Qualcomm Reportedly Loses Interest In Intel Takeover 51
Qualcomm's interest in acquiring Intel is cooling due to the complexity of the deal, Intel's debt, and regulatory hurdles. However, according to Bloomberg, Qualcomm may still explore acquiring certain divisions of Intel to expand into markets like PCs and networking. Tom's Hardware reports: [T]he proposed acquisition faced significant obstacles, including Intel's $50 billion debt, dropping CPU market share, and its struggling semiconductor manufacturing unit, an area where Qualcomm lacks expertise. A deal of this magnitude would also likely trigger extensive regulatory scrutiny, particularly in China, a key market for both companies.
Intel is undergoing significant restructuring under CEO Pat Gelsinger to reclaim its competitiveness in the semiconductor market in terms of products and process technologies. Still, for now, both Intel and Qualcomm are quite successful standalone companies. While the combination would make a formidable firm (probably facing unprecedented antitrust scrutiny), it does not make much sense for Qualcomm to make such a massive takeover. These factors have collectively made a complete takeover less appealing to Qualcomm. Meanwhile, selling off a part of the company to Qualcomm may not make sense for Intel.
Qualcomm aims to generate $22 billion in annual revenue by 2029 by expanding into markets like personal computers, networking, and automotive chips. Although Cristiano Amon, Qualcomm's chief executive, has stated that his company did not need a major takeover to achieve this goal, the company initiated preliminary discussions with Intel regarding a potential acquisition in September. Yet, it does not look like the deal is going to happen.
Intel is undergoing significant restructuring under CEO Pat Gelsinger to reclaim its competitiveness in the semiconductor market in terms of products and process technologies. Still, for now, both Intel and Qualcomm are quite successful standalone companies. While the combination would make a formidable firm (probably facing unprecedented antitrust scrutiny), it does not make much sense for Qualcomm to make such a massive takeover. These factors have collectively made a complete takeover less appealing to Qualcomm. Meanwhile, selling off a part of the company to Qualcomm may not make sense for Intel.
Qualcomm aims to generate $22 billion in annual revenue by 2029 by expanding into markets like personal computers, networking, and automotive chips. Although Cristiano Amon, Qualcomm's chief executive, has stated that his company did not need a major takeover to achieve this goal, the company initiated preliminary discussions with Intel regarding a potential acquisition in September. Yet, it does not look like the deal is going to happen.
And crappy tech... (Score:4, Interesting)
Qualcomm is just too nice to say that. Intel is a lemon at this time. A lemmon nobody sane wants.
Re:And crappy tech... (Score:4, Informative)
Oh, how the mighty have fallen.
I remember when Intel was a tech behemoth, and AMD was a struggling upstart nipping at Intel's heels. Today, Intel's market cap is half of AMD's.
"Bankruptcies happen two ways. Gradually, and then suddenly." -- Ernest Hemingway
Re: (Score:2)
Re: (Score:2)
AMD went through rough water as well before.
AMD has fallen behind and then caught up with its next generation.
AMD can do that because it's fabless, which makes it fast and nimble.
Intel's problems are structural.
Intel does both design and fabbing, and is only as strong as the weakest link. The design side and the fab side are like two drowning men hugging each other.
They will catch up with TSMC.
Not likely. More importantly, Intel shouldn't be trying to compete with TSMC in the first place.
Re: (Score:2)
Re:And crappy tech... (Score:5, Informative)
From memory, AMD was at a certain point not fabless.
AMD spun off its fabs in 2008.
The spin off became GlobalFoundries [wikipedia.org].
In hindsight, it was a smart move.
Samsung is the only company that successfully combines designing and fabbing, and that's due mainly to Korean cronyism.
Intel is the best the US has.
TSMC is building a big fab in Arizona.
Re: (Score:2)
TSMC is building a big fab in Arizona.
Yeah, and the yields there are already better than Intel's.
Intel won't have another process out until next year at the earliest, and that's assuming they actually achieve it successfully unlike their last process advance.
Re: (Score:2)
Intel is the best the US has.
False [tomshardware.com].
TSMC got yields up in the USA before Intel did, after starting later.
TSMC will become an American company after Taiwan is invaded by the PRCCCP.
Intel says their next process is coming sometime next year. We'll see.
Re: (Score:2)
TSMC will become an American company after Taiwan is invaded by the PRCCCP.
There are a *lot* of pretty wild assumptions there.
When/if the PRC takes Taiwan, the US has options:
Accept that it happened:
TSMC is now a Chinese company.
Do not accept that it happened:
Warfare. It all becomes pretty moot, then.
Economic warfare (we seize TSMC assets in the name of "national security"), which likely leads to real warfare.
The first is what is going to happen.
TSMC will become a Chinese company.
The powers that be aren't willing to start a hot war with China over an island 80 miles
Re: And crappy tech... (Score:3)
"There are a *lot* of pretty wild assumptions there."
TSMC and ASML have both stated that the EUV machines will shut themselves down without frequent blessings and also tweaking from ASML.
China can conquer Taiwan all they want and what they will get is nothing of use to them.
You may try to keep up at your leisure.
Re: (Score:3)
TSMC and ASML have both stated that the EUV machines will shut themselves down without frequent blessings and also tweaking from ASML.
See above. Blessing will be given. The kills switches act as a threat.
China can conquer Taiwan all they want and what they will get is nothing of use to them.
Yes, they will.
1) economically crippling the western world, who buys 92% of its silicon from TSMC. Unless of course the kill switch isn't used.
You may try to keep up at your leisure.
You may continue fantasizing at yours.
Re: And crappy tech... (Score:4, Insightful)
1) economically crippling the western world, who buys 92% of its silicon from TSMC. Unless of course the kill switch isn't used.
Huh? TSMC produces 16 million 12-inch-equivalent wafers annually, or 1.8 billion square inches. Worldwide silicon production is 12.6 billion square inches. TSMC produces about 15% of the world's silicon chips. And no, it isn't even 92% of the world's "advanced silicon" [semiwiki.com], unless you arbitrarily pick 3nm as the dividing line for "advanced", and even then, only if you do so in a way that excludes Intel and Samsung's 3nm processes.
And not all of TSMC's chips come from Taiwan. In addition to Taiwan and China, TSMC also has overseas manufacturing sites in Kumamoto, Japan; Camas, Washington; Phoenix, Arizona (opening next year); and Singapore (SSMC), which makes up somewhere between 10 and 20 percent of TSMC's capacity, and that number will probably increase as more overseas fabs come online in the U.S. in 2025 and 2027.
Yeah, losing the TSMC plants in Taiwan would *really* suck, but the alternative could very well be worse from a geopolitical perspective, depending on who you ask. It's a tough call. So I fully expect them to go through with it and brick the systems if China invades.
And to play devil's advocate, it seems likely that western tech companies will be wary of CPUs being made in China from a security perspective, knowing that China would then have access to the process technology required to produce modified CPUs that look like the originals, but are back-doored. If China invades Taiwan, without extremely expensive levels of oversight by the tech companies themselves, all products coming out of those fabs will be seen as untrusted within a matter of weeks at most, along with products manufactured in China because of the risk of them pulling a part substitution on the assembly side.
In other words, if China invades Taiwan, the tech industry will spare no expense to pull all production out of China in O(months), so continuing to operate the fabs would buy TSMC just a few months of profit. And the tech companies that depend on TSMC protecting their intellectual property will likely never trust them again afterwards. If they don't use the kill switch, stick a fork in it, and you'll see that they're done. So there's really very little reason for TSMC to *not* use the kill switch, and a lot of reasons for them to do so.
Re: (Score:2)
Huh? TSMC produces 16 million 12-inch-equivalent wafers annually, or 1.8 billion square inches. Worldwide silicon production is 12.6 billion square inches. TSMC produces about 15% of the world's silicon chips. And no, it isn't even 92% of the world's "advanced silicon" [semiwiki.com], unless you arbitrarily pick 3nm as the dividing line for "advanced", and even then, only if you do so in a way that excludes Intel and Samsung's 3nm processes.
Sorry- I should have been more accurate.
TSMC produces 92% of the high-end silicon.
China produces most of our low end (100nm+)
And not all of TSMC's chips come from Taiwan. In addition to Taiwan and China, TSMC also has overseas manufacturing sites in Kumamoto, Japan; Camas, Washington; Phoenix, Arizona (opening next year); and Singapore (SSMC), which makes up somewhere between 10 and 20 percent of TSMC's capacity, and that number will probably increase as more overseas fabs come online in the U.S. in 2025 and 2027.
10 and 20%.
Yes, a reduction of 80-90% of capacity instead of 100% totally changes this conversation.
Yeah, losing the TSMC plants in Taiwan would *really* suck, but the alternative could very well be worse from a geopolitical perspective, depending on who you ask. It's a tough call. So I fully expect them to go through with it and brick the systems if China invades.
I don't think it's a tough call at all.
You can't look 360 degrees around you without seeing dozens of things containing chips made there.
And to play devil's advocate, it seems likely that western tech companies will be wary of CPUs being made in China from a security perspective, knowing that China would then have access to the process technology required to produce modified CPUs that look like the originals, but are back-doored. If China invades Taiwan, without extremely expensive levels of oversight by the tech companies themselves, all products coming out of those fabs will be seen as untrusted within a matter of weeks at most, along with products manufactured in China because of the risk of them pulling a part substitution on the assembly side.
Untrusted? Sure. But what's your options? Stop the production of cell phones? Anything with an EUV CPU that isn't an Intel?
In other words, if China invades Taiwan, the tech industry will spare no expense to pull all production out of China in O(months), so continuing to operate the fabs would buy TSMC just a few months of profit. And the tech companies that depend on TSMC protecting their intellectual property will likely never trust them again afterwards. If they don't use the kill switch, stick a fork in it, and you'll see that they're done. So there's really very little reason for TSMC to *not* use the kill switch, and a lot of reasons for them to do so.
Fucking absurd.
The kil
Re: (Score:2)
Huh? TSMC produces 16 million 12-inch-equivalent wafers annually, or 1.8 billion square inches. Worldwide silicon production is 12.6 billion square inches. TSMC produces about 15% of the world's silicon chips. And no, it isn't even 92% of the world's "advanced silicon" [semiwiki.com], unless you arbitrarily pick 3nm as the dividing line for "advanced", and even then, only if you do so in a way that excludes Intel and Samsung's 3nm processes.
Sorry- I should have been more accurate. TSMC produces 92% of the high-end silicon.
How do you define "high-end" to get that high a number? It's more like two thirds in terms of high-density chips, and if by high-end silicon, you mean PC chips, it's likely less than one third, though Intel having their 2024-release chips made by TSMC is likely to shift that needle towards TSMC a lot in the short term.
Yeah, losing the TSMC plants in Taiwan would *really* suck, but the alternative could very well be worse from a geopolitical perspective, depending on who you ask. It's a tough call. So I fully expect them to go through with it and brick the systems if China invades.
I don't think it's a tough call at all. You can't look 360 degrees around you without seeing dozens of things containing chips made there.
Yeah, but other than the five Apple products I see, most of those TSMC chips are made with a larger process size, not the advanced stuff, which means other fabs could produce the chips. You'
Re: (Score:2)
How do you define "high-end" to get that high a number? It's more like two thirds in terms of high-density chips, and if by high-end silicon, you mean PC chips, it's likely less than one third, though Intel having their 2024-release chips made by TSMC is likely to shift that needle towards TSMC a lot in the short term.
Christ, stop it.
Intel makes only Intel parts- and for the relevant market segment- only CPUs. So laptop and desktop CPUs. They do technically make some GPUs, but they're insignificant in the market currently, so I don't think we need to dig down deeper in that technicality.
So for all high density (let's call it 7nm and below- and we'll include Intel 10, just because it has similar density to TSMC N7) there are exactly 3 players.
Samsung, Intel, and TSMC. 2 really, if we're talking non-Intel.
Samsung and
Re: And crappy tech... (Score:2)
"Sorry- I should have been more accurate."
Why don't you just make that your sig? It always applies.
Re: (Score:2)
The context was the effect of losing TMSC Taiwan.
I said, "we lose 94% of our silicon capacity"
Sure, I meant 95% of our current-CPU silicon capacity. Technical difference- large. Effective difference? None.
Re: (Score:2)
Re: (Score:2)
It is questionable about, "fallen behind". If you overclock and just throw excess voltage at a chip for higher clock speeds, and you do it as the standard, then sure, on the surface it's better, but you are using 100W of extra power to get the higher speeds. On the other hand, with Zen4 and Zen5, PBO(which was on by default for Zen4), would automatically clock itself higher based on temperature as the guide. Hitting 5.6GHz without manually overclocking on 16 cores and having that be fully safe and sta
Re: (Score:1)
The problem is that Intel never had good CPU engineering, being a memory-company originally. AMD comes from signal processors...
Re: (Score:2)
Intel has had world-leading CPU engineering and shitty engineering, a lot like AMD.
Re: (Score:2)
Have an actual look at tech history instead of going with your gut-feeling.
Re: (Score:2)
Right now, this very moment, Intel cores are flatly superior to AMD cores in terms of performance per clock. That's good engineering.
The problem they have currently, is a process that is producing faulty silicon.
In a world where Intel were producing its cores with the same TSMC process as AMD cores, teh Intel cores would be the better buy in all cases, except for where you're lookin
Re: (Score:2)
Right now, this very moment, Intel cores are flatly superior to AMD cores in terms of performance per clock. That's good engineering.
The problem they have currently, is a process that is producing faulty silicon.
The problem they currently have is that compilers and dynamic recompilation have finally caught up, and RISC won the war against CISC. Modern ARM CPUs are now emulating Intel CPUs faster than the native performance of Intel's mobile CPUs from just a couple of years earlier, at a fraction of the power consumption. The x86-64 architecture is an architectural dead end at this point, with modern x86-64 CPUs competitive only for specialized workloads that are critically dependent on performance per core, and e
Re: (Score:2)
The problem they currently have is that compilers and dynamic recompilation have finally caught up
What in the fuck are you talking about?
and RISC won the war against CISC.
Only truly ignorant people say this.
The distinction is meaningless.
Modern ARM CPUs are now emulating Intel CPUs faster than the native performance of Intel's mobile CPUs from just a couple of years earlier, at a fraction of the power consumption.
This is a stupid fucking metric.
There is 1 arm CPU, period, that competes with current x86 core performance. And Apple designs it, and only TSMC can make it.
To say that says anything about arm in general is ludicrous.
The x86-64 architecture is an architectural dead end at this point, with modern x86-64 CPUs competitive only for specialized workloads that are critically dependent on performance per core, and even that is only because Intel and AMD are pushing for faster clock speeds while pretty much the entire rest of the industry is pushing for better efficiency.
Your opinion is fucking idiotic.
x86 CPUs are the benchmark. Other things compete with them, not vice versa. And ultimately, only arm cores made by Apple do.
You live in a fucking fantas
Re: (Score:2)
Re: (Score:3)
and RISC won the war against CISC.
Only truly ignorant people say this. The distinction is meaningless.
Not really. The only reason Intel has been able to keep up as well as they do is because they throw a huge amount of silicon at instruction cracking, which results in higher instruction density at a huge power cost. When you reach the point where that power cost exceeds the cost of extra cores, you can't ever be competitive in performance-per-watt except in single-core metrics. And once the single-core performance reaches a threshold where it is good enough, there's rarely a reason to go with the power m
Re: (Score:3)
Not really. The only reason Intel has been able to keep up as well as they do is because they throw a huge amount of silicon at instruction cracking, which results in higher instruction density at a huge power cost. When you reach the point where that power cost exceeds the cost of extra cores, you can't ever be competitive in performance-per-watt except in single-core metrics. And once the single-core performance reaches a threshold where it is good enough, there's rarely a reason to go with the power monster chips other than momentum.
Wrong. You have no fucking idea what you're talking about.
CISC and RISC are meaningless terms.
arm isn't even a very RISC-like architecture. It has included variable length instructions for almost ever now.
People have thrown around the claim that "large instruction decoders" are non-competitive for decades, and they're as wrong now as they were then.
There are benefits that come with large instruction decoders- particularly in a lower amount of instructions that need to be fetched.
On terms of work/watt,
Re: (Score:2)
Today, Intel's market cap is half of AMD's.
That bad already? Impressive!
Re: And crappy tech... (Score:1)
Heh (Score:2)
I'd laugh my ass off when AMD would gather the funds to buy Intel in some years or so.
Re: (Score:2)
I'd laugh my ass off when AMD would gather the funds to buy Intel in some years or so.
That'd be hilarious, but there's no way it'd pass anti-trust muster.
Re: (Score:2)
Then they can buy the dead corpse.
Re: (Score:2)
Re: (Score:2)
I'd laugh my ass off when AMD would gather the funds to buy Intel in some years or so.
That'd be hilarious, but there's no way it'd pass anti-trust muster.
Right now, no, but the trend towards ARM-based computers is accelerating. By 2027, 25% of PCs are expected to be ARM-based, and 50% by 2029. So in five or six years, unless Intel and AMD pivot away from the x86-64 architecture, they'll be minority players in the market, and there would likely be little objection to them merging.
Re: (Score:2)
If AMD liquidated every asset it had right now (went out of business), they *still* couldn't afford to purchase a struggling Intel. It would have to be a stock swap.
Why am I seeing ads on Slashdot? (Score:1)
I have the 'Ads disabled' checkbox checked. Until 2 days ago, this was sufficient to keep the Slashdot home page free of ads.
2 days ago, the homepage stopped loading for me. I had to set NoScript to allow scripts from html-load.com to load, in order to see the homepage.
With this change, ads started to be displayed on every Slashdot page.
Re: (Score:2)
Re: (Score:2)
yeah, this happened on a system with AdBlockPlus installed instead of uBlock. Time to migrate, perhaps.
Re: (Score:2)
I have fsdn.com enabled and I'm not getting ads
I have 6sc.co, and ml314.com blocked and I am not seeing any.
I am using only noscript and ublock origin for ad blocking.
Wow! Qualcomm?! (Score:5, Funny)
They must have made a fortune selling licenses for Eudora.
Re: (Score:2)
They made a fortune selling patent licenses under extractitive terms.
They would have good and truly killed Intel's x86 under similar terms.
I run AMD here but I also don't want AMD to have no competition.
Intel has certain mobile patents that probably QC was trying to get to ensure market monopoly in the Western world.
China don't care if we let government detonate our economy. And USG gave Intel $8B in corporate welfare so Intel may not feel like giving QC a sweet deal now.
Intel's management still ignores eng
Re: (Score:2)
Intel's management still ignores engineering so that will be burned through quickly.
This is an idiotic take.
Intel has the best performing x86 core in the world right now. A crown that they have often had, due to good engineering at times, and willingness to push cock rates through the roof at times (which is also good engineering, to a degree, since if you can push clocks higher than your competition, your part is still better at something)
There are problems with their process that they're using to produce that core. Big problems.
Claiming that this implies "Intel's management ignores
Re: (Score:2)
Re: (Score:3)
That's like saying "Biggest Truck in its Class." We've (hopefully) switched over to 64-bit instructions and not Intel's.
You fucking moron.
Do you think anyone was referring to 32-bit x86?
If Intel's performance is best, why do supercomputers mostly use AMD? I know the answer, but do you?
No, you don't.
The answer there is simply economics.
While AMD has lower performing cores, they do have the capability to cheaply put more cores on a die.
That means if you're in the market for X number of x86 TFLOPS, AMD is your cheapest solution.
Re: (Score:2)
Wrong answer. [top500.org]
Intel continues to provide the processors for the largest share (63.00 percent) of TOP500 systems, down from 67.80 % six months ago. 156 (31.20 %) of the systems in the current list used AMD processors, up from 28.00 % six months ago.
You could have said, "Why are the top 2 in the TOP500 currently using AMD?" and at least would have been accurate.... but who knows what the point of that question would have been.
Re: (Score:3)
Wait until Biden leaves (Score:1)