Will there ever be a new CPU manufacturer?

[Buffu]

Currently in desktop CPU manufacturing, there are only 2 companies that hold practical all of market. I have read it is because Intel and AMD own licenses to x86 and x64 architecture ( Let's assume VIA is non-existent).
Now if a new company wants to make CPUs they need to make some new CPU architecture or get the licenses, as far getting licenses goes I don't think Intel or AMD, particularly Intel, will ever want a competitor to the market. So only option left is to make a new architecture which is also impossible as it would require entire uplift of PC market like new motherboards, compilers, OSs and other essential software. If that is not enough that new architecture need to have significant improvements over existing ones for obvious reasons.

In conclusion, I don't see any of these things happening unless tech giants like Apple, Google or hardware manufacturers like Qualcomm (I named them because they make mobile CPUs and are only inhibited by Intel into entering PC market) made it their "do whatever it takes" goal and invested sh!t ton of money into it.

So the question is whether it is possible to break monopolies of Intel and AMD in CPU market ? Or we are struck with AMD and Intel until end of time ?

 

[rcgldr]

In the past, Apple desktops were based on Motorola cpu's and later Power PC cpu's, but by 2006, Apple switched to X86 / X64 cpu's.

 

[Buffu]

In the past, Apple desktops were based on Motorola cpu's and later Power PC cpu's, but by 2006, Apple switched to X86 / X64 cpu's.

Still that is a decade old.

 

[jim mcnamara]

Other cpus:
Sparc stations - sparc; Itanium HP - IA64(discontinued). Sparc is used in desktops and small midrange systems, itanium sometimes in desktop special designs.

I don't think there is all that much of a monopoly, especially considering AMD currently faring quite well against Intel. PC's in general are lagging in sales
http://www.amd.com/en-us/press-releases/Pages/press-release-2017jan31.aspx
http://www.oracle.com/technetwork/server-storage/sun-sparc-enterprise/overview/index.html - some sparc servers are in the desktop cost range.

 

[Buffu]

Sparc stations - sparc; Itanium HP - IA64(discontinued). Sparc is used in desktops and small midrange systems, itanium sometimes in desktop special designs.

Itanium was developed by Intel, so we should leave it.

I don't think there is all that much of a monopoly, especially considering AMD currently faring quite well against Intel.

Yes AMD got 10% increase in market shares in this year's Q2 (Good News I guess) but for last 5 -9 years AMD has only got 20-25% market share, I guess after the launch of Core 2 Duo processors. That certainly is monopoly by Intel. I think AMD needs at least 40-45% market share for breaking monopoly of Intel.

Performance wise also AMD is lagging behind, Ryzen is certainly a good news though.

PC's in general are lagging in sales
http://www.amd.com/en-us/press-releases/Pages/press-release-2017jan31.aspx
http://www.oracle.com/technetwork/server-storage/sun-sparc-enterprise/overview/index.html - some sparc servers are in the desktop cost range.

Not a big deal, smartphones can't compete with today's PCs.

 

[phinds]

The problem with introducing a new CPU is not the technological challenge. For example Nvidia produces GPU's with thousands of cores and could enter the general purpose CPU market if they chose to compete in that space. The problem is who's going to buy them? First, somebody has to take on the challenge of writing an operating system to compete with Windows/Unix/Linux. Who's going to do that on spec? Who's going to write the apps? It's just too big an entry challenge to potential new CPU makers.

 

[Buffu]

The problem with introducing a new CPU is not the technological challenge. For example Nvidia produces GPU's with thousands of cores and could enter the general purpose CPU market if they chose to compete in that space. The problem is who's going to buy them? First, somebody has to take on the challenge of writing an operating system to compete with Windows/Unix/Linux. Who's going to do that on spec? Who's going to write the apps? It's just too big an entry challenge to potential new CPU makers.

That basically if saying that somebody needs to make another architecture because AMD and Intel are not going to give their licenses to them. Should not these licenses should be free for other companies to use since they are so ubiquitous and important ?

An interesting fact about Nvidia is that it is no less of a monopoly in GPU market than Intel in CPU against a single common competitor AMD/ATI. From what I have read, in GPU market technological challenges are the main barrier.

 

[phinds]

That basically if saying that somebody needs to make another architecture because AMD and Intel are not going to give their licenses to them. Should not these licenses should be free for other companies to use since they are so ubiquitous and important ?

Well, if we were in Russia or China or some other communistic country, maybe. I think here in the land of capitalism, that's a terrible idea. Do you think all companies should give away their intellectual property if they become one of a few big players in a particular arena?

 

[jim mcnamara]

Moved to General Discussion.

 

[DrZoidberg]

Especially in a capitalistic economy it's important to make sure that there is a enough competition. Once a patented technology becomes so important that giving a single company the monopoly over using it leads to that company having a large monopoly in the economy, it becomes a huge problem.
There already are antitrust laws that allow a big company to be broken up forcefully to prevent it from having too big of a monopoly. Naturally it should also be possible to take away some of a companies patents and make them public domain in the interest of a free open market.

Do you think all companies should give away their intellectual property if they become one of a few big players in a particular arena?

If it's necessary to make sure there is enough competition, then yes, of course they should.

 

[dipole]

There are lots of applications for new CPU designs, and many exist, but they're mostly in high-performance computing, which is generally restricted to scientific, business and industrial usage.

So they're not the kind of thing people are going to use at home, but they will be used by people who need extreme performance and can spend millions of dollars to achieve it.

 

[russ_watters]

Especially in a capitalistic economy it's important to make sure that there is a enough competition. Once a patented technology becomes so important that giving a single company the monopoly over using it leads to that company having a large monopoly in the economy, it becomes a huge problem.

Um...by nature of what a patent is, every patented product is monopolized by its creator/owner.

 

[DrZoidberg]

Um...by nature of what a patent is, every patented product is monopolized by its creator/owner.

Exactly. And as I said, once this leads to a large monopoly on the market it poses a serious problem.

 

[256bits]

Exactly. And as I said, once this leads to a large monopoly on the market it poses a serious problem.

Just out of curiosity, with a monopoly, what would the serious problem be?

 

[russ_watters]

Exactly. And as I said, once this leads to a large monopoly on the market it poses a serious problem.

So patent protection should only be for mediocre ideas? Wouldn't that stifle innovation if good ideas weren't patentable? That doesn't really make a lot of sense. In any case, what do you mean by "serious problem"? Perhaps more to the point; what benefit do you see in bringing in another cpu manufacturer?

 

[DrZoidberg]

Just out of curiosity, with a monopoly, what would the serious problem be?

Are you asking me here what the downsides of a monopoly are? Well, this article about US antitrust law for example mentions many important points. But you could also look at this summary of the biggest problems. Interestingly that site also shows that there can be exceptions where monopolies are beneficial or even necessary but those are rare and CPU manufacturers certainly don't qualify for such an exception.

So patent protection should only be for mediocre ideas? Wouldn't that stifle innovation if good ideas weren't patentable? That doesn't really make a lot of sense. In any case, what do you mean by "serious problem"? Perhaps more to the point; what benefit do you see in bringing in another cpu manufacturer?

Quite the contrary. Patent protection should only apply for really good ideas of course. In my opinion trivial or even mediocre ideas shouldn't be patentable at all.
Whether protection of good ideas has a positive or negative impact on innovation is a difficult question by the way. Companies making money of of patents will of course claim that their effect is positive overall but actually there is no conclusive evidence either way. However there are cases where even a patent on a good idea has a clear negative effect e.g. CPUs or things like audio/video codecs or web technologies. There is a reason why many technology companies prefer free open standards over proprietary formats. When big companies like Google/Alphabet want to introduce new web technologies, they have to make them open/free otherwise they would not be accepted by the industry and for very good reason. No one likes to pay royalties for something just because it's the industry standard and they are forced to use it even though they themselves or some other company could easily develop something comparable or better. This is also a form of monopoly btw. because there is no real choice.

And the reason for bringing in other manufacturers is of course to increase competition, which should bring down prices and improve the rate of innovation.

 

[gleem]

The problem with introducing a new CPU is not the technological challenge. For example Nvidia produces GPU's with thousands of cores and could enter the general purpose CPU market if they chose to compete in that space. The problem is who's going to buy them? First, somebody has to take on the challenge of writing an operating system to compete with Windows/Unix/Linux. Who's going to do that on spec? Who's going to write the apps? It's just too big an entry challenge to potential new CPU makers.

Intel lost out to ARM in the mobile business. Now it is reacting to the AI revolution in which Nividia has a leading role. More users of computing technology are going to task specific accelerators for their speed like application-specific integrated circuits, or field-programmable gate arrays . Intel is addressing this by building a more powerful CPU to work with chips like Altera's FPGA.

This article discusses the above in more detail. At this point the winner is ...?

 

[256bits]

Are you asking me here what the downsides of a monopoly are? Well, this article about US antitrust law for example mentions many important points. But you could also look at this summary of the biggest problems. Interestingly that site also shows that there can be exceptions where monopolies are beneficial or even necessary but those are rare and CPU manufacturers certainly don't qualify for such an exception.

My opinion falls along these lines from the wiki article:

Alan Greenspan argues that the very existence of antitrust laws discourages businessmen from some activities that might be socially useful out of fear that their business actions will be determined illegal and dismantled by government. In his essay entitled Antitrust, he says: "No one will ever know what new products, processes, machines, and cost-saving mergers failed to come into existence, killed by the Sherman Act before they were born. No one can ever compute the price that all of us have paid for that Act which, by inducing less effective use of capital, has kept our standard of living lower than would otherwise have been possible." Those, like Greenspan, who oppose antitrust tend not to support competition as an end in itself but for its results—low prices. As long as a monopoly is not a coercive monopoly where a firm is securely insulated from potential competition, it is argued that the firm must keep prices low in order to discourage competition from arising. Hence, legal action is uncalled for and wrongly harms the firm and consumers.[57]

Thomas DiLorenzo, an adherent of the Austrian School of economics, found that the "trusts" of the late 19th century were dropping their prices faster than the rest of the economy, and he holds that they were not monopolists at all.[58] Ayn Rand, the American writer, provides a moral argument against antitrust laws. She holds that these laws in principle criminalize any person engaged in making a business successful, and, thus, are gross violations of their individual expectations.[59] Such laissez faire advocates suggest that only a coercive monopoly should be broken up, that is the persistent, exclusive control of a vitally needed resource, good, or service such that the community is at the mercy of the controller, and where there are no suppliers of the same or substitute goods to which the consumer can turn. In such a monopoly, the monopolist is able to make pricing and production decisions without an eye on competitive market forces and is able to curtail production to price-gouge consumers. Laissez-faire advocates argue that such a monopoly can only come about through the use of physical coercion or fraudulent means by the corporation or by government intervention and that there is no case of a coercive monopoly ever existing that was not the result of government policies.

Judge Robert Bork's writings on antitrust law (particularly The Antitrust Paradox), along with those of Richard Posner and other law and economics thinkers, were heavily influential in causing a shift in the U.S. Supreme Court's approach to antitrust laws since the 1970s, to be focused solely on what is best for the consumer rather than the company's practices

A monopoly is not necessarily an evil, and very few would not have any competition of some sort. Any business or venture always should always be looking over its back to see what is coming up, and be innovative to not be overwhelmed. Having a bold assumption that they cannot be touched is to the detriment in the long run, for any venture whether a monopoly or not. The AT&T research arm, Bell Laboratories was financed by a so-called monopoly, https://en.wikipedia.org/wiki/Bell_Labs, see the research done before the breakup of the company by Reagan.

Some examples of monopolies that you have to agree in the present day are necessary, and are accepted as being necessary.
Federal government of the USA ( or any government of any country, state, city, ... )
National Defence - with a paid standing army.
Federal Reserve Bank, Bank of Canada, Bank of England, ...

With just those three examples, one can envision stability of some sort with their presence, or total anarchy with their demise.
Even though they are monopolies in their own right, competition from around the world is evidently present.

 

[Buffu]

My opinion falls along these lines from the wiki article:A monopoly is not necessarily an evil, and very few would not have any competition of some sort. Any business or venture always should always be looking over its back to see what is coming up, and be innovative to not be overwhelmed. Having a bold assumption that they cannot be touched is to the detriment in the long run, for any venture whether a monopoly or not. The AT&T research arm, Bell Laboratories was financed by a so-called monopoly, https://en.wikipedia.org/wiki/Bell_Labs, see the research done before the breakup of the company by Reagan.

Some examples of monopolies that you have to agree in the present day are necessary, and are accepted as being necessary.
Federal government of the USA ( or any government of any country, state, city, ... )
National Defence - with a paid standing army.
Federal Reserve Bank, Bank of Canada, Bank of England, ...

With just those three examples, one can envision stability of some sort with their presence, or total anarchy with their demise.
Even though they are monopolies in their own right, competition from around the world is evidently present.

@DrZoidberg did wrote in his reply that some monopolies can be beneficial but not most and certainly not in CPU manufacturing. Just compare the prices of Ryzen 3/5/7 CPUs to 7th gen core i3/5/7 CPUs.

 

[DrZoidberg]

There is a Sentence in the text you quoted that I find particularly interesting.

Such laissez faire advocates suggest that only a coercive monopoly should be broken up, that is the persistent, exclusive control of a vitally needed resource, good, or service such that the community is at the mercy of the controller, and where there are no suppliers of the same or substitute goods to which the consumer can turn.

While the x86 CPU market is not a real monopoly it seems to me that that sentence partially applies here. CPUs are a vitally needed resource. A huge number of people and companies are dependend on those CPUs. The existence of these patents makes the market entry for potential competitors very difficult to impossible and in this way it could be called coercive.

 

[russ_watters]

Quite the contrary. Patent protection should only apply for really good ideas of course.

Except that you are suggesting that if an idea is too good (too profitable), protection should be revoked. So only those ideas that stay under your chosen level of usefullness will retain their patent protection.

However there are cases where even a patent on a good idea has a clear negative effect e.g. CPUs...

Can you explain why you think this is true for CPU manufacturers? It just seems to me like you are 15 years late for this party: PC CPU technology maxed-out 15 years ago, so bringing in another player using the same technology isn't going to change anything. What is needed to move it forward is an entirely different way of processing.

 

[russ_watters]

@DrZoidberg did wrote in his reply that some monopolies can be beneficial but not most and certainly not in CPU manufacturing. Just compare the prices of Ryzen 3/5/7 CPUs to 7th gen core i3/5/7 CPUs.

No, you do it: please make your point yourself.

 

[Buffu]

No, you do it: please make your point yourself.

I never said monopolies can be beneficial. Can you cite my quote saying so ?

 

[russ_watters]

I never said monopolies can be beneficial. Can you cite my quote saying so ?

Huh? I'm asking you to justify your claim that the current cpu climate is not "beneficial" (setting aside whether it really is a monopoly). "Just compare..." is not sufficient: you need to show us the comparison you are referring to. That would be true even if I did see your point, which i honestly don't.

 

[russ_watters]

@DrZoidberg did wrote in his reply that some monopolies can be beneficial but not most and certainly not in CPU manufacturing. Just compare the prices of Ryzen 3/5/7 CPUs to 7th gen core i3/5/7 CPUs.

I'll be more pointed: it appears to me that you are suggesting that the Intel processors are higher in price for the same performance due to monopoly power. But it seems to me:
1. That isn't true.
2. Even if it were true, it wouldn't suggest monopoly power to me, it would suggest brand loyalty (see: Nike).

 

[jim mcnamara]

It is clear that the OP started with the assumption of a monopoly:
From post #1, last paragraph.

So the question is whether it is possible to break monopolies of Intel and AMD in CPU market ? Or we are struck with AMD and Intel until end of time ?

So let's get past this point, or the thread is going to be closed. It seems to me, Russ is trying to get past that point as well.

 

[wukunlin]

Well, Qualcomm is trying to push their SoC to the laptop market running Microsoft Windows
https://www.laptopmag.com/articles/windows-10-snapdragon-835-laptops

Looks like Windows makes running x86 apps on those hardware possible through emulation, and intel isn't too thrilled about it
http://www.androidauthority.com/intel-rattled-windows-emulation-snapdragon-835-779490/

Assuming this will keep going on, it doesn't seem far fetched that there will be ARM CPU's powerful enough for desktop purposes in the future. I mean they are already building supercomputers with it, aren't they? Though, I have to admit that I know nothing about supercomputing so this probably isn't exactly relevant...
https://www.top500.org/news/cray-to-deliver-arm-powered-supercomputer-to-uk-consortium/

 

[StoneTemplePython]

This thread seems to be quite light on relevant facts.

Fact: AMD is not profitable. Look back over the past 5 years. Look over the last 10 years, 15 years, whatever. Average things in some way that seems appropriate and anyway you look at it, the owners of that enterprise have not had a good time.

The general view in economics is that firms enter a market when they think it is profitable to do so. If AMD has incumbency, seems to try extremely hard, and does not reward its capital providers, why would some new potential entrant expect to fare any different? This is the first question to address long before anyone brings up licensing or politics or anti-trust or whatever -- i.e. why would you expect entry? What I'm hinting at is: the CPU business is a tough one for everyone except Intel. (Why?)
- - - -
Some of the people in this thread may want to look at Bruce Greenwald's Value Investing. It feels a bit dated in the high tech world, but never the less it has a very nice analysis of Intel vs AMD (long story short: mild demand preferences combined with scale economies all but ensures Intel does quite well). Greenwald runs the Value Investing program at Columbia and unlike basically everyone else teaching this stuff actually invests in companies, and further he even tells people what he's investing or not investing at risk of occasional major embarrassment. (Last I read, he has made a considerable amount of money off these investments and donates proceeds to charity.) Every now and then he gets something way off, but he's thoughtful and does the theory and practice quite well. His views have a strong flavor of industrial organization to them, plus a good bit of real world / investing pragmatism.

If you look at the set of people with strong views economics, the set of people who are quite sophisticated with economics, and the set of people who put material amounts of their own money where their mouth is (i.e. betting on their predictions)... the intersection is quite small.

 

[russ_watters]

What I'm hinting at is: the CPU business is a tough one for everyone except Intel. (Why?)

Well focused. My answer is that it is an R&D heavy business and the bigger the company is, the more able they are to put money into R&D to build newer, better products.
[reverse order]

The general view in economics is that firms enter a market when they think it is profitable to do so. If AMD has incumbency, seems to try extremely hard, and does not reward its capital providers, why would some new potential entrant expect to fare any different? This is the first question to address long before anyone brings up licensing or politics or anti-trust or whatever -- i.e. why would you expect entry?

Obviously, you wouldn't.

 

[Vanadium 50]

The stated topic of this thread is overly broad and still not well-focused.

There are four major CPU manufacturers: Intel (US), Global Foundries (US/Dubai), Samsung (Korea) and SMIC (China). AMD doesn't actually make chips - they design them, and send the designs to Global Foundries to actually make them.

Additionally, not all CPUs are x86-compatible. Roughly an order of magnitude more ARM processors are sold than x86s, and roughly an order of magnitude more smaller processors are sold than ARMs. For $5 you can get an 8080 compatible SOC and build your own 1980's vintage personal computer. Intel has ventured into this market, never very successfully.

AMD exists despite its unprofitability because of the need of large customers to have a second source besides Intel. If their market share falls below perhaps 10%, sales will be propped up by these large customers who don't want a true monopoly, or well-bankrolled third source emerges: VIA owns a license already, for example, as does AMD. A company like Samsung could make a move. But right now, given that this market is at best marginally profitable, there is no desire to compete with AMD for the crumbs.