, and 10nm in multiple variants, in nearly a dozen high volume fabs.
Thats not fair to TSMC. even their 16nm is denser than Intel's 14nm so to be fair you'd have to also include TSMCs 16nm, 12nm and 10nm nodes.

As for those nodes, The Revenue from 5nm and 7nm is rapidly growing (source: TSMC's Q4 2020 report)

And they will be shipping over 120 000 5nm wafers a month in 2H2021, that's not an insignificant amount by any means.

This is how rapidly things changed in just 2 quarters, 1Q21 results will be very interesting.
1616783639983.png
 

Doug S

Senior member
Feb 8, 2020
525
716
96
Ian says that 16FF is theoretically 28.2 whereas Intel 14 is 37.5. And as I said 10FF was just a stopgap which is probally not being produced anymore.
If TSMC was as incompetent as Intel and was shipping 16FF+++++ due to an inability to get any newer processes working, it would probably be denser by now.
 
  • Haha
  • Like
Reactions: Tlh97 and moinmoin

ASK THE COMMUNITY