P
Portland
Guest
Intel is exploring options not asml.
Will Intel have enough EUV for 7nm?
- Thread starter Daniel Nenni
- Start date
As they should be, especially now that ASML is deep with TSMC.
What options would they be looking at? Re-engineering DUV similar to 10nm or some other alternative?
Push more 7nm products to TSMC N3.
kvas
Member
This does sound like a reasonable and simple plan. However, if Intel goes with it, they would be locking themselves into outsourcing cutting edge chips to TSMC for the foreseeable future. That in itself is not necessarily a bad idea, but they have just announced the whole IDM 2.0 thing, and outsourcing everything that matters wouldn't rhyme with that. At times I wonder if IDM 2.0 was just a nice story while actually they are slowly but surely going fabless.
I think Intel considers outsourcing its leading edge node while it fixes its manufacturing challenges (which as you note, isn't clear when that is fully solved) with the intention of taking back leading edge manufacturing once challenges subside. The interesting question is how much capacity TSMC allocates for Intel - the capacity may likely come at the cost of either long-term partners (NVIDIA, AMD, etc.) or capex to build out the additional capacity, which would not generate great returns if Intel is a (relatively) short-term customer.
TSMC upped CAPEX to accommodate Intel so there will be plenty of N3 for all. Intel stays on nodes for a long time so I'm sure TSMC will make plenty of money, absolutely. N3 in general should be a long node depending on how N2 GAA turns out. So far Samsung GAA is not doing well, hopefully TSMC has a better take on it.
I'm currently a TSMC shareholder so my views can be somewhat bias. That said I believe it'd be tough to claim that TSMC is currently a monopoly.
Yes, TSMC fairly recently ended their discounts and raised wafer prices (Source: https://www.taiwannews.com.tw/en/news/4081055)...but so did many of its peers (Source: https://www.taiwannews.com.tw/en/news/4209957)! There's a chip shortage after all and you can't fault TSMC for trying to maximize its profits while it's advantageous. If you're a shareholder, that's what YOU would want a company you invested in to do!
TSMC could argue that its global semiconductor market share is only 28% (Source: https://semiwiki.com/forum/index.ph...l-semiconductor-manufacturing-industry.13693/) for 40nm and below. Now, let's apply the FTC's definition of monopolization (Source: https://www.ftc.gov/tips-advice/com...ws/single-firm-conduct/monopolization-defined)..."Courts look at the firm's market share, but typically do not find monopoly power if the firm (or a group of firms acting in concert) has less than 50 percent of the sales of a particular product or service within a certain geographic area. Some courts have required much higher percentages."
I'm sure you would point to TSMC making 92% of the world's most sophisticated chips (Source: https://www.wsj.com/articles/the-wo...zmy6nstkrbi&reflink=desktopwebshare_permalink). But even here, TSMC would have a case according to the FTC..."Obtaining a monopoly by superior products, innovation, or business acumen is legal; however, the same result achieved by exclusionary or predatory acts may raise antitrust concerns." I would argue that TSMC got to where it is today through innovation, business acumen and a whole lot of luck (Thank you Samsung for screwing your relationship with Apple!). The bleeding-edge of the foundry business seems to be a winner-take-all market. It's not in TSMC's interest to sit around and wait for Samsung or Intel to catch up or leapfrog them. South Korea just announced a $450 billion push to help unseat TSMC as the world's top chipmaker (Source: https://spectrum.ieee.org/tech-talk...billion-investment-latest-in-chip-making-push). Europe just pledged more than $175 billion to develop manufacturing capacity for chips using 2nm process nodes (https://www.techspot.com/news/87907...join-forces-bolster-region-semiconductor.html). Intel has started talking and hiring for its IDM 2.0 plan. It certainly seems like TSMC will have healthy competition in the most advanced process nodes very soon!
Here's what Nvidia CEO Jensen Huang recently said about TSMC (Source: https://venturebeat.com/2021/06/12/...chain-chip-shortage-arm-deal-and-competition/)...
Daniel Nenni and many others here know the foundry business much better than me so they may have a different view than me.
I appreciate your points. Standard Oil controlled 91% of oil production in 1904. I would say 28% is the comparable figure for TSMC.
Still 28% is heading north, and is likely to breach 50%, within 5 years.
Taking a step back: NEC was dominant in the 1980s, Motorola in the 1990s, Intel in the 2000s, and Samsung was the dominant firm of the 2010s. TSMC appears to be the firm of the 2020s. So the semiconductor industry seems to rotate dominance.
The dominant firms have abused their positions. Intel paid Dell not to buy AMD. Samsung (and others) fixed DRAM prices. I would expect TSMC to abuse their position to some degree, it comes with the territory.
What is different today is the systemic influence of semiconductors on the economy. This is being felt by the chip shortage today. TSMC influence over GM, Ford, Volkswagen, it's surprising. (Also surprising: Not Toyota, they stockpiled).
I think regulation is coming.
Dominant but not a monopoly. Everyone uses the same equipment yet the results are wildly different. To me it really comes down to corporate culture and business model. If you want an analogy I think the race between the tortoise and the hare best applies.
tooLongInEDA
Well-known member
If I understand correctly (which I may not !), US antitrust legalisation exists to protect consumers (and business) from abusive behaviour from monopolies and does not attempt to prevent natural monopolies. The fact that TSMC, or Intel, or Microsoft achieves for some period a natural monopoly through their own efforts and by producing products customers want does not automatically mean there is a problem.
There are certainly arguments that Intel and Microsoft did push the envelope. I don't have first hand knowledge, but it seems that TSMC hasn't got anywhere near that level yet. Of course, as a foreign company, the trigger threshold for US regulation is likely to be lower ... . That said, the US is not a company that goes out of its way to punish success (the "tall poppy syndrome" rarely occurs).
But the US and TSMC's competitors trying to blame TSMC for their own failings in regulation (US) and technology (competitors) just doesn't cut it with me. Likewise customers who didn't plan ahead.
There are certainly arguments that Intel and Microsoft did push the envelope. I don't have first hand knowledge, but it seems that TSMC hasn't got anywhere near that level yet. Of course, as a foreign company, the trigger threshold for US regulation is likely to be lower ... . That said, the US is not a company that goes out of its way to punish success (the "tall poppy syndrome" rarely occurs).
But the US and TSMC's competitors trying to blame TSMC for their own failings in regulation (US) and technology (competitors) just doesn't cut it with me. Likewise customers who didn't plan ahead.
You are assuming TSMC does not bully its suppliers. How far is that from the truth?? If you are a TSMC supplier, I bet you will NEVER say that.
Many of you are right and I agree with the frustrations.
Intel, the lion, dumped many billions of dollars with big talents leading the EUV research at its primitive early days, but sadly, now it turns out that TSMC, the hyena, is enjoying the kill and reaping all its benefits, and Intel does not even get priority in getting the machines. It's sad, sad, sad and absolutely a huge mismanagement on the part of Intel company from top to bottom. Their arrogance will eventually kill them. They have no one to blame but themselves.
But I am still hopeful of some sort of resurrection of US semi manufacturing, be it Intel or someone else. Maybe not the same broad scale as TSMC, but hopefully catching up at the most leading edge. We certainly have the resources and the talents. We need the will. I even think we would do just fine if we have just half of the boldness of China in semi, because technologically we are still far ahead today. We can create powerful tech companies like Apple, Qualcomm, Google, why we can't compete in leading edge manufacturing?
If we swap "TSMC" here by "Apple"," Samsung", "intel". It is still valid. If we take ASML as another example of supplier, it seems a little different. I would say it is supply chain competition but not "bully".
I work with TSMC suppliers and bullying has never come up. Follow TSMC's supplier guidelines and you will do fine.
Daniel: tsmc treats you well. I worked in tsmc and supplier side. The vendor management strategy is well executed there.
tooLongInEDA
Well-known member
One more thought on this.
"Regulation" almost certainly means forcing - by some means or other - some TSMC customers to choose inferior technology from another supplier. Unless TSMC can be forced to cross-licence their process (even assuming that was possible, there is no guarantee that the licencees could produce to the same quality and yield), limiting TSMC market share means someone has to use an inferior process.
I don't see how that's a benefit to the end customers. At best you can hope that the cost of this is less than the [alleged - unproven for me] cost of TSMC monopoly pricing].
In 2014 Intel made the decision to design out EUV for 10nm because it wasn't ready. 10nm was slated for 16/17 at this point, who knew it would be 2020.. TSMC on the other hand with it's "new" node every year was able to slot in EUV as soon as it was ready. Also, Intel in a classic "frenemy" move loved to dual source and support Nikon. Europeans prefer friendsUnder ASML 2012 co-investor program, Intel invested $4.1 billion, TSMC invested about $1.4 billion, and Samsung invested just less than $1 billion in ASML to accelerate EUV development.
Why Intel invested the largest amount of money among the big three semiconductor companies and had all the insider information yet ended up to buy the least number of EUV?
It's clearly a leadership and management failure by Intel themselves.
