This take on it from the AMD subreddit makes at least for me more sense than the "evil patents" way of thinking. Though, nothing confirmed yet.

It seems strange at first, but the selection of patents and the timeline lead me to think of a very particular plausible line of events that would absolutely merit a lawsuit and also explain the delay to implementing said lawsuit.

AMD, IIRC, has access to Global Foundries' patent portfolio - enabling them to use those patents freely. Makes sense, they were once the same company and have a very tight agreement in effect for at least a few more years. AMD goes to TSMC for 28nm after GF's stumbles and takes GF's patents to TSMC to improve their process tech for AMD's needs. All legit, all legal. TSMC does not have permission to use those patents on non-AMD products, but does so anyway.

Fast forward a couple years and investigating TSMC's process as it applies to shipping products by other vendors reveals patent infringements. Good-faith efforts are made to stop the infringement, but TSMC either doesn't acknowledge the infringement or simply claims compliance and then does nothing. Ergo, the lawsuit we see today.

https://www.reddit.com/r/Amd/comments/cvqu5l/comment/ey6fje5

TMSC has always been a contract foundry with customers who have potential IP conflicts, I would be kind of surprised if they were sloppy like this.

The other side of this question is, which other foundries in the world would have tech that TSMC would be interested in?

I'm not well versed in the foundry universe but a quick guess would be a very limited number of companies. Maybe Intel, GF, Samsung and a company or two I don't know about.

Unfortunately, patent offices think the courts will sort it out if they mess up, and courts think patent offices probably did an OK job vetting, and they both defer to each other more than they probably should.

That said: there are only a handful of successful patent defense strategies, and some of them merely prevent bad things from happening to you (e.g. arguing noninfringement), others invalidate the patent (e.g. showing prior art). So, just because the process stumbles sometimes doesn't imply that patent can not encourage progress and it does not imply patents can't encourage progress in general. But it does lead to expensive lawsuits that even bona-fide in-the-right actors will settle to avoid.

To take the counterpoint: we're not seriously suggesting GF is a patent troll right? (Though a recent court case argued "patent troll" can't be defamatory because no-one knows what it means and so it must be opinion ;-))

TSMC was the first to propose the general idea to use water. Others [ASML and Nikon] made it work in actual implementations that required hundreds of other developments. TSMC doesn't even make lithography machines, it's a completely different business. I don't see how anything is unfair here.

Real world innovation and who runs to the patent office first are often two very different things.

Yet many, often very rational, people defend the process as necessary for innovation.

Much like minimum wage laws there's a lot of sub-industries where it legitimately helps (studies show fast food mega corps don't have lower employment rates after small bumps) and many where it hurts the very same people it claims to protect (unskilled youth and disabled people who just want to work in the summer/after school/part time for cash strapped small businesses for experience and pocket change, and don't benefit from a 'living wage' or nothing arrangements). I'm a big proponent of having these laws be specific to certain fields where it makes sense like biotech and not presented as a panacea that's pigeonholed onto every industry and easily exploited by companies who contribute little more than hiring good lawyers.

That's not how it works. You have to invent in order to get a patent. The first ro file rule happens when two people independently invent the same thing.

It might not be how parents work but it is how innovation works in most industries, so what are we really rewarding here?

Being first to claim it doesn't mean you're the first to "invent" something and are the only one who should be given the power who gets to turn it into a real life thing and make it actually useful.

So the reason they abandoned their 7nm process was because they were going to make money patent trolling instead? Guess it doesn't require as much investment as building fabs.

Actually producing stuff is just a liability for a patent troll, that leaves them open for counterattacks.

I think it's more likely that they abandoned their 7nm process because they couldn't figure it out, and then resorted to patent trolling to still make some money, somehow.

I have a feeling that eventually we are going to be facing basically a completely paralyzed tech industry because of what basically amounts to a patent singularity.

As more and more formerly innovating and profitable companies that invented and patented key technologies either start to struggle, get bought out, or go out of business, there will be more and more pressure to monetize their huge patent warchests, (and competing companies are even more driven to build up their own warchests as a result) we are going to see wave after wave of massive industry wide patent lawsuits like this that are going to completely disrupt things.

We absolutely need massive patent reform in the USA before things reach this point and any incentive to bring new products to market is destroyed by the tidal wave of patent suits that it is virtually guaranteed to attract.

This stuff has very extreme chilling effects, for instance a technology that has almost limitless potential (forward error correction) receives very little attention because of how crazy the patent situation surrounding it is.

Since the technology moves on so fast, shortening the patent expiration to eg. 1year would solve most of the problems.

Companies get 1 year head start (which is a lot), and after 1 year, we all benefit.

If you're suggesting a 1 year patent period for all utility patents, you might as well just get rid of patents. In the industry I work in (Biotech - DNA sequencing) it takes maybe 5 years to mature an approach and get it to market.

Once your patent is published, there isn't much to stop others developing the same method [1] (they just would run into trouble if they sold it commercially). So at the same time that the patent owner is developing their platform, a competitor can be developing theirs and launch at the same time or nearly the same time as you.

Certain areas, like software patents, I'd possibly agree. I don't think software/algorithm patents are a good idea for a number of reasons however...

[1] General consensus seems to be patents don't cover academic work. I've also heard that they don't cover commercial research, and that in any case it's hard to provide damages. http://softlib.rice.edu/ques.html#2

Patent validity should not be one size fits all like it currently is. It's hard to imagine but the patent system was actually designed to encourage growth, not slow it. The validity period should be part of the application process and awarded based on things like time-to-market and the length of the research effort involved. Pharma is an example of a long patent period and a software patent like two finger zooming should be shorter (maybe five years).

About encouraging growth or not. I'm not sure, it would be interesting to see what would happen without patents. However, I'm reasonably sure scientific startups wouldn't get VC funding as easily.

Don't get me wrong, it's totally possible that this is a good thing, and it might make it easier for companies to bootstrap if they don't need to compete with VC funded startups. But it's a very different model than we have now.

But...I don't think it's obvious that patents don't encourage growth. It's possible that they do. They do encourage disclosure to an extent (I've learned a lot reading patents). Without patents, you might find companies relying on trade secrets more.

Do you trust the patent examiners to set the validity period? I don't particularly to be honest...

> It's hard to imagine but the patent system was actually designed to encourage growth, not slow it.

And that was done by people completely oblivious to how real world economics work, and reinforced in that delusion by later generations of not much smarter statesmen.

It is not humanly possible to imagine how economic growth and efficiency is encouraged by intentional introduction of economic inefficiency through imposition of artificial scarcity (which is what patents are, literally a license on economic inefficiency)

It actually is quite humanly possible. The fundamental issue is that while replication of information is effectively free, generation of (valuable) information most certainly is not. Which takes us right back into limited resource allocation and value decision territory. And markets are the best tool we've found so far for societies to use to perform generalized efficient resource allocation towards societal goals. It's also often quite possible to keep information secret for a significant time period, but efforts to do so (and efforts in turn to gain that information) are themselves examples of the "economic inefficiency" you complain about. However, generalized efforts to restrict information require the cooperation of society overall and costs to society in a way that real physical property does not.

"Intellectual Property" is one possible way to try to answer all these points, a hack to try to couple market allocation to the information generation issue while also recognizing that it's not real property and is funded by and for the public. It offers a chance, not a certainty, to make money off an investment on information generation through convincing other parts of society to pay for it piecemeal during a limited pseudo-physical-property replication stage. In exchange, it shouldn't be secret, shouldn't cover basic facts of the world or certain other areas where additional generation incentives are unnecessary, and should have limited times taking into account the ongoing and ever increasing societal cost.

I think it's very foolish to mistake implementation issues with fundamental ones and throw out the baby with the bath water unless you've got something better to fund information generation equally effective. And I have yet to hear any proposed scheme by total-anti-IP folks to take its place, merely complaints about the current implementation. I think it'd be far more productive to first direct efforts towards fixing mismatches in historical IP in each of the 3 areas above vs modern times. On the secrecy issue for example, encrypted source code escrow should be a requirement for having copyright protection for everything that isn't open or available source already, and have-your-cake-and-eat-it Trade Secret laws should be eliminated. Patent protection for mathematics (all software patents) should be completely eliminated, copyright is more then enough already for software and better aligns incentives. And new, better ways to reflect ongoing and increasing societal costs, and the divergence in areas of generation, should be added rather than mere fixed times. Ongoing yearly renewal beyond an initial short period at non-linearly increasing cost (either fixed or an increasing percentage of revenue) would be one possible approach.

But all this seems far more useful than anything else I've seen. Total command and control at all levels of the economic stack hasn't proved any more effective for information generation than it has anywhere else in economics. IP has overall worked fairly effective, and it seems to me that the problems it has caused are more down to warts and rust that could be fixed.

> The fundamental issue is that while replication of information is effectively free, generation of (valuable) information most certainly is not.

This is a complete delusion. This is what about Goebbels said, "the bigger the lie, the easier it is to believe in."

People are so indoctrinated in that way of thinking, that some times I have to show this though experiment:

1. Set up a machine that will "destroy" the IP value of something by copying it.

2. Make it do so a billion times, and watch if the company whose IP being "infringed" goes bankrupt.

Now, will the company go bankrupt? Would they "run out" of their "IP generation resource?" Definitely not, that would be completely inconsequential to them. They nor lose nor gain anything. And then, all those guys run to make up some smart "economic theory" to explain that, often mutually contradicting, one sillier than another.

My idea is: if though process coming under IP generation is an unproductive process, that does not make money, why does the society goes nuts to encourage and facilitate that? A basic understanding of economic principles tells this is just like throwing money away.

See, US always brags of its informativeness. What was the last "big thing" an American company ever came with? I will say it was transistor, and it was more than 60 years ago, when no monstrous IP protection regime was in place.

It was not American companies who came up with: hoverboards, e ciggaretes, selfie sticks, stick computer form factor, an untold number of innovations in construction materials (MgO board is a Chinese invention, as well as FRPP piping, modern foam concrete manufacturing machinery,) an untold number of advances in materials science which allowed for $200 laptops being made from carbon fibre (A US made analogue will cost you $2000 just for the casing)

Just how many new products are coming from China in virtual anonymity and never given acknowledgement for their innovative nature are made without any consideration for getting remunerated for their "IP"

Now whose system make more innovation US or China's?

I clearly see, the moment IP protection began verging on extreme, US began stagnate in science and technology, and Chine went ahead.

>This is a complete delusion.

Um, no. It's objective fact. Whether it's a video game or R&D on drugs or semiconductors, it does in fact require the expending of limited resources. Are you for real? This stuff doesn't pop into existence out of thin air.

>This is what about Goebbels said, "the bigger the lie, the easier it is to believe in."

Ah yes going the Nazi quote angle, always the sign of someone who is definitely not a crackpot.

>1. Set up a machine that will "destroy" the IP value of something by copying it.

>2. Make it do so a billion times, and watch if the company whose IP being "infringed" goes bankrupt.

Did you, you know, actually READ anything I wrote, including the very first sentence that you yourself quoted? Because what you just described is information replication, which as I wrote in, again, that very sentence you quoted, is effectively free. It does not cover the step 0 of GENERATING that information to be copied in the first place. That's the problem part, which IP attempts to solve by hacking on synthetic limits on the unlimited part of the equation to handle allocation to the limited part.

>What was the last "big thing" an American company ever came with?

Basically the entire semiconductor and software industry? Most modern medicines? Much of modern genetic technology? Reusable rockets? Like, are you for real?

>and it was more than 60 years ago, when no monstrous IP protection regime was in place.

Uh, yeah when it comes to patents yes, IP is about the same as it was 60 years ago, beyond the Republicans continually trying to cut the budget of the PTO. That's part of the problem in fact.

I keep wondering if you're being sarcastic or engaging in satire here yet you seem to be entirely serious, but it's utter nonsense stuff.

> Um, no. It's objective fact. Whether it's a video game or R&D on drugs or semiconductors, it does in fact require the expending of limited resources. Are you for real? This stuff doesn't pop into existence out of thin air.

You can do a lot of normally valueless things for money, if somebody gives them to you for that, whether by legal compulsion or own volition. Not denying that. The question is whether giving money for that is a rational thing to do.

>It does not cover the step 0 of GENERATING that information to be copied in the first place.

I put that example exactly to demonstrate that it does not influence whether the person/business in question has his capacity "of GENERATING that information" diminished.

No, that person or a business is 100% capable continuing doing so in the same capacity.

The capacity "of GENERATING that information" gets diminished if it gets regulatory limited though IP monopoly (I copyrighted this mathematic formula/program, thus nobody else can use it in their mathematic formulas/computer programs)

China for example has its own non-insignificant pharmaceutical industry making original drugs, low profile semiconductor industry which nevertheless makes own developments, and gigantic videogame companies all with close to none IP enforcement.

> If you're suggesting a 1 year patent period for all utility patents, you might as well just get rid of patents. In the industry I work in (Biotech - DNA sequencing) it takes maybe 5 years to mature an approach and get it to market.

So have a two phase validity. Once it's in the market you have a year.

This encourages companies to withhold products from the market. In many, cases they will withhold products for the maximal allowable period before releasing them into the market. This will give them the maximum period of patent coverage, and prevent other players from releasing products. Market leaders in particular will want to extend their coverage using this technique.

Aside from this, what is "on the market" most novel scientific equipment undergoes early access. This period often lasts up to a year, usually equipment is sold at a loss or near cost. The product is validated. So... they recover none of their R&D during the patent period (making the patent largely useless).

So... perhaps you structure things such that you give this patent covered product away for free during early access. But you use the product to recognize revenue elsewhere (consumables, reagents... whatever)... Thereby extending your coverage period.

So... how will all this be policed/structured so companies can't find ways round it?

If you can find a way of enforcing it, it still doesn't work out for the companies. In my experience a scientific startup needs at least 5 years in R&D.. an then at least 5 years on the market to cover those R&D expenses.

If this is going to get funded by a VC it then needs a high enough return to cover all the VCs investments. This is then looking like a minimum of 10 years of patent coverage, and something like 20 years to make it worthwhile for VCs.

Yea, you can get rid of patents all together. Probably the investment/funding sources will need to change and that might not be a bad thing. But I can't see that what you're suggesting is workable.

You’d want some limitations on the first phase too, so that companies don’t hold onto their patents indefinitely while waiting for the perfect market conditions for their one-year monopoly.

A real solution is don't apply for patent protection. The down side is if someone else also figured it out, you don't get anything. Makes sense to me.

Yes, I agree. I'm not a fan of patents. While I understand the desire to protect ones investment, it doesn't make sense to me that it should be done at the expense of someone else's investment, since multiple people coming up with similar ideas around the same time isn't particularly uncommon and unfortunately there's no easy way to differentiate those from plagiarism.

As someone who researches at a western company, this would actually get rid of a lot of research, because business moves so slow to commercialize due to the highly regulated environment.

Regulation and legal arbitrage, where companies (particularly in Asia) can take what is newly invented on a silver platter and sell to the global market due to lack of environmental protections etc. would be all that is left.

» As someone who researches at a western company, this would actually get rid of a lot of research, because business moves so slow to commercialize due to the highly regulated environment.

I don't know about you but I'm perfectly ok with that.

I like technological progress, but I would also be okay digging ditches in case our economy crashed towards a commodities-only enterprise, so I’m strange in that regards :)

Some people think patents are required for innovation. They are mistaken. It is human nature to innovate.

I agree, but it is not human nature to take their innovation to the world. Post WW2 the US govt had tens of thousands of technology patents but no one to bring it to the world.

Patents are risk mitigation tools for funding entities to actually support commercialization. Innovation will happen always, but most of it stays on a dusty shelf a la an academic library. It’s not necessary that anything interesting and innovative (and sufficiently complex) gets sold.

Patents exist to prevent the sale of innovations. The notion that I can't make a buck without a patent because someone else will undercut me may be true, but it also says that innovations will get to market without patents.

Yes, the enforcement is the manifestation of their use as a risk mitigation tool. Some innovations will get to market without patents, complex innovations requiring some threshold dollar amount will not find funding and will not (this was learned by the world of govt. science policy over the last century).

It is, and people would keep coming up with ideas.

It would get a lot harder to put together a few hundred million dollars in funding to build the vacuum chambers, ovens, lithography machines, and other things you need to actually test your idea and get it working, though.

I would think that setting a fast increasing tax (say doubling each year) on patents would solve lots of issues. Once you do not want to pay the tax, the patent moves to public domain. Same tax should apply also to copyrighted work.

Shortening patents (when it can take years to gain traction) or making them more expensive only plays into the hands of large companies

Wouldn't the slowness of the courts and the lengths of processing lead to virtually no protection at all? If going through the courts takes sometimes 3+ years but parents expire after 1 year, every final court ruling would be retroactive.

So if you prevail you get payed retroactively for the damages. Whats wrong with that?

You can get six years back damages, and often you have a substantial portion of a patent's 20-year term in front of you. And frequently cases settle well before you are three years into a suit.

I'm sure a national emergency declaration would resolve an issue like this.

I predict this will work itself out by TSMC paying some money to GF. The amount will have some some vague connection to the validity of the case and the mood of a judge.

GF licensed 14nm from Samsung. My hunch is that they’re trying to get their hands on TSMC 7nm with cheap terms.

> My hunch is that they’re trying to get their hands on TSMC 7nm with cheap terms.

I can't imagine a less diplomatic strategy for negotiating cheap licensing terms. Good luck to them if that's the angle.

Pretty sure IBM 7nm was functionally complete but they decided not to expand facilities to support it. Why would they do it for TSMC 7nm instead of, say, licensing something from samsung again?

I think it's weird that most of the IP is developed and owned by US companies, yet everything is being built overseas. Of course I'm talking about core chips, like CPUs, SoC, RAM, flash memory, etc.

https://en.wikipedia.org/wiki/Semiconductor_industry#/media/...

I don't really understand why Taiwan has so many of those chip factories. I can understand why developing countries like China have been assembling electronic devices for so long, because the labor is cheap, but I don't think that chip making follows the same rule. Those generally are high tech factories, and I don't see why Taiwan has so many of them.

Because Taiwan recruited Morris Chang to help promote Taiwan's economy, and Morris Chang decided to found the first foundry (i.e. fab-only) in the world.

Ever since, TSMC has been more-or-less leading innovation in semiconductor manufacturing. Taiwan has a lot of high tech factories because it is a high tech country with smart people.

Taiwan started investing heavily in semiconductor industry since the late 70s. (https://en.wikipedia.org/wiki/Hsinchu_Science_Park)

I'm pretty sure that's decades before mainland China.

Why would Taiwan not have high tech factories? It has similar or higher level of development than - as a non-random example - France.

https://en.wikipedia.org/wiki/Human_Development_Index

A big reason seems to be the difficulty of financing manufacturing operations in America.

https://americanaffairsjournal.org/2019/05/financing-advance...

For CPUs specifically, many of Intel's chips are fabricated in the US: https://en.wikipedia.org/wiki/List_of_Intel_manufacturing_si...

Taiwan has also put together a pretty dominant position in industrial automation, with companies like Hiwin making the robots and components that go into those factories. They also are very close to customers in Japan, Korea, and China.

GlobalFoundries(GF) was part of AMD and now AMD relies heavily on TSMC. If this ban they are seeking goes through, does it affect AMD?

I think GF gave up on 7nm node a while ago, So maybe AMD is safe here.

Yes, and they still have AMD's orders going on for legacy products, auxiliary ICs, and interface logic.

That's entirely them giving up on existing as a proper market player, and the "farewell gift" to the industry before they self-destruct.

Having an idea of mentality of GlobalFoundries' new owners, I can say they care about having a return on their investment any way possible, and how they make that is of no importance to them.

Unlike a tech player with stake in the well being of larger ecosystem, or genuine fear of being denied the market by retaliatory lawsuits, those guys have no problem "setting the house on fire"

This is why they chose Germany out of all countries to file a patent lawsuit. Germany famously allows for instant summary injunctions in patent and other IP cases, which will be completely devastating for a company making actual physical products. Second to that, Germany controls the supply of few critical chemicals for semiconductor manufacturing, and they may opt to go for arm twisting with them too.

I can say that this is what we will see for year and years to come, unless something dramatic will finally happen to the patent system.

I wouldn't be surprised if AMD kept some kind of patent license to GlobalFoundries patents when they spun them off.

Makes sense to do so.

is it hip now to try and ban competitors from doing business :/ . grow up. where are the days of trying to beat competitors with a better product?

That time never actually existed. Microsoft, IBM, Apple, Facebook... and practically every large business since Rockefeller and before, did everything in their power, including regulatory capture, to hobble their competition.

It's the nature of doing business at scale.

The invention of the telephone, for example. https://en.wikipedia.org/wiki/Elisha_Gray_and_Alexander_Bell...

Microsoft and Apple do not make good products?

That wasn't the claim. The claim is they tried to squash any competition possibly making a better product. They've literally been convicted of that.

I was replying to- "where are the days of trying to beat competitors with a better product?" and then Apple and Microsoft were listed in response as offenders.

So they have inferior products than the competition?

> So they have inferior products than the competition?

In some cases yes, they did and had & tried to squash that competition. Less so today, since there's little competition where they dominate, but MS for example has a well documented history of doing just that[1].

I think you're conflating such behavior with not making any good products, which is obviously not the case.

1 - https://en.wikipedia.org/wiki/United_States_v._Microsoft_Cor....

I know about the past- I still tend to believe that the better product wins.

You're pointing at Internet Explorer in 2001 as to why Windows is the #1 desktop operating system today? It was a better product- with or without this shitty business practice 20 years ago.

> why Windows is the #1 desktop operating system today?

Because nobody's even attempting to compete in that space anymore. Many would argue BeOS was better, or some UNIX variant was better etc. The point is MS has sprung into its monopoly position via many at least questionable tactics & then used that position to keep others at bay. That message has been received & nowdays nobody on the desktop is even trying because going against per-installed Windows on every PC you can commonly buy is hard to beat.

Intel had a worse product when AMD came out with 64bit CPUs and so they used their position & partnerships to play dirty, Apple arguably does not play fair against Spotify etc.

That does not mean the better product never wins. Nobody said that. It's just that it hardly wins every time, like it should.

always everyone only looks at few examples such as these companies. who actually do make decent products. (good luck making better OS than them for example... )

but look at things like ZOOM. they make product which has no new features. it just works better and that's why people use it. it does exist. just not in the minds of people who numb themselves with clickbait.

Of course, back in the day people did make better operating systems than Microsoft, and Microsoft deliberately broke their other offerings to put them out of business.

And seriously, Apple’s no better. iTunes didn’t survive because it was the best way to put music on your iPod, but because they sued the competitors out of existence.

Microsoft and Apple make good products, but most of the money comes from locking you into inferior products.

I said the same thing when JK Rowling prevented me from publishing my pirated copies of Harry Potter. Why aren't I allowed to do business but she is?

Those days only existed in the stories of Horatio Alger and other capitalist mythologies.