That was the original plan when he took the corporate iSub, but got lost along the way because a long deleted map was restored after the recent software update. Eventually he had to go back to Cupertino and take a regular plane
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple COO Jeff Williams Reportedly Visits Taiwan to Secure 2nm Chips
- Thread starter MacRumors
- Start date
- Sort by reaction score
Business is still done in person for a deal like this.
Out of respect for TSMC to show good faith and that TSMC feels properly treated and respected, and, for Apple's optics for investor, client, and other potential clients, going in person (and making that trip public) for closing a significant deal (in this case billions of dollars) is a no brainer.
Out of respect for TSMC to show good faith and that TSMC feels properly treated and respected, and, for Apple's optics for investor, client, and other potential clients, going in person (and making that trip public) for closing a significant deal (in this case billions of dollars) is a no brainer.
HobeSoundDarryl
macrumors G5
Maybe, but the point was not so much about having multiple vendors but a "what if?" about the loss of that production. It's not like any event that shuts off TSMC supply could result in Apple just continuing production with Samsung or Intel or others. There would be this sizable gap in time for the ALT option to "catch up" to what would have been TSMC production.
I would imagine that must be their contingency plan if China goes belligerent.
Because costs of labor would be so much higher in the U.S. (even Arizona), I figure that would be a last resort.
I would speculate that Apple would want to have TSMC's expertise, executive or otherwise, one way or another. They should "learn" TSMC's fab know-how, and have the capability to either build their own fab to develop their chips, or partner with Intel (shudder) or another partner who could take that expertise and get it done. Even with the best of timetables, this would be a 5-10 year effort, minimum.
HobeSoundDarryl
macrumors G5
I had no trouble seeing the linked post this morning when I posted but yes, now I go there and it is paywalled for me too. However, there are a thousand articles online about China's ambitions on Taiwan.
I agree with the positive benefits within your response but it doesn't change the reality. If any scenario interrupts TSMC production, all of Apple's prime products don't have "fresh brains"... and there is nobody else to step up and deliver equivalent brains for likely some period of time.
This is not a case of having OS-X running on PowerPC but also having a version running on X86 as a secret backup. In those days, if anything had killed the source of PowerPC, Apple could have rolled out the X86 Macs whenever they liked.
There is no backup here. As a thought experiment: take out TSMC for whatever reason. Then what? And how long for "next man up" to deliver. And is that going from a target of 2nm to 7nm from replacement partner? Or waiting for partner to replicate the 2nm capability and thus having a LONG delay between new hardware availability? It's all just ugly in that "what if?"
HobeSoundDarryl
macrumors G5
Marketing will never let the laws of physics get in the way of sales-driving copy. So when we get too near to the minimum, the goal posts will get moved and something else will get the new focus. For example, we used to really, really, really care about maximum POWER of Macs when Intel was the brains inside. However, Silicon couldn't always win that contest, so the goalposts were shifted to PPW, where silicon can always win. Once Apple starting spinning PPW over Power, "we" just rolled with them... making PPW arguments even for desktop Macs, where battery life has no importance at all.
Similarly, spin like "one handed use" was gigantic in terms of bashing phablet-sized phones when 3.5" and then 4" were the "perfect" size screens... until Apple went phablet themselves and now we ridicule anyone wishing for smaller phones than what Apple sells. Hands did not grow overnight. Instead, the goalposts were moved to highlight what was to be sold vs. what had been sold... as if we forgot about a million posts about one handed use, fragmentation, abominations, pants with bigger pockets and man purses.
So Marketing will move the goal posts to something else... or take whatever is the last bit of space for shrink and change the measure. Maybe angstroms take over. And when angstroms are near their final measures, there's even tinier fractional measures below those. Pico, Femto? And we'll make up another if needed to keep this train running.
A simple analogy is to think of things in terms of dollars. Marketing measuring things in dollars would work from say- $100 down to $1 over a hundred generations of releases. OMG, what happens after it gets to $1? Switch the measure to cents for another 100 generations. But what happens when it gets to 1 cent? Switch to tenths of a cent like that last number for a gallon of gas... to get 10 more generations... or just go ahead and go to hundreds or thousands for another hundred or another thousand generations. Easy!
Last edited:
I’d think the person at the top of making the chip orders would visit the customer face to face to work out some of the many details. There are probably well over 1,000 people involved in this process.
I know that you guys think that TSMC just builds a factory and then Apple orders a bunch of chips, but that’s not how it works. It’s critical for both parties to align on volumes and schedules and those schedules are very complex. They don’t just need to have the fab working, there are chemicals and raw silicon and packaging materials and substrate compounds and so forth.
He's an Engineer/MBA and COO with previous experience at IBM, and with a low key and very hands-on approach, apparently well-appreciated by most--doesn't sound like a "bozo". "He's not a product guy"? He shepherded the Apple Watch design through development. Seems like it's a success--or maybe to you about 35% of the smartwatch market is a failure.
The Chief Operations Officer probably is the one that signs off in chip orders at Apple. And who's to say other engineers arent there in this case?
The Apple employees that need to be aware or, or there for, the logistics and technical peice are either already there, have been there working with TSMC for years, and or will probably be shuttling back and forth to there, unannounced over the months ahead, as they have for previous chips, and for ongoing client relations and liasing with TSMC.
Original point I was making is that the COO (or any high level Apple employee going to sign this or any other deal with suppliers or partners) going over there to officially sign the deal is largely for optics as a public /industry display of professional courtesy for its business partner.
Showing that respect still matters.
If TSMC is shuts off production, not many companies will ship products. And foundry isn’t something you diversify. Samsung and Intel don’t have capability nor scale to match TSMC. If TSMC goes down, there won’t be AMd/Nvidia/Apple among others. Why would TSMC stop production? If we are talking about dooms day scenario, you can assume Intel and Samsung will be done too.
TSMC has around 100+ engineers working closely aT Apple to ensure smooth integration/testing before they churn out chips in large volumes. Ita lot more involved then ordering something like a camera, or memory chips.
Errr. Not really what happened. Intel had problems getting Xeon SP4 running on their own fab ( went through
" ...
With a total of 12 (!) steppings, they haven’t exactly covered themselves with glory either, and I can’t remember any project so far that needed so many steppings before you could even use it to some extent. ..."
Never Ending Story: Intel’s Sapphire Rapids (maybe) comes in the 12th Stepping - Planned shipping, a growing bug list and possible availability | Exclusive | igor´sLAB
Intel’s new Xeons, i.e. the scalable server processors of the 4th generation, are delayed again. Once again, one might say, and the reasons are very diverse. The problems and errors are collected…
www.igorslab.de
Intel had a phase were they were trying multiple strategies all at the same time. They diluted their focus too broad ( sometimes too much cash leads to too many projects. ) and lots of projects slid into longer delays. The particular fab process was the primary factor. The only coupled factor there was that Intel was shifting to using more 'open' design tools ( and less 100% internal custom stuff). That really didn't just apply to TSMC; design to their own future fab processes was linked to that also.
Making several product lines very chunky chiplets and pour "3D" construction ( EMIB & Foveros ) on several packages didn't speed things up in the short-to-intermediate term.
Intel has also 'flip flopped' a bit on exclusively doing Arrow Lake on TSMC N3B. There are reportedly Intel 20A alterantives f
" ...
In the case of Core Ultra 5 240F, a change of CPU tiles may also mean a change of fabrication node. The 6+8 version is said to be using Intel 20A, while the 8+16 is rumored to use either Intel 20A or TSMC 3 nm node. ..."
Intel Core Ultra 5 240F "Arrow Lake-S" CPU to use either 8+16 or 6+8 core tiles - VideoCardz.com
Core Ultra 5 240F, expected to use two versions of the Arrow Lake-S silicon A new rumor suggests Intel will continue its ‘tradition’ to mix silicon dies within the mid-range CPU segment. The next-gen Arrow Lake-S series is on track to debut on the desktop platform this year, introducing the...
Lunar Lake (which is also a Ultra 200 series variant) is also deeply committed to TSMC N3B so there is more than a decent chance there is a N3B 'CPU+GPU' variant tile there also. So decent chance some 'overhead' here is the time to do the double design work (not can't design for TSMC N3. The GPU stuff was always on a TSMC track . So the notion that Intel had no skill set exposure there is dubious. )
[ Intel doing Lion Cove / Skymount on two different fab processes is expensive ( not going to help Intel's short term margins). But it likely will the an insightful "learn" process ( paying for an expensive education. ) . The other problem is Intel just can't make enough volume EUV wafer to do 'everything' in-house. ]
Back in 2023
Intel Reportedly Delays TSMC 3nm Orders for 15th Gen Arrow Lake CPUs
Intel is slightly adjusting its Arrow Lake rollout as it looks to cut costs and navigate weakened demand.
There has been reports that tried to couple "14th gen" Meteor Lake's GPU tile to 3nm , but those are clearly not correct at this point ( It has shipped without and isn't classified "14th gen" ).
And if Meteor Lake doesn't come out "on time" it was going to be very hard for Arrow Lake (the follow on) to do launch on time also.
Whether Launch means "shipping in Q3 also" is a bit up in the air:
Intel confirms Core Ultra 200V "Lunar Lake" is on track for Q3 2024 launch, Arrow Lake update at Computex - VideoCardz.com
Core Ultra 200V “Lunar Lake” arrives next quarter The first Lion Cove and Skymont-based CPU series coming soon, also the first Xe2 based product. Intel is sharing new details about its upcoming client architecture, Lunar Lake, aiming to compete with today’s Microsoft’s and Qualcomm’s next-gen...
videocardz.com
but since the lead time on "baking" N3B chips is relatively long there is likely already in production. Even if wanted them in Q4 would need to have started now to have any substantive quantities by then.
MediaTek launched on TSMC N4 before Apple did. Samsung has their own S3 process to shepard.
If not blocked, Huawei probably would be (the used to be first in line with Apple until they were restricted).
AmpereComputing just announced they are ready to go.
" James and Wittich also both highlighted the company’s upcoming new AmpereOne® platform by announcing a 12-channel 256 core CPU is ready to go on the N3 process node. ..."
Ampere Scales AmpereOne Product Family to 256 Cores, Announces Joint Effort with Qualcomm on CPUs and Accelerators
Ampere Computing® today released its annual update on upcoming products and milestones, highlighting the company’s continued innovation and invention around sustainable, power efficient computing for the Cloud and AI. The company also announced that they are working with Qualcomm on a joint...
amperecomputing.com
AMD appears to be going for "quantity" over "density" so the Zen 5 TSMC N4 variants are going to be the 2024 production targets. If pressed AMD could have tried , but they can still take share away from Intel with N4 variants so that is a safer route for them. And it looks like although AmpereOne is ready to go they are not going to ship a large quantity of the N3 variant this calendar year. AMD may have pick some speciality chiplet for early N2 though ( as competition in AI/ML is only going to get tougher in 2024-245).
Arm's "ready to go" Neoverse CSS chiplets might be ready for some TSMC N3 variant by end of year.
"...
Finally, underscoring the quick turnaround times that Arm is envisioning with the V3 CSS IP, the company is already announcing a design win with Socionext, who is designing a 32 core V3 CSS chiplet to be fabbed at TSMC. ..."
AnandTech Forums: Technology, Hardware, Software, and Deals
Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
www.anandtech.com
It wouldn't be surprising if that 'waited' until N3P was in production though. That would bring along Amazon and Google since they are leveraging Arm Neoverse ecosystem. Arm has a working relationship with TSMC to cooperatively develop Arm chiplets going forward. Arm doesn't put in the wafer orders but being 'ready' for orders is more layered on Arm.
Until Apple is ready to dump them. Never a good idea to depend too much on Apple always keep all doors open.
no, no backup, but that is/was a business decision and informed risk taking. And, Apple is not alone, neither NVIDIA nor Qualcomm nor AMD nor Tesla (just to name a few) have dual sourcing for their high-end products.
TSMC and all the other foundries have been very reliable in terms of output, even including when there were natural disasters, for decades now.
If I take TSMC "out" for political reasons, some risk mitigation is underway, namely TSMC building 2 fabs in AZ and possibly 1 in Europe. But again, if that were to happen, we have much bigger worries than these chips ... what if Samsung, or GF, or Intel would be "out" for whatever reason, same issue, maybe not for some high profile chis/companies but the impact to our lives would be very measurable ...
Everything depends on China and the Chinese, one way or another, whether they live in Taiwan or in Mainland China.
If TSMC disappears then their new ASML EUV fab machines don't have anywhere to go ( so Samsung and Intel can get more). It isn't the loss of "2nm" that is core issue. It would be the 'volume' of 2nm. The production capacity of ASML is limited so it would take a long while to replace all of those machines if there were destroyed.
If TSMC 'disappeared' because they were 'acquired' the production capacity is still there. Lots of folks stopped buying Russian oil a couple of years ago , but other folks just bought more. For the most part there was just a slightly more redistribution of resources.
It wouldn't be that hard for companies to just not have yearly product releases. If more folks kept their phones for an additional year generally the sky wouldn't fall.
It’s not that easy, unless TSMC slacks off like Intel. Apple isn’t in foundry business, and there is literally no one who can match TSMC right now. Nvidia/AMD also use TSMC, in fact AMD is their second largest customer, and Nvidia isn’t too far behind. Samsung nor Intel can match up well with TSMC, unless they fumble badly in the future.
