Hacker News new | ask | show | jobs
Qualcomm to give all its 3nm AP foundry work to TSMC instead of Samsung (thelec.net)
110 points by jrwan 1572 days ago
4 comments

Well not unexpected. What is also missing in the report is that TSMC 3nm also performs better. And that is 2023, for anyone that is wondering.
>> What is also missing in the report is that TSMC 3nm also performs better.

Was recently reading some non-authoritative news that TSMC is also having yield issues on their 3nm. From the sound of it probably not as bad as Samsung. Taken together this doesn't look good for Intel's optimism toward catching up in just a few years.

>some non-authoritative news

Which is why those news are clickbait title. For a node that is expected to enter volume production by the end of 2022 / early 2023, having yield issue in Q1 2022 is completely normal. Compared to Samsung which is still having yield issues now on their current 5nm / 4nm node.

And their 5nm actual size is much closer to TSMC N7+ or Intel 10nm rather than N5
It can’t come soon enough after rounds of terrible Samsung fab results. Their latest and greatest node is still somewhere around TSMC 7nm or intel 10nm, but with worse performance characteristics and current leakage.
huh? Read somewhere, I think from wikipedia, that Samsung 3nm is not ready, and not expected to be ready before 2024.
35% yield for mobile SoCs sounds...crazy low? Is that normal for mobile SoCs or are Samsung's fabs that inefficient?
35% would be bad yields with a 650mm^2 GPU.

These are likely only around 80-100mm^2 (likely because it’s impossible to find die size numbers for recent snapdragon processors — reviewers need to solve this).

Abysmal would be damning their process with faint praise. I wonder if Samsung is even turning a profit or just trying to offset losses.

>Samsung's fabs that inefficient?

This has been the case for sometime, I pointed it out way back in 2017 on SemiWiki. They are fine on mature node though, this is only a problem on leading edge.

But credit where credit are due, Samsung didn't give up. Leading edge is hard, and that is saying the least. If you have been wondering where those Samsung NAND and DRAM profits gone, Foundry services. And if you have been asking why TSMC just doesn't hike price. Samsung has been keeping them at bay some what. And if you think TSMC is a monopoly and has no competition. You should spend a weekend doing more reading on Semi-Conductor industry.

In 6 NM crazy low in 3 Maybe reasonably, each time the process shrik tend to be more harder to sustain the yield
Note that this is for 2023, so Samsung still can improve this for mass production.
Phones with the Snapdragon 8 Gen 1 with the referenced 35% yield are launching this month (Samsung S22 is being delivered to preorders this week), so SS must have churned out quite a few wasted chips to meet the launch.
lack of competition is depressing
There is a positive feedback loop in fab progress: the more customers you have the more yield experiments you can do, the better the yields get, the more customers you get. This explains why the number of leading edge fabs is decreasing, ending up with a monopoly:

https://en.wikichip.org/wiki/technology_node#Leading_edge_tr...

I really hope Intel can get back on track and find success with IDM 2.0. Not long ago they had a process advantage… up until 10nm yield woes.
Just wait for TSMC to become fully Chinese.
Is this likely to happen ?
The Chinese government wants it to happen and probably can make it happen, but it will not be easy. China would have to take the island by force and an amphibious invasion is logistically difficult and almost certain to face heavy casualties on both sides.

US and the west place a lot of strategic importance in Taiwan so they may come to the defense, but it is unlikely to be defendable for long.

We are seemingly moving towards another Cold War: China and Russia vs US and NATO.

Those fabs are not ore mines. If you blew up a mine the ore is still there. If you blew up those fabs there is nothing left.

If Taiwan doesn't already have plans to destroy them in case of an invasion I'm sure they are the target of some missiles from outside the country.

The equipment they need to manufacture the latest chips is made by a Dutch company. The US has convinced the Dutch government to ban the export of that equipment to China. If China managed to gain control of TSMC they would loose access to that equipment.
Plus the US can trivially exfiltrate all the IP related to TSMC before China properly takes the island. Insiders will happily turn it over for any number of reasons.

The next step if China were to move on Taiwan, other than getting all the data and IP from TSMC's internal networks, would be to spin up a US division and assign that IP to that new corporation. And then the US should try to get as many of TSMC's employees to the US that they can.

Operation Paperclip TSMC.

The US didn't just convince them to, the US was a big part of the development of technology with a government/industry consortium in the late 90s (EUV LLC) with DARPA being a big part. That partnership worked with ASML under terms that are now letting the US have a big say in export controls.
Part of the equipment.

Sure an important part. But lets not forget that there is a lot of knowledge in TSMC too. And China is trying to slowly buy their engineers.

Intel and AMD (now Global Foundries) both had access to same Dutch company's product and they didn't or barley managed to keep up.

Not gonna happen. Tibet will get lots of freedom fighters courtesy of Uncle Sam and Aunty Indira. Same will happen at Xinjiang with groomed mujahideen fresh from Middle East plus American weaponry courtesy of free gift left in Afghan. China now is riped for breakage as inscribe in 5000 years of belief. One of American strategy to reduce China dominance is to accelerate breakage to 7 pieces (Xinjiang, Tibet, HK, North China, South China, Inner Mongolia and Heilonjiang). Attack on Taiwan, will have all the CCP children in America under internment similar to American Japanese during WW2. Would any parents able to meekly ignore that? Kill the emperor and save your children is a persuasive slogan. An economic blockade on the sea front and devastating India recovery of the western Aksai with Heilongjiang getting eyed upon by Putin while Japanese reconsider Manchuria. Most Chinese soldiers now are product of 1 child policy with very weak will to fight. Furthermore a breakage of world biggest dam would enough to wipe off 1/4 of Chinese capability on political and economic front. And remember the last few years strange prophecies happening in China. One notable one is June snowing in Beijing. Pandemic occuring in China history also signaling a change of dynasty. All in all, it is extremely unfavorable for Chinese to do a Putin Ukraine invasion. Putin is incredibly bright. You can go watch some of Putin's speeches on Youtube. Current Chinese leader is widelynknown on China to be not that well educated (due to Cultural Revolution) and lack eloquence. In short, attempt to take back Taiwan by force will end the China that we know now. People will care more about GPUs and CPU than the lives and prides of Chinese people.
Russia tried joining NATO multiple times and was rebuffed - this is entirely on the west. Obviously China is the new global rival to be concerned about, not Russia.
It does not seem unusual that an organization (NATO) created solely to defend/protect against one particular country would not allow that country to join forces with them.
There's a very straight-forward solution to the threat, which if the US Government were smart it would be pursuing asap.

First, we have TSMC working on building in the US, that should be continued and accelerated if possible.

Second, work with Taiwan to force TSMC to split itself into a US and Taiwan division. The US division will be granted all the same intellectual property and will be IPO'd (with TSMC retaining a super majority ownership position; if China ever gets ahold of TSMC, we'd simply force the parent to sell its majority ownership position on national security grounds). But how will Taiwan force that outcome on TSMC? Nation states have fun ways to make things like that happen when it comes to their domestic corporations. But why would Taiwan do that? The same reason they're getting TSMC to build fabs in the US (which is a political move): because we're asking nicely and the US is the only thing keeping Taiwan an independent liberal democracy. TSMC is so big and strategically important, they're a political entity as far as Taiwan is concerned, they're even more outsized vs Taiwan than Samsung is vs South Korea.

Why would Taiwan do that though? TSMC is the goosr that lays the golden egg: giving it up means America has much less reason to protect it.
it's enough for a government to say it won't buy chips manufactured abroad and presto, you have TSMC fabs being built on US ground as we speak; EU is close behind.
I think it's more likely that Taiwan would destroy their own fabs before allowing China to take them.
This is most likely a bluff.

If the invasion comes, it will come because China will think it will succeed. And if it looks like it's going to succeed, why blow your own factories? After the war, you will have to return to ordinary life, and hopefully a job. Blowing away your own factories, to spite your enemies doesn't make sense. Sure you deny China access yo them. But you deprive your people who will have to live after the war there even more.

More likely scenario is that US is going to try to get them blown up (probably not directly)

Perhaps. I agree that scuttling the factory wouldn't entirely deter China from attacking Taiwan, but it would definitely dent the value prospect of doing so. TSMC under Chinese ownership would be a remarkably powerful bartering chip (companies spend billions just getting their foot in the door around there), so I could really see it going either way. I know people from Taiwan, their general consensus is that under Chinese rule there would be no "ordinary life", and their goal is to make China respect their independence to the bitter end. There's definitely insufficient economic motive here, but the politics of the matter run much deeper than that.
Probably. Make no mistake. China wants Taiwan for tsmc.
China want Taiwan with or without TSMC. If China take over Taiwan the US will probably block ASML from doing business with TSMC.
Depends on how critical TSMC will remain, there's also chance US will have to work with PRC to ensure production continuity. IMO preserving fabs as leverage is probably top priority for all parties concerned, no one gains from it's destruction. The scenario where US controls inputs, PRC controls fabs, TW controls production is still preferable to scenario where destroyed fabs grinds advanced industries around world (mostly in the west) to a crawl.
The US cant block a european company from doing anything
Haven't people been floating the idea of Taiwan destroying TSMC's facilities in the event of an invasion?
I would imagine there would at least be a sharp decrease in yield rates. Hard to imagine all those Taiwanese TSMC workers working hard to ensure the technological dominance of mainland China.
Not sure if it's true, but I've heard that Taiwan has already rigged all of TSMCs facilities with explosives specifically to deter Chinese invasion.
Imagine the chip shortages then!
So much consolidation in a single entity. Lessons learned from covid are already forgotten?

What’s plan B?

The only thing we learn from history is that we learn nothing from history.
Fabs have been consolidating for thirty years as each node has become exponentially more expensive than the last.