I have worked for certain major semiconductor companies. At least two of them.
The question is not “Why did AMD hand over all its IP to China,” the question is “Why did our incompetent government allow this to happen and why did they turn a blind eye to the strong arming that was going on?”
A further question is: “Why did only some (always smaller) tech companies fork over the IP while very large ones did not?”
Let me tell you a story about a fab in a distant country that begins with a C. The rumor was that it was a very expensive server fab. One day, the wafer (the thing that the server chip is produced with and is extremely valuable, multi billions in the case of a server chip) went missing. They locked down the factory for 48 hours and brought in a search team, scoured the building from bottom to top and then found nothing.
Then, as if by magic, the wafer appeared three feet away from where it went missing. Industrial espionage. And that is what you get when you are big.
What happens when you are small? They put a gun to your head (metaphorically) and say “Split off a company and hand it over to entities controlled by the Chinese government. All your IP will be shared with them. They will own the market in China and send you a royalty check and have the ability to modify your IP to build custom products. If you don’t like it, fuck off, no China for you.”
What lazy, incompetent and useless western journalists have been shockingly non curious about is why certain companies like Microsoft and Apple magically don’t seem to get this treatment.
It is almost as though....some other arrangement was getting made. No lazy western journalists to my knowledge have yet bothered to investigate why it is that Intel, Microsoft etc have not been forced to do these IP sharing agreements while AMD and others have.
One day, Lazy western journalists might be curious and get off their asses from
Covering trump but somehow I doubt it.
Blaming AMD for “handing over the keys” is fucking stupid. It is the United States government’s lack of a spinal cord that is the problem. These companies had no choice. Either play the game or get shut out of the market.
Bad journalism, everywhere along with a lack of curiosity.
"Apple hasn’t provided any information on the matter and did not respond to requests for comment. But analysts said the most likely interpretation is that the company is giving Beijing access to its operating system source code in return for being able to continue to do business in China—arguably Apple’s most important market, but one that has been imperiled by regulatory obstacles"
Darwin doesn't even remotely resemble MacOS. It's the kernel and open-sourced user-land components. It resembles(but is not at all) a very old FreeBSD-ish system if loaded.
Just out of curiosity, what else if not the kernel would be interesting to copy from MacOS? I use a Mac but just don't have the inclination to learn about it under the hood (unlike Linux). Besides the various apps they include and the graphical / window system, I can think of LaunchD, drivers, but not much more that would be worth a lot to China.
Good high-level mental mode, but it's not a clean separation like that.
Think Open Source Android vs. Google's Android, or Chromium vs. Chrome. It's not just dragging and dropping some UI libraries on top. There's a ton of glue, and sticking shims into non-obvious places that make a lot of the magic happen.
That seems like the same sort of difference between source available and free software. Seems like a minute difference from a cursory glance, but it turns out the graph of dependencies and derived works is more important than the the code itself.
That and software can be easily decompiled. It's not that hard if you really want to know how something works. I learnt enough asm to modify a modern app that was essential to my work flow but had been abandoned by the developers. It didn't take me that long to learn.
A nation state can easily do it if they want. The US govt (NSA) has even devloped a reverse engineering tool and open sourced it.
Sure, but the USSR just openly ran modified versions of OS/360. https://en.m.wikipedia.org/wiki/ES_EVM China certainly has the capability to do the same to all of Microsoft's products. It's not a fundamental part of software that China isn't bilking Microsoft blazenly, there's something deeper going on.
There's very little special sauce in an operating system; at least, it's hard to hide state of the art techniques in one. Bleeding-edge hardware is chock full of trade secrets which are hard to simply copy.
Absolving AMD -- who actively made the choices in this case to pursue a market that they certainly didn't need to chase, and which would certainly end up biting them in the medium term -- while blaming "lazy western journalists", is simply incredible. How in the world it became the fault of journalists is an incredible stretch. The Trump bit makes your motivation suspect.
The bit about big versus small -- China pulls the same tact with virtually all foreign companies, even making Apple and others play the China game. Intel was expressly forbidden by congress of sell or licensing server chips in China, and AMD went in to try to eek a small profit in the void.
> One day, the wafer (the thing that the server chip is produced with and is extremely valuable, multi billions in the case of a server chip) went missing.
Nah, I can see a wafer, pre-etcing or soemthing, being way more valuable than a mask set. We as digital logicians are way more at the grace of the process/foundary engineers than we'd like to admit.
Also, you can steal important masks without anything being missing.
If I were going to steal someone's IP and make unlicensed copies, I'd put a bit more effort in upfront and steal the systemverilog source code. Should be easy enough to do - send someone to work in the IT department responsible for the source code management system - as well as having access to all the code, they'll have knowledge of what protections are in place.
On the one hand, I agree. On the other hand, the investment in the physical design itself is quite significant at the high end, and you can't see that from the HDL source. Frankly, if you're going to steal, you'd want to steal both.
> One day, the wafer (the thing that the server chip is produced with and is extremely valuable, multi billions in the case of a server chip) went missing.
LOL sorry but a wafer isn't worth billions, single digits millions, maybe 10-20 million tops
Agreed.
While I agree with OP's criticism of journalists, even as a developer with an interest in low-level programming I'm not nearly knowledgeable enough in what is ultimately an extremely niche field. I understand what all of the components are when reading about them, but know nothing about the manufacturing process, and even less about the politics that goes on.
If you know something, enlighten the right people. It might be asking too much to expect people to figure this out on their own.
The govt, rightly so, doesn't interfere with every single private deal. While the US govt is wising up to the threat China poses to the West and is blocking IP xfer, it's a balance that needs to be struck.
The US doesn't have China's system of rigid state control, and that's a good thing.
But I think there actually are many similarities between China and US. China doesn't allow some Western companies to do business in their country, but US is also banning some Chinese and Russian companies without even a court decision, just because some guy from Trump's administration have decided so. And if we compare the volume of international sanctions imposed by both countries, China will not be a winner here.
The U.S. has a better tactical position in the tariff war, but people exaggerate how dependent China is on exports to the U.S. Exports account for less than 20% of China's GDP and have been steadily declining for over a decade; exports accounts for more than 12% of U.S. GDP and have been slowly increasing.
Conservative pundits are living in the 1990s. One reason Obama had to sideline hawks to cut the Iran nuclear deal is because China (among some other countries) made it clear that they were about to walk away from sanctions with Iran unless the U.S. sat down to hammer out a nuclear deal independent of Iran's missile programs. And as Trump's trade war has proven, China was fully prepared to make good on that threat.
Here's a story from yesterday which explains why Beijing made HK Chief Executive Lam apologize to Hong Kong protesters:
TL;DR: Beijing is starting to dig in for the long-haul, changing their domestic political strategies with the expectation that they'll need to cut their U.S. export exposure sooner rather than later. Part of that is relaxing their stance toward Hong Kong. (Think about it: they made Lam apologize. That's not only a huge loss of face to her and her administration but to Beijing, and in the calculus of Chinese politics a not insignificant threat to CPC's power.)
Beijing has been carefully managing China's transition to a domestic, consumption-oriented economy since the aughts (2000s). That's the background to the corporate espionage and two steps forward, one step back trade liberalization. Now they're having to accelerate that process; it's not what they wanted, but they're fully prepared to do it over caving to U.S. demands. They're making concessions, just ones that strengthen their tactical position in the tariff war and their long-term strategic position. In effect they're doubling down--the anti-competitive behavior was always a means to an end, and they're recommitting themselves to that end.
That U.S. demands are reasonable is beside the point. Long-term it's China with the winning hand. The irony is that it's the so-called realist conservatives, who proclaim that might makes right is a fact of nature, that fail to see that according their own rules their strategy is doomed to fail.
The U.S. is entering a phase, much like the U.K. before it, of managing its fall from power. If done well (better than the U.K. managed, not they they did an especially bad job of it), the U.S. can minimize the loss of relative strength and maximize its absolute strength, improving the lot of the world in the process. But if we don't accept this basic fact and instead overplay our hand, the consequences will be devastating relative to the best possible outcome--far less relative strength and, at best, treading water in terms of domestic wealth. Right now we're literally hastening our own demise--we're accelerating our descent by setting the stage to prove yet again (after Iraq, Afghanistan, Syria, North Korea, etc) our increasing impotence while rapidly depleting the goodwill, structural leverage, and other resources needed to manage the decline.
None of these points are novel, nor even (IMO) seriously contentious. With sober consideration they're all fairly self-evident. But there's a ton of motivated thinking going on in the U.S. right now.
> the U.S. can minimize the loss of relative strength and maximize its absolute strength
It's not an optimization problem that can be neatly solved. It going to be very chaotic. While the train tracks of where the story goes have been laid out, what happens on the train is mostly madness.
So basically you want US companies and economy to be controlled by the government to the same extent that Chinas companies are? Where the government needs to approve every deal with foreign suppliers and manufacturers?
That sounds very much how our socialist economy worked, glad you've decided to try the same approach we did from 1945-1990. Maybe THIS TIME it'll actually perform better than the very successful free markets of 20th century. ;)
Seriously - the most successful US companies are global corporations which have major presence in EMEA and APAC markets. They also have rather large control over communications and news fields. What do you think will happen in those non-US markets if those corporations start to be a vessel for American propaganda and foreign policy bullying? How long do you think it'll take before they find themselves banned and pushed out of non-US markets?
> So basically you want US companies and economy to be controlled by the government to the same extent that Chinas companies are? Where the government needs to approve every deal with foreign suppliers and manufacturers?
It's not the companies that are the problem, it's the requirements being imposed on them by foreign governments. The failure is on the part of the US government in not protecting US companies from unreasonable technology transfer requirements etc.
And why would foreign governments agree to that while still tolerating US companies? Especially when the power blocks like EU have an option to side with China - especially after attacks from Trump administration?
What gain does the US think to get here except having everything spiral into regressive and dangerous pre-WW1 mercantilism?
Realpolitik international relations. Different parties having different interests doesn't mean there is nothing the US government can do to achieve better conditions for US companies.
> So basically you want US companies and economy to be controlled by the government to the same extent that Chinas companies are? Where the government needs to approve every deal with foreign suppliers and manufacturers?
I don't think that sales volume is the only metric to be taken into consideration here. No matter what degree of culpability you assign to each party, the US and China are countries with an increasing level of hostility towards one another. Why should America permit what are ostensibly American companies from openly trading high technology with an increasingly hostile nation? Especially technology which is of military significance.
I'm sure there were similar restrictions in place during the Cold War. I'm not advocating that the west returns to similar policies, but this might be the direction that we're forced to progress in for our wellbeing's sake.
> The partnership with the Santa Clara, Calif.-based chip maker was a game changer for China, which has long been unable to match the U.S.’s supercomputing power because of its inferior chips, one product the country has so far struggled to master.
Author forgot about Sunway TaihuLight, the top supercomputer in the TOP500 at the time the transfer happened, using a completely domestic CPU architecture.
> The AMD deal gave China access to state-of-the-art x86chips, which are made by only two companies in the world: AMD and Intel Corp.
Author forgot about Zhaoxin. The new LuJiaZui in particular is really interesting.
I wouldn't say that invalidates their achievements though. It's not like they took a 20+ year old architecture, built it on 14nm to bump up clock speed and were done. Improving something existing still requires you to thoroughly understand the technology. There wouldn't really have been a point in starting completely from scratch (why not even reinvent electricity while you're at it), other than some additional street cred.
Interesting. Almost all leading-edge US chip designs have a piece of their pedigree I can trace from the engineers that worked on the Alpha in that Littleton, MA facility. Did Sunway hire their people or just copy their designs?
The author also omitted that China has no bleeding edge foundries (like TSMC & Samsung's 7nm fabs) and it is unlikely they will be able to build one any time soon.
Both Intel and Global Foundries are trapped on older nodes, and China is in a slightly worse position (with their most advanced node being 14nm).
All of AMD's new cores are built on TSMC 7nm, and with outdated nodes & 3 revision old designs, China is unlikely to be able to significantly improve on the designs they licensed from AMD.
Arguably, China doesn't have the homegrown talent to do chip design. Allwinner yeets barely altered SOC designs from ARM into existence, and Hygon's chip design talent is in large part not China based.
I wouldn't say they don't have the talent. Whether they do or not, they could get it very quickly. What is holding back Chinese foundries has more to do with legal architectures than human talent. A 7nm facility is a huge investment, takes years to develop, and is very location-specific. The few corporations capable of creating such a facility are hesitant to sink the required capital, energy and tradecraft into a facility in China. The long-term risks are too much for this very sensitive business.
> The author also omitted that China has no bleeding edge foundries (like TSMC & Samsung's 7nm fabs)
Is that supposed to make Americans feel any better? :) Seriously the Chinese government still thinks of Taiwan (and on a bad day Korea) as one of its provinces and given its location it could easily become one again.
So the reason China might win the trade war is that... they'll just start a world war? That seems a little out of scope. Yes, they might invade Taiwan. That would be rather worse for everyone than if they won/lost the race for CPU dominance.
russia solved their oil supply problem anexing a neighboring country and no sign of a world war. why do you think china anexing taiwan would start one (also, remember china already sent a column of tanks there one nigth and again, nothing happened with the world)
International apathy is at an all time high, thanks to most countries dealing with internal political issues (which in most cases involve rigth wing groups and fake news)
> remember china already sent a column of tanks there one nigth
Taiwan is an island. Sending a column of tanks there isn't something the PLA could just do overnight. They'd need to get a warship past Taiwan's naval and aerial defenses in the first place before they could even think of invading with tanks.
He is just saying that Chinese dependence on Taiwan for cutting edge semiconductor manufacturing is not that big a vulnerability. Even disregarding economic self interests, cutting China off carries huge political risk for Taiwan. You don’t need to bring in world war to see his point.
It doesn't need to happen in a day. China has a gigantic spatial advantage here, and as their navy gets stronger and ours doesn't (or, worse, gets cannibalized and weakens), influence shifts. It's a good bet that as this trend continues Taiwan's choices will align more and more with Chinese interests and less and less with US interests.
And again I repeat: if you want to talk about radical changes to global geopolitical balance, you're talking about a rather different subject than "Can China Make a Good CPU?".
Yeah yeah, sure, it could happen. And if it did the last thing any of us would be concerned about is whether or not AMD shared innapropriate technology.
China can train chip designers just like Israel does, but they are on a path to end up in Intel's predicament, where they don't have the ability to fabricate competitive chips as their foundries are one to two generations outdated.
Not a coincidence the US usually has a carrier battle group somewhere between the two.
Cross straight relations is a definite hotspot, China (PRC) maintain Taiwan is still their territory.
The US has traditionally sat as a buffer to prevent China asserting that militarily but with the US becoming more isolationist and the Chinese rapid expanding their blue fleet capabilities it's only a matter of time until things get...tense.
> Not a coincidence the US usually has a carrier battle group somewhere between the two.
The USA hasn't sent an aircraft carrier through the Taiwan Strait since 2007. Barring some major geopolitical shakeup, it is fairly likely it will never do so again.
The PRC maintains Taiwan is part of their territory and the ROC maintains the mainland is part of theirs.
The US has traditionally used Taiwan as yet another military foothold in the Asia-Pacific to project and exert its own influence in the region for geopolitical objectives, the containment of China being one of them.
Why would China need to annex Taiwan when they can simply follow the HK strategy of supporting leaders who are pro-China? Let's face it, if push comes to shove, the US isn't going to do anything except make symbolic statements about Taiwanese sovereignty.
A far more likely scenario would be where the US undertakes overt military action against Iran, wasting money and resources while China sits back and consolidates power in its own sphere of influence.
I dunno, HK was legally handed over to China with international approval (albeit with China bending/breaking the 50 year agreement). Messing with Taiwan’s sovereignty would be a far far greater geopolitical problem.
Although SMIC has a rather poor track record, China is continually funding it. Tsinghua Holdings is also funding another foundry as well I cant remember its name. And they have slapped quite a few TSMC / Samsung people to work for them.
UC Berkeley receives hundreds of millions of dollars in funding from the U.S. Federal government, and some of that is certainly going to the BAR (and they are by far not the only ones). India is outright commissioning the research and design of at least four distinct RISC-V chips (SHAKTI) through IIT Madras. And (although the absolute amounts are smaller, the proportion is not really) many smaller countries through institutions like ETH Zürich are spending a lot of money on microarchitecture and IC design research, most with RISC-V involved.
When it comes to IC design more broadly, just put in Google/DDG the name of an industrial nation and a couple other related words, and enjoy the results.
I could think of more if it was my job, but I don't see why you'd dismiss it out of hand. Why would it be so personally important to you that the Chinese government is seen as greatest state sponsor of microarchitectural research using RISC-V?
Germany is investing 3 billion in deep learning accelerators and all the big automotive companies are investing large amounts aswell. They want to own the supply chain, just as they used to. There are even some army guys sitting in the negotiations (unusual for Germany).
Though I don’t trust this. I guess, this money will be accessible only to companies having dedicated departments writing grants (like the big car companies) and it will be wasted at the end. This country should start immediately increasing salaries in research institutions and stop that nonsense with many many 1 year contracts for scientific staff.
Given the number of bad articles regarding China lately (NYT,Bloomberg,WSJ), It seems that the journalists of the land of freedom have the same level of integrity than the Chinese journalist.
Simple question: Chinese money saved AMD in 2013 when the stock was less than 2.5$ and the company almost dead. Why did the USA not pay for their own technology, rather than complaining 6 years later when this investment is now becoming fruitful?
I remember wondering if AMD was about to die back then from Intel strangling it to death. They totally did this out of desperation, and I'd imagine they were not the only Western company that were faced with choosing between impending doom or fire sale to China.
It's probably worth remembering that WSJ is a Murdoch property.
> Three weeks after getting the top job, Ms. Su, a Taiwan-born New Yorker, jetted to Beijing to meet officials at China’s Ministry of Industry and Information Technology. A Chinese vice minister urged her to partner with China “to achieve mutual benefits based on AMD’s technological strength,” according to a ministry press release at the time.
I assume the message is: ZOMG, foreigners are giving away US technology!
Which is true. These dual venture deals are very stupid to make long term. China is being very smart here, while the groupthink towards outsourcing and the stock market demanding Chinese access has created their own doom.
A good subset of HN commenters defend union-busting ("right-to-work" laws), outright regulation breaking (Uber's "greyball", Airbnb) and poor domestic working conditions (Amazon fulfillment centers), if it results in "value for the shareholders" and "frees companies to innovate".
Of course, since this involves selling chips to a CCP-adjacent entity, it's no longer a story about how AMD saw a 10x boost in its stock and brought competitive new products to the market.
I'm in favor of right to work, only because I don't want to be required to join a union. Of course, I'm also in favor of much greater restrictions on Corporation "Freedom" and bringing IP legislation back to something reasonable.
China made investments in AMD when they weren’t doing well. If they wanted to, they could have hired away AMD’s team instead. Maybe people miss the notion that this particular strategy had likely kept jobs in the US in the longer term. It is a similar strategy to Geely’s ownership of Volvo.
But of course you can expect the WSJ to spin it this way.
Perhaps some might have gone, but the majority in Silicon Valley would have just stayed there, going to Intel, IBM, Apple, and other ARM makers. People don’t change countries randomly (and so drastically) because they lost their jobs.
The counterpoint is, you have to remember the timeframe when a lot of these hardware companies were not doing too well. I had many friends who got laid off around the same time. Many had mortgages. As a rule, hardware companies of this sort do not just start up.
AMD was about to go bankcrupt in 2012-13. CEO Rory Read laid off a large number of non-engineering staff and sold off real estate to stabilize the financial situation and managed to stave off imminent bankruptcy. When Dr Lisa Su took over in 2014, AMD was desperate for a cash infusion to plow into R&D for the Zen architecture.
There was no U.S. investor willing to make a $300 million investment in a company that held key CPU and GPU technology and world-class engineering expertise. Apple, Google, Microsoft could have benefited from AMD's CPU/GPU tech but they would rather make multi-billion dollar offers for companies like Groupon and Snapchat or Dr Dre's Beats.
An American company was saved by this joint venture with China while Wall Street was gleefully shorting AMD.
Yes that is correct. Abu Dhabi investment kept AMD alive for a while but by 2012-13 AMD was at rock bottom and needed
more cash to implement a turn-around. There was a lot more financial jugglery going on but to go into detail would be off context to the discussion.
> There was no U.S. investor willing to make a $300 million investment in a company that held key CPU and GPU technology and world-class engineering expertise. Apple, Google, Microsoft could have benefited from AMD's CPU/GPU tech but they would rather make multi-billion dollar offers for companies like Groupon and Snapchat or Dr Dre's Beats.
Or the US government. Given the national security implications, and investment may have been a wise thing to do.
This just smells like a rat like some others said. It's probably true that the deal while might not be essential probably did help Chinese cpu development. But it's unlikely that US govt. at the time didn't understand the implication or couldn't have stopped it if they really wanted it. And it's not like reporters are finding out about this just now.
I don't know the financial situation for AMD was at 2014 and they might have ceased to exists for all I know. But the article writes as if the money itself turned the company around. It's the R&D and design of a new chip that heralded a new era for AMD. And if I'm reading correctly the deal actually ended up being done in 2016 by which time most of the work on Zen was already done I presume.
As someone else said in another comment it is kind of shameful how investment works. Forget about China/US thing, chip development is important work and no one invested in it properly. For big cloud companies simply the pressure over intel (with a competitive AMD) should have been worth it by itself.
I'm also confused why it's even a problem if China has top class fabs. There'll be more competition, so it's better for us. Are we worried about Intel's profit margins?
A decent chunk of US's success in the past 70 years has been reliant on the US having no real competition.
By real competition I mean someone with both the same skill level and the same heft/weight.
If you look at how the US reacted when Japan was rising, it wasn't a pretty view. It only died down because Japan stagnated 2 decades ago.
China's kind of the same story, but if it stagnates at about the same level at Japan, the Chinese economy will be 2-3x the US economy.
That's super scary if you haven't competed on equal footing for more than half a century.
Obviously, China being an autocratic regime doesn't help either, or that China's national interests in the region directly conflict with the US interests there (Taiwan, South Korea, etc.).
I'd bet dollars to cents that even if China had the most pristine Swiss democracy, the US reaction would be 99% the same.
I'm sure it's just a coincidence that this is being published in multiple outlets now, when AMD is kicking Intel's butt, instead of when it was originally known. Why would anyone suspect otherwise?
> That technology is helping China in its race with the U.S. to build the first next-generation supercomputer—an essential tool for advanced civilian and military applications.
Are we entering a space-race with China for supercomputers or something? Did I miss the memo?
"Red danger" does indeed seem to be the narrative that western politicians, the US in particular, are pushing today.
There always has to be some fear-based narrative involving the boogeyman de jour - the Russians, the Chinese, Islamic extremists, cyberwar whatever. Any one will do, anything to deflect from real problems at home.
The fear of Islamic terrorism is mostly BS, and the cyberwar stuff is a manageable risk.
However, the "red danger" of the Soviets was (IMHO) an existential risk to freedom--just as the Poles why they fought for free multi-party elections. The rising influence of China, a totalitarian state, is a cause for concern. Why would you think a dictatorship growing more powerful is a good thing?
> Any one will do, anything to deflect from real problems at home.
As a society/civilization, I would think it would be possible to walk and chew gum at the same time: raise concerns about bad actors abroad and try to solve domestic problems.
> However, the "red danger" of the Soviets was (IMHO) an existential risk to freedom
"was" being the operative word, although I think the reality was more nuanced than "those commies are gonna get us!".
> The rising influence of China, a totalitarian state, is a cause for concern. Why would you think a dictatorship growing more powerful is a good thing?
It's not totalitarian state or a dictatorship. I also didn't say the rising influence of China was a good thing - but I'm not saying it's a terrible thing either.
What I do find to be of concern, is western politics, corruption, FUD spreading and war-mongering.
Well, they call them themselves a dictatorship. See Article 1 of the CN Constitution:
> Article 1. The People's Republic of China is a socialist state under the people's democratic dictatorship led by the working class and based on the alliance of workers and peasants. The socialist system is the basic system of the People's Republic of China. Sabotage of the socialist system by any organization or individual is prohibited.
China is neither totalitarian nor is it a dictatorship. It is an authoritarian single-party state, but let's not play fast and loose with classification here.
From https://en.wikipedia.org/wiki/Totalitarianism:
> Totalitarianism is a political concept of a mode of government that prohibits opposition parties, restricts individual opposition to the state and its claims, and exercises an extremely high degree of control over public and private life
> Article 1. The People's Republic of China is a socialist state under the people's democratic dictatorship led by the working class and based on the alliance of workers and peasants. The socialist system is the basic system of the People's Republic of China. Sabotage of the socialist system by any organization or individual is prohibited.
How would you propose to fix these; what system, according to you, would not have the same flaws as the current ones? There being an Us and a Them has been the way forward for most of human civilization's existence.
> an Us and a Them has been the way forward for most of human civilization's existence.
and so was slavery for a long time.
Humanity needs to shed the tribalism from which it was forged. The current political paradigm requires that the politicians to divide and conquer, because most people aren't able to really see what's real and what's propaganda (which makes propaganda so effective). I don't have a solution to this problem unfortunately, but until this is solved, world climate change or inter-planetary colonization (or any other big endeavour) won't happen.
But western countries are slowly-slowly creeping towards something else, something bordering on Orwellian.
They hype the boogeyman de jour to the nth degree, using fear, uncertainty and doubt to control the populace and justify ever-more overarching anti-privacy, anti-encryption, and mass surveillance laws. It's at the point where it seems like most people now accept such sweeping controls, because of course it's all to keep them safe.
They start wars based on faked evidence and the flimsiest of reasoning. War crimes are covered up and whistleblowers treated as the villans - non-western lives mean nothing to these people, who will do everything they can to grab more power, and enrich their coffers and that of their cronies.
So, it's not just about the choice of political system, or "Them and Us" - it's also about the profiteering, corruption, cronyism and borderline megalomaniac that have somehow become normalised. Perhaps worst of all is the simply staggering hypocrisy.
Yes you did miss the memo, but most of America did. My first awareness came about five years ago through friends who are in (or adjacent to) the HPC space.
What has surprised me most is the difference in awareness and framing between the Chinese and Americans I have spoken with (though obviously "data" is not the plural of "anecdotes").
When I speak with Chinese friends, and contemporaries in tech sectors, many view the current state of US-Chinese economic conflicts in two parts: a trade war and a related-but-separate "tech war". Speaking with Americans, among those who work in the technology sector, less than half see there being a separate "tech war", outside/beyond the context of the tariffs and related trade negotiations.
Outside of tech, no Americans I have spoken to see there being a separate tech war, and it is entirely contextualized in President Trump's current trade war framing.
As for Chinese nationals outside of tech, I have far fewer conversations. I did recently speak with a gentleman from a maritime BRI project in Africa (i.e. very removed from tech). He nonetheless seemed to hold a pride for recent Chinese technological advances as a part of his identity (or personality?) and viewed it as outside/beyond the specific incidentals of the current trade war. He was happy to be working on the BRI, but he thought that it was less important than the technology contest.
My guess is that a lot of it comes down to intentional framing and emphasis in popular news media?
Maybe because people realize that despite this "tech war" that China is apparently winning, Apple still makes the vast majority of profits in the smartphone world, and the vast majority of those profits find their way to the US. Google still.makes the vast majority of OSes installed in the world, followed by Microsoft and Apple, and the money there also largely goes to the US. Intel still makes the vast majority of money in the semi conductor world, and the vast majority of that flows to the US. Their biggest threat is ARM, a British company owned by the Japanese, also an ally and open societies.
American software companies dominate tech and internet in both mindshare and profits throughout the world, excluding China. Despite them "stealing" all this IP.
Sure, the US may not be able to compete in China, but the openness of the US meant that the rest of Asia, Europe, Africa America's were willing to buy their products.
When the US engages in nationalistic trade wars, it loses China's market which was never really available to it. But it also gives Europeans and Asians and Africans an excuse to exclude the US from their markets.
Any country wanting to do that, which I suspect a lot of them will now that Trump has decided he just wants to fight trade wars despite the US being the biggest winner of the current economic system, simply needs to point to the well known American spying to ignore American suppliers and software companies and developers.
The US may come out net better in the relationship with China but it's gonna lose a lot more. (it won't even here...China had been giving America TVs and computers and what not for pieces of paper. If you told an alien this deal they wouldnt believe the US is complaining. The US's problem is entirely that it hasn't been able to distribute this massive wealth surplus it has enjoyed throughout it's society, because it has prioritized the ability of the richest to hoard that wealth to the point they don't even have islands left to buy, instead of providing healthcare to the poorest).
Correct, China may regress further backwards under the new system that Xi has installed (and with the continued tightening down on everything across the board that he is pursuing in the name of absolute control).
Most likely the US will increase its advantage overall during this time, as China's Government puts itself into a straight-jacket along with its most talented people. As opposed to unleashing them in Deng-style with further liberalization.
China can never have a full answer in terms of exports vs the US software juggernaut. That's inherent to their system of extreme government controls and limitations. Why would anybody in the West ever use Baidu, Ctrip, or Chinese equivalents of Salesforce, Workday, ServiceNow, Slack, DocuSign, Fortinet, Github, GitLab, Splunk, Twilio, CrowdStrike, Palo Alto Networks, et al? There are dozens of booming enterprise companies in the US in that mold, China inherently can't compete with most of them globally. That also goes for the burgeoning software companies in the 'West' outside of the US, such as Atlassian and Shopify (both with $30b market caps). There is a reason why AWS is globally dominant, while China's cloud providers are lagging by five years or more (both in service offerings and sales size) - businesses in the West mostly won't go anywhere near China's cloud services and never will. It's another case of their companies being firewalled inside of China.
To the parent's point re allies, Trump is a mess, however he is gone after one or two terms. US allies generally understand this is a speed bump in a very long-term relationship. Xi's authoritarianism is practically guaranteed to increase by the year for a long time to come. It's going to get worse. Dictatorships become more aggressive and paranoid with time, rather than less.
I agree, without holding their people as slave workers they will not be able to compensate the lack of efficiency in resource allocation in the long run. People are accepting this right now because there is so much progress but if that progress stalls for a couple of years, because of some economic cycle e.g. they might lose their patience. This will be answered with more force/control of the government which in turn kills even more of the efficiency.
In my view this is the most probable scenario but there are also some advantages they have. One I find particularly valuable is that most of its government is constituted of engineers (70% if I remember correctly) vs. in the US most are some sort of Lawyers (?).
It is really interesting how this will play out in the mid/long term. I am kind of bullish on the western countries though they are all (especially here in europe) shooting their own feet by killing their most valuable asset: their free markets.
I don't think engineers have any special insight into running governments, or are even particularly likely to beat lawyers at running governments. I know it's tempting to believe this, given how many of us here are engineers, but I suspect the mentality of lawyers might be better suited to being democratic politicians than engineers and democratic politicians are clearly better suited to running countries than dictators - history is very clear on that point.
What matters in a politician is primarily gathering wisdom from around the population and combining, synthesising it into policy. Reality is complex, nobody can see the big picture, there are millions of factors that can affect any nation-level decision. See it as hand-crafted logic (dictatorship/engineers) vs machine learning (votes/lawyers). Lawyering is likely especially suited to creating decent leaders because the heart of being a (trial) lawyer is debate and being able to represent people who you may not only disagree with, but actively suspect are criminals. You have to do your best to be the devils advocate anyway - which is a good skill to have if your job is or should be primarily implementing the decisions of the electorate.
This whole story of later years, maybe it will bring more attention to the lack of progress in commercial hardware design tools, their ridiculous price, and existing efforts to create modern open source alternatives, like SymbiFlow [1], Chisel3 [2]/FIRRTL[3], The Open ROAD[4] project, etc.
:') oh noes, china also has business men and women with money who want to invest in things. lets find any reason to be negative about it and try to get it back in our western hands
The strategy summarized as "introduce a foreign technology to the market, absorb it, and then innovate to make China a leader" is doubtlessly true, and I am for re-leveling the economic playing field (so-called) between the US and China as much as the next person, but this article over-sells and under-informs in a few ways.
For starters, there is a throwaway graph (in that it is not directly addressed in the article) of the Top500 over time. As Patrick Kennedy at STH has rightly complained on over recent years [0], there is a rising issue with taking the lower-to-mid ranks of Top500 to mean anything at all. A lot of the systems are actually web hosting platforms, being temporarily leveraged to run an HPC benchmark for marketing purposes. Most recently he called out two Chinese companies (Lenovo and Sugon) as among the largest offenders.
My other nitpick is more core to the article: it is not as if no one in China knew anything about x86 chip design before talking to AMD. Though I will admit I am underinformed on the NatSec legalities, to my knowledge there has a similar x86 license-and-joint-venture structure between Zhaoxin and VIA for many years.
The article goes on to say:
>Chinese versions of AMD chips already have been rolling off production lines.
Which brings up some ominous analogues to displacement of other American products by Chinese manufacturing. That may be inaccurate imagery, though, as it isn't clear which manufacturer's fabs are being used for the Dhyana chips. I am no industry insider but from what I can tell Hygon is fabless (as is Huawei's HiSilicon, for that matter) and I could not find any clear information as to who is actually manufacturing the chips [1]. Does anyone here know?
I'll keep my uninformed speculation to myself, but knowing whose fabs and whose process technology actually led to working Hygon designs being realized in silicon is an important detail to fully tell this tale.
Now I have no doubt China is working to catch up in this area as well, but it seems that the entire fabrication ecosystem (including companies like ASML, not just Taiwanese and American fabs) represent technology gaps equal to x86 chip design. It takes many actors to put on this play, and laying it all at AMD's feet, or using phrasing like "gave away the keys to the kingdom" feels like a solid overstatement.
It’s not about control logic as supercomputers are used to simulate explosions so you don’t have to do it in the world releasing a lot of radioactivity.
But you can not put a tzar bomb on a rocket, that can be fired from a silo or even from a submarine. And to threat America and Europe your rocket need to flight 10000 miles or more and reach Mach 5 or even faster. All possible only with high yield, compact warheads, which the Chinese do not have yet.
Sure, so that takes you into 80s technology. A Nintendo switch destroys a cray y-mp on perf.
The processing power at oak ridge et al AFAIK is spent trying to recover old warheads without testing them, which is a massively harder problem.
Edit: and the Chinese have had their DD-ZF publicly for years now. I feel like they have to have warheads for it. Given where we've known about what warheads they have when, it'd make sense for that to be available currently.
They already have low yield nukes but to develop high yield, compact warheads without much nuclear testing (which can be easily detected and would undoubtedly draw critics from other countries) precise simulation on super computer is necessary. Also to solve fluid dynamics problems (for example to develop advanced jet engines or silent submarine propeller, which the Chinese desperately need), detailed simulation on super computer is absolutely a must. Now running LinPack to show off some nice PR numbers is one thing but actually writing an own efficient (and correct) simulation) on massive parallel systems is a thing of another level. It surely helps if one can draw on the massive highly optimized numerical libraries built for the x86/64 instruction set. It certainly will not hinder either in case one has stolen properties software and code from western companies!
But you can not put a tzar bomb on a rocket, that can be fired from a silo or even from a submarine. And to threat America and Europe your rocket need to flight 10000 miles or more and reach Mach 5 or even faster. All possible only with high yield, compact warheads and advanced rocketry, which the Chinese do not have yet.
The focus on supercomputing seems odd. x86 is not nearly as dominant in supercomputing as it is in servers. A quick look at the TOP500 shows many other architectures (top 3 aren't x86). Is there something a homegrown x86 chip would give China over top of the line Intel silicon? Does Intel have secret IP on the die that only the US govt can access?
If Chinese manufacturers infringe on patented technologies, then it will be easy to ban import of their products. If they don't infringe and make something innovative then there is no problem. Having more competition is better for consumers.
Also, as AMD entered the deal voluntarily, it means that it was considered profitable despite the risks.
And I don't see the problem with building a supercomputer. Is it a privilege that shouldn't be available to developing countries?
there is a flood of anti-china articles because china threatens the US' monopoly on high-tech profitmaking, and the US companies and rich people are laying the basis to protect their profits with war.
Maybe this is because I’m not a US citizen, but why do Americans dislike China so much?̊̈
I just cannot understand why China building it’s own jets is a problem.
US already builds it’s own jets and nobody hates that to the point ‘Let’s not partnership with US to not leak any info’.
Why is this particular dislike to China?̊̈ Is it just the ‘Communism is bad...’ thing?̊̈ Or is it something much more?̊̈
Americans always have someone to dislike. It was Soviet from 50s to 70s, Japan in 80s and early 90s, terrorists in Afghanistan, Iraq and Libya in 2000s, and now it's China.
I'm not sure if it is a coincidence that Americans always come across some bad guys, or finding an enemy is a necessity to justify US military industrial complex.
I dislike China for running concentration camps for Muslims, organ harvesting, social scores, no free speech, and having made Xi a dictator for life. I’m pretty sure Americans have a lot of other reasons like job losses, real estate money laundering, etc.
The US supports numerous regimes guilty of the exact things you're describing, with financial aid, military security guarantees and arms sales. What's your stance on those?
China's economic threat has magnified several fold because the US has spent the past 2 decades mired in conflicts that have been stoked by its allies in Saudi Arabia. Meanwhile, China has had a free hand in developing trade relations with its neighbouring states and African countries, further weakening the US' position there.
Whatever billions the US has earned selling weapons to KSA are a drop in the bucket compared to what it's spent in Iraq, Afghanistan, Yemen, Libya and Pakistan. These are all countries where KSA has funded extremist political movements that continue to destabilize the region.
> I just cannot understand why China building it’s own jets is a problem.
Even if China building its own jets isn't a problem, China building jets based on US designs (acquired through Israel IIRC) is a problem. That expensive design process was underwritten by the US government - i.e. by the taxpayers - to achieve the strategic goal of maintaining a certain balance of military power. If those designs then end up in the hands of our rivals, then the taxpayers' money was totally wasted.
BTW this has nothing to with whether the strategic goal is valid, whether that money should have been spent, etc. My own answers to those questions would tend to be negative. What I'm getting at is that having the designs leaked or stolen destroys even what value might have existed in the minds of proponents.
The Chinese government is authoritarian and anti-freedom of speech. As a Western person who agrees with Western values, I do not view the ascendancy of such a government to be a positive thing.
So can you explain us why Saudi Arabia an absolute monarchy that oppress at least half of its population (women), kill everyone speaking against the king is one of the best US ally?
Having such double standard undermine greatly US credibility.
There is no chance of Saudi Arabia becoming a great world power any time soon. Fear is the combination of disagreement and power. The USA may disagree with how Saudi Arabia runs things but the KSA has no real power, so it doesn't get much attention and can be an "ally of convenience".
> There is no chance of Saudi Arabia becoming a great world power any time soon
No in the conventional sense, but they have spread their ideology to such an extent that most people on earth won't draw a cartoon of a particular historical character using their real name.
Westerners dislike Chinese because they challenge us in our perceived superiority.
This goes for all aspect of society. From how we raise our children to how big ecological challenges are met.
Plus right now in the US. There is the upcoming election where I think Trump has figured out he can win by being tough on China and the slew articles you are seeing is a part of a political campaign.
Tl;dr: AMD turned itself from a domestic $5b company with 2,000 employees and no capital into a global leader in semiconductors while simultaneously enduring 30 years of crushing (and illegal) anti-competitive practices by it's primary competitor of 25 times its net worth and 50 times it's workforce strength. Now that AMD has managed to take those 2,000 American workers and push back against 80k American Intel workers + 20k Israeli Intel workers the US has decided it's not fair for "AMD to license it's highly proprietary, highly impressive, highly competitive products to undemocratic nations."
The actual manufacturing is done in Taiwan (TSMC) and South Korea (Samsung). USA can do it too (Intel, Global Foundries) but these are either not leading edge or only for private use.
Everyone uses equipment from the Netherlands (ASML).
> they couldn't manufacture ballpoint pens domestically until 2017
They were already producing 38 billion ballpoint pens a year. Just the tips had to be imported because domestic steel was too low-grade.
Since they weren't given assistance in learning steel production techniques it had to be rediscovered. Chinese steel is now high-quality enough to produce the ballpoint tips, but building up R&D is important for reasons beyond pens; the responsible state-owned-enterprise developed several new techniques, and is (apparently) the only company in the world able to produce e.g. http://www.chinadaily.com.cn/a/201904/24/WS5cbfc4a9a31048422...
> Everyone uses equipment from the Netherlands (ASML).
This is funny, because besides ASML, Europe is very light on the chip manufacturing. Sure, there's ARM, but they don't have foundries. All this talk about geopolitics and computing, chips, foundries, and every single company and thus country is dependent on this one equipment vendor from a tiny and otherwise unimportant country, and nobody seems to bat an eye.
It makes me wonder whether maybe all this geopolitical scheming and fearmongering is way overblown. Why does it matter if country Y can make chips themselves or buy them from country X? What is this, a world full of little soviet unions? I mean, we have trade, and trade works! I think in the long run trade will always win over protectionism.
High precision, high volume, low labor input manufacturing is much harder than you'd think and not exactly what China's economy specialized in until it became an explicit focus of the central government. The electronics and computing world outsourced everything to Asia back when those sectors were dirty and labor intensive. Manufactured items such as ballpoint pens (particularly the ball and tip which we the biggest stumbling block for China) were made with processes so highly automated and done so with decades of gradual refinement that there was no natural incentive for existing companies to invest in Chinese production facilities. Certain manufacturing niches were so competitive and easily optimized in the decades before global outsourcing became the rule that there were practically no benefits to moving production overseas and without the sort of "free" knowledge transfer that comes with such a move, there was no economic case for Chinese industrialists to try and build native capabilities from scratch.
Techniques involved in producing pens are unrelated to producing silicon chips. You can easily have one without the other, one is not a condition to have the other.
I guess you mean China didn't have the technological framework to produce hightech products. I am not sure I agree, but I don't have numbers to support any claim.
If you show up with bags of money, ASML will sell to you.
So yes, it's a major achievement and there's a lot of knowledge layered on top of that equipment, but it's not like they need to reinvent everything from scratch.
They're worried about foreign influence and subversion. They know chip making is a critical activity underpinning all sorts of other activities. They want to build the tech it depends on themselves so it's in their control. Maybe backdoor-free, too.
Genuinely curious - equipment-wise how much beyond the photolithography gear that ASML will sell you do you need to set up a cutting edge fab? Or are the ASML offerings pretty much one-stop shopping?
As far as I know (and anyone in the field please correct me), ASML focuses on photolithography equipment. Which leaves where you put it, how you feed it, and how you validate the output.
In a sense, it's kind of like a nuclear plant -- you can't not consider the site and physical structure as part of the operation.
I wouldn't be surprised if those parts are the secret sauce for a lot of cutting edge fabs. How do you isolate your space while moving materials about? What do you clean, when, and how?
We gave industrial engineers crap at my school, but for semiconductor fabrication the holistic approach seems necessary.
I remember a case study from the 90s where a Canadian company was able to export chopsticks to China, and undercut the domestic competition on price.
The Chinese juggernaut is based on cheap labour (or until very recently at least), if you automate something enough, they cant make an inroad. Chopsticks and pen tips are examples of things that are very highly automated.
My bic byros have not particularly precise balls, made from not particularly hard balls. If you pop one out and wash it, you can see the surface roughness and imperfections. Some other brands of pen use glass balls!
The article seems to confirm that the issue is not, in fact, an inability to produce a ballpoint pen (tip). Rather, they could not produce a competitive high-end one:
"A year ago Premier Li Keqiang went on national television and bemoaned the failure of his country to produce a good quality version of this seemingly-simple implement.
Locally-made versions felt "rough" compared to those from Germany, Switzerland and Japan, Mr Li complained."
> Premier Li Keqiang recently made a shocking revelation about the industrial capabilities of China on national television: despite the fact that the country is widely known as the “world’s factory” and produces everything from iPhones, aircraft carriers, high-speed railways to spacecraft. Until now there is not a single manufacturer in China that is able to produce the tiny rotating ball fitted to the tip of a ball pen that disperses ink as you write.
> Each of these tiny metal balls has to be imported by Chinese pen manufacturers from overseas suppliers.
I assume the 90% figure was rounded up to "all of them", potentially by Li Keqiang himself, who was probably not interested in low-quality ballpoint pens that could be made without high-precision manufacturing equipment.
I can't find an original source for that. Maybe he said it, maybe there have been several misquotes. More anecdotally, we all know how much people like to mock Chinese manufacturing as purely copy-cat and low quality. You can almost see the glee dripping off the articles about the ballpoint. I'm pretty sure you can imagine someone with a failed ball-point saying "probably made in China". This doesn't mean it's untrue, but it does mean there's likely a bias.
It's not entirely unbelievable that they couldn't do this to a sufficient standard at all. I'm just saying it's more believable that they could, but couldn't do it to an excellent standard.
The question is not “Why did AMD hand over all its IP to China,” the question is “Why did our incompetent government allow this to happen and why did they turn a blind eye to the strong arming that was going on?”
A further question is: “Why did only some (always smaller) tech companies fork over the IP while very large ones did not?”
Let me tell you a story about a fab in a distant country that begins with a C. The rumor was that it was a very expensive server fab. One day, the wafer (the thing that the server chip is produced with and is extremely valuable, multi billions in the case of a server chip) went missing. They locked down the factory for 48 hours and brought in a search team, scoured the building from bottom to top and then found nothing.
Then, as if by magic, the wafer appeared three feet away from where it went missing. Industrial espionage. And that is what you get when you are big.
What happens when you are small? They put a gun to your head (metaphorically) and say “Split off a company and hand it over to entities controlled by the Chinese government. All your IP will be shared with them. They will own the market in China and send you a royalty check and have the ability to modify your IP to build custom products. If you don’t like it, fuck off, no China for you.”
What lazy, incompetent and useless western journalists have been shockingly non curious about is why certain companies like Microsoft and Apple magically don’t seem to get this treatment.
It is almost as though....some other arrangement was getting made. No lazy western journalists to my knowledge have yet bothered to investigate why it is that Intel, Microsoft etc have not been forced to do these IP sharing agreements while AMD and others have.
One day, Lazy western journalists might be curious and get off their asses from Covering trump but somehow I doubt it.
Blaming AMD for “handing over the keys” is fucking stupid. It is the United States government’s lack of a spinal cord that is the problem. These companies had no choice. Either play the game or get shut out of the market.
Bad journalism, everywhere along with a lack of curiosity.