Semiaccurate: Intel the target of an acquisition
Posted by symmetry81@reddit | hardware | View on Reddit | 207 comments
Posted by symmetry81@reddit | hardware | View on Reddit | 207 comments
DYMAXIONman@reddit
They lose the x86 license if they are bought if I'm not mistaken.
chx_@reddit
you mean the x86-64 license
they can't lose the x86 license because they invented that shit :D
FuturePastNow@reddit
If something drastic happens to Intel the corporation, we'll probably see the creation of a shell company with the name and enough of the IP to keep those agreements intact.
Ravere@reddit
The tags on the article are hoc tam and broadcom.
Hock E Tam is the CEO of broadcom
Although I'm sure broadcom has previously been mentioned as a possible bidder.
Helpdesk_Guy@reddit
It's most definitely $AVGO (Broadcom).
They are insane enough to pull it off, and their current evaluation of +$1T would allow that – Broadcom WILL capitulate upon their current valuation of one Trillion one way or another, rest assured!
Them buying Intel as a whole using a stock-for-stock buy-up, would absolutely allow them to pocket the Intel-disaster, only to divest the most valuable assets afterwards – Would a good fit to be honest, since both are cut-throat companies, and Broadcom would have no scruple to slaughter Intel and sell it off into a million pieces just for the laugh of it.
… an with that, Intel would ultimately meet its final boss in their own end-game.
Vushivushi@reddit
https://twitter.com/dylan522p/status/1880379652054901175?t=OeVPwPiHa6nPt2TlEThTUA&s=19
Or not. Oh boy. What's happening?
imaginary_num6er@reddit
Investing in Intel is going to be just setting fire to a pile of cash. All of its talent have long left the company and hiring a bunch of new engineers will not magically create a new TSMC and Nvidia
Vb_33@reddit
Florida, home..
Correct-Witness8233@reddit
Didn’t the CEO of Broadcom make a statement that he’s not interested in buying Intel?
https://siliconangle.com/2024/12/22/broadcom-ceo-hock-tan-claims-no-interest-buying-intel/#:~:text=The%20boss%20of%20%241%20trillion,into%20an%20artificial%20intelligence%20powerhouse.
Helpdesk_Guy@reddit
Yeah, no. Sounds like a tactic, to keep the price-tag down for the time-being!
That's so original, it looks like the investor's handbook chapter II.
Vushivushi@reddit
DYMAXIONman@reddit
Which would be funny because Intel had more revenue last year than they did.
Vushivushi@reddit
And Broadcom had more revenue than Avago when they got gobbled up.
DYMAXIONman@reddit
Really the only buyers that make sense are Qualcomm, Nvidia, or Apple.
Nvidia would likely only want them if they can keep the x86 license, same with Qualcomm (though they might just be okay with the design team). Apple would want the fabs.
Ravere@reddit
It would be kind of poetic
Invest0rnoob1@reddit
Was my guess since Pat killed his previous company also(VMware) and sold it to Broadcom.
sambull@reddit
When Pat came I told my wife they'd try and piece it up and sell it off.
Helpdesk_Guy@reddit
You hopefully shorted INTC before the story, right? You did, right?!
PeakBrave8235@reddit
Why is everything about stock prices with you guys lol
Helpdesk_Guy@reddit
It isn't, though if you have the knowledge, one could leverage, right? Knowledge is power.
I mean, Intel's downfall has been a slow-motion disaster one could see coming from miles away for literally years…
PeakBrave8235@reddit
Ohhh I thought your original comment was meant to be snide or something because you liked Intel or something lol.
Invest0rnoob1@reddit
Good thing he acted like he’d try to save the company 🤡
PeakBrave8235@reddit
And failed. Miserably.
Adromedae@reddit
So confidently wrong, LOL.
Auautheawesome@reddit
Patty G the secret Broadcom agent
AbhishMuk@reddit
Microsoft agents are out
Now friendship with Broadcom agents
I mean, who needed a Nokia anyway? 🥲
Complainer_Official@reddit
Note: The following is analysis for professional level subscribers only.
bubblesort33@reddit
What happens to my stonks if this is the case?
only_r3ad_the_titl3@reddit
if they sell the whole company - you will get paid out. Usually slightly more than the current stock price.
If they sell part of the company - maybe it drops even further or people are happy about the change and it could increase even.
Due_Calligrapher_800@reddit
No Intel shareholders would vote to accept an offer “only slightly higher” than the current share price.
The average offer is usually at least 30% premium when the company is rising and doing well, but no shareholders would vote to accept that.
daekdroom@reddit
"massively undervalued" lol
The entire tech ecosystem is on a bubble. Intel used to be overvalued until everyone figured out they were going for profit over long term sustainability.
Due_Calligrapher_800@reddit
Intel is trading well under book value. Name any other semi that is trading even <2x book value.
Exist50@reddit
I.e. they've been spending a massive amount of money with no clear ROI.
They absolutely are not as advanced as TSMC. Solidly a node behind. Which is the root of the problem.
Due_Calligrapher_800@reddit
The point is, the comment I replied to, said they are maximising profits over long term sustainability. It sounds like they haven’t really reviewed the Intel situation in over 5 years as it’s literally the opposite. As you said, they are burning through cash to invest in fabs to ensure there is some longer term sustainability instead of becoming irrelevant
Exist50@reddit
I think they were referring to the stagnation from the years prior and attributing Intel's current issues to that. But I agree that, at least in terms of absolute amounts of spending, underinvestment is not to blame for Intel's recent issues. Actually the opposite. Gelsinger spent money Intel didn't have in a reckless bet that foundry mattered the most. Turns out he was dead wrong.
For GAAFET, they basically tie TSMC. More importantly, and as you've clearly heard me say before, customers don't pay for transistor shapes or backside metal. They care about PPAC and schedule certainty. And Intel's significantly behind there.
Given how basically every major company with access to that info has written them off, and Intel themselves still continue to rely on TSMC for flagship products (even post ARL/LNL), I think the story is clear enough. If Intel actually had node parity, much less leadership, they wouldn't be in this mess.
tset_oitar@reddit
Broadcom later stated that 18A evaluations weren't concluded yet, so... Then there was a rumor that there were some issues with the initial pdk v0.9 or 1.0 offered to customers, which Intel supposedly fixed shortly after. Also if 18A and its derivatives were in such a poor state, I doubt Broadcom would be bothered to put out that clarification.
One reason for them to use N2 for client cpus could be related to potential thermal constraints imposed by 18A's backside power delivery. After all the improved Fmax thing was only shown off using Atom at 3ghz speeds and not on a 250W+ DT chip with all core clocks over 5.5Ghz.
Exist50@reddit
They put out some boilerplate statement, but notably did not explicitly contradict the reporting. It seems like they just wanted to clarify that they're not permanently closing any doors, but we've certainly not heard of Broadcom as a customer...
Yes, Intel's PDKs have been a dumpster fire, from everything I've heard. That's bad. I think they publicly lied about initial 0.9 availability, for that matter.
I don't think that should be a significant problem. It sounds like N2 is just straight up the best node available, and the incremental perf is worth disproportionately more selling price. Like, let's say it's 10% better perf vs 18A. That's the better part of a CPU generation's worth.
Due_Calligrapher_800@reddit
Until we have that performance data in the wild I don’t think you can say it’s due to performance. There is massive hesitancy about Intel’s ability to deliver as essentially a start up fab with limited contract manufacturing experience. If they roll out 18A and it goes well for their own products and limited external customers, 14A and onwards will have a massive customer influx.
Exist50@reddit
Intel needs to deliver a node on schedule before customers will even start considering them. That ship already sailed for 18A, so I'd say 14A is optimistic for 3rd party adoption.
only_r3ad_the_titl3@reddit
if they offer 40 $ per share i would more than happily accept lol
Due_Calligrapher_800@reddit
$40 for Intel is selling it short massively. It will be one of the most valuable companies in the world in the 2030s and will overtake TSMC as the number 1 global foundry next decade. Calling it now.
Auautheawesome@reddit
I have a cost basis of $36~ so I will happily take $40+
only_r3ad_the_titl3@reddit
i did 0 analysis and bought 20 shares at around 20 euros, 40 usd/share would be an quiet a W
ExtendedDeadline@reddit
They become GPUs
bubblesort33@reddit
Hopefully not Intel GPUs they'll be paying me out in.
ExtendedDeadline@reddit
If you genuinely held their stock, you should be happy for some b770s!
bazhvn@reddit
nV has the chance to do the funiest thing
a5ehren@reddit
they don't want the product team *or* the fabs
tecedu@reddit
Pretty sure they’d love their fabs and even having their dedicated processors lineup would be great for them. Like they could dominate networking hard now if they also some routers and firewalls, intel’s old nodes are also perfect for them, but everything nvidia does needs the latest nodes
Exist50@reddit
They're losing billions a year. Why would Nvidia want them?
Intel themselves don't even use their old nodes. Chipsets and stuff are at TSMC or Samsung.
tset_oitar@reddit
Intel 16E exists. UMC 12 is a thing or is that not the case anymore? Not exactly an old Intel node but it'd still utilize some of the 14 and 10nm capacity
Exist50@reddit
Intel 16 is their best legacy node available, and it still doesn't seem good enough to bother using for much. Heard there's maybe a WiFi chip or something on it? Just don't think it matters in the big picture.
Well yeah, UMC 12nm exists (I'm presuming "still") because of this exact problem with Intel's legacy nodes. And maybe that will help, but it does seem kind of late. And to get back to the original question, it's not something Nvidia would care about either way. Their networking gear and stuff will likely need better than 12nm.
Dangerman1337@reddit
Nvidia is going into Robotics as Dylan Patel of Semi analysis, Intel's fabs, at least the most promising stuff off it can get Nvidia wants at scale than haggling with TSMC. There's logic to it. Because Apple basically dominates latest nodes at TSMC anyways while Nvidia would love to have them ASAP.
Exist50@reddit
How do you figure Intel's fabs are desirable for robotics?
tecedu@reddit
I specifically said networking, you don’t need the latest and greatest nodes for them. Imagine x86 dpus with some lower powered atoms chips themselves, they could quite literally replace the cpus part out of their biggest clusters
Exist50@reddit
And I'm pointing out that Intel's legacy nodes are all so crappy that Intel themselves don't use them. It's cheaper and easier to use TSMC or Samsung. No sense buying a fab for that, certainly.
If you're proposing basically a big iGPU, how is that different than what Nvidia can do with Grace?
VenditatioDelendaEst@reddit
IIRC Intel's old nodes only have proprietary Intel non-standard PDKs
DYMAXIONman@reddit
Yes and no. They are getting killed at TSMC with pricing and might want 18a for their new cards.
That being said the status of the x86 license would need to be sorted out prior to any purchase. Nvidia buying Intel would scare the shit out of AMD though. Nvidia would make sense as the lack of the x86 license locks them out of many laptop deals, console deals, and server deals.
Exist50@reddit
They'd have Samsung if they just want a cheaper node.
That's pretty rapidly changing. Nvidia might not see much value in x86 going forward.
SherbertExisting3509@reddit
I wouldn't put my hopes in Samsung being a viable competitor to TSMC because their nodes look even more unhealthy than Intel's.
Their 3nm node is only being used in smartwatches and most of their fab customers have jumped ship to TSMC
tssklzolllaiiin@reddit
meanwhile, nvidia has been on a hiring spree poaching intel engineers as they abandon ship
Hendeith@reddit
If anyone would be able to whip Intel products and fabs into shape it surely would be Nvidia. With their manufacturing volume for consumer and data centers, own advanced fabs could maybe make sense.
I doubt they want to buy Intel, but it's a funny what if scenario to consider.
gnollywow@reddit
The article tag has the name broadcom in it.
Wyvz@reddit
FTC in response: lol, that's funny.
Exist50@reddit
Will the FTC exist in a few weeks?
constantlymat@reddit
US regulators aren't the only ones who have a say when it comes to takeovers that impact the global supply-chain of semiconductors.
From-UoM@reddit
Intel is valued at 90 billion.
Jensen's net worth it 120 billion.
Let that sink in for a moment.
6950@reddit
Random semi material manufacturer that no one has heard of worth in millions if they say no TSMC Nvidia can't produce a single GPU let that sink in
Accomplished-Snow568@reddit
This is just another rumour as many many times earlier.
Dyslexic_Engineer88@reddit
Intel is trading near its book value. It has many extremely valuable assets and is 100% an acquisition target.
The only issue is that it's a strategic asset to the USA, and any buyer has to have the funds to turn Intel around, not be seen as monopolistic, and not be seen as a threat to national security. That rules out a lot of potential buyers.
With the new administration coming in, the government may be more lax about mergers and acquisitions, which increases the possibility of snatching up Intel.
Exist50@reddit
There's also the question of whether funds to "turn Intel around" wouldn't be better spent elsewhere. Why buy Intel with all its baggage when you can just hire its employees? The x86 mote seems smaller by the day.
Area51_Spurs@reddit
Their fabs
Exist50@reddit
The fabs are what's sinking Intel to begin with, and are probably the biggest barrier to an acquisition. They are viewed as a liability, not an asset.
Dyslexic_Engineer88@reddit
Intel fabs are still near the cutting edge and worth a ton!
Also, x86 isn't going anywhere; having two leading architectures competing on design is good for consumers and the tech industry as a whole.
X86 isn't as bloated as you might think, it can stay competitive with arm indefinitely
Exist50@reddit
The market clearly thinks otherwise. It's a business losing billions of dollars a year with a single major customer, and it hasn't released a node shrink on schedule in a decade. I think Intel Foundry is widely viewed as a lost cause at this point.
The encroachment of ARM in both client and datacenter comes at the direct expense of x86 market share. x86 may not truly die for the foreseeable future, but it's weakening. And that will only accelerate with Intel's downfall.
Maybe if properly invested in, but Intel is not doing that. Nor is that justification to acquire Intel. If you can accomplish the same with ARM, why not just do that instead?
Dyslexic_Engineer88@reddit
No intel is one full step behind TSMC right now. they can still catch up; their 18A could be it, but it's too early to say.
Many times in the past, we have seen different fab companies fall behind and catch up, again.
We have also seen many fab companies fall behind and not catch up, but it's too soon to count Intel out yet.
Companies are moving to ARM because some ARM chips offer competitive products geared toward their specific application. Many applications are becoming agnostic to the CPU instruction set and depend more on the actual underlying design and architecture.
A lot of these new ARM-based CPU companies can undercut intel and AMD in price to performance in specific applications, because they dont need to achieve the same profit margin to support their business, and they are trying to gain market share at all costs.
Other large companies like Apple are switching to ARM because they have the R&D budget to make their own chips, and X86 instruction set is not licensable.
The growth in ARM market share so far has not meant a decline in volume for x86; both will live side by side in different niches.
The CPU instruction set isn't as important as it was 25 years ago, and ARM is now nearly as bloated as X86. What matters more are the architectural designs. Most improvements come from better implementation of out-of-order execution and allocation of resources to specific instructions CPU functions.
Getting these performance gains requires massive investment in architecture and some serious engineering talent.
Intel's new Arrow Lake is very promising like Zen was when it launched, but it will take a few iterations to see if the design pays off.
It's WAY too early to count intel out yet.
Their financial trouble comes mainly from years of under-investing in R&D to maximize profits.
Their undervalued stock results from losses in market share and restructuring costs. The AI hype has also caused investors to move allocations from Intel to Nvidia, which exacerbates the issues for Intel stock.
A return to form for them may require an outside acquisition to shake things up.
Even if they don't return to the cutting edge for a while, they will still crank out CPUs and, hopefully, GPUs, just like AMD did for years when they underperformed.
Exist50@reddit
18A is an N3 competitor years later, when TSMC is rolling out N2. Still a node behind.
Actually, we haven't. This would really be unprecedented. But I don't think that needs to factor into the argument either way.
Yes, exactly. And in such a scenario, x86 is nothing but a liability.
It has though. In many cases, it's a direct substitution.
ARM isn't remotely as bad an ISA as x86, in bloat or anything else. It's probably the best ISA available today, by a decent margin.
What? Arrow Lake has been a disaster. 3 year jump in core architecture, 2 full node shrinks, expensive advanced packaging, and it's commonly beaten by Raptor Lake...
They've been slashing product RnD over the last 1-2 years. The only thing they've been consistently investing in is foundry, with nothing to show for it yet.
FloundersEdition@reddit
ARM has many issues. they dropped support for Arm32 with X2 and A720, resulting in many sticking to old cores like the A78 (Switch 2) that lack modern ISA-extensions like low precision data types, SVE and SME.
Qualcomms Oryon doesn't support SVE/SME either and relies on Neon. in effect, Arms vector ISA-extension is still 128-bit NEON/Arm v8.2A and for the next 8 years this will likely not change for any Arm code (unless you give a shit about code portability). FIY they specificed 8.3-8.9 and 9.0-9.6 already. it's even worse than Intels AVX-512 mess.
you typical don't run assembly on Apples architecture either and I don't know, what there secret vector extensions are. I really don't know, if you can really call it an Arm ISA anymore, you basically always use APIs.
Apple got burned 3 times on CPU, PowerPC, x86, Arm32 and with GPU as well (Nvidia, Imagination, AMD, moving to an in house solution). they really had them all and don't want to have any more friction for devs. they basically don't allow you to do anythings outside of their APIs.
you also don't have a real baseline. if you decide Zen 1, you basically always now, newer cores are better. but from A78 to A710 as well as from A55 to A510 to A520 they decided to cut back on the width of the core or cache width and often run slower. many cores have multiple potential L1I, L1D and L2 cache setups. at one point they removed their micro-OP cache.
so you never know, how code will run, even if a core is more modern. your best baseline is an old Arm core, but don't believe you will have ISA-compability for much longer!
MC_chrome@reddit
God forbid someone tries to push the tech industry past 2003....
The rabid insistence that 32 bit must stick around till the end of time is starting to get a bit old now, 20 years after 64 bit was originally introduced by AMD.
FloundersEdition@reddit
GameBoys/Nintendo DS used Arm32. Saving the last transistor isn't always worth it. GameFreaks and co refurbished many old IPs in the past like Pokemon Firered. Now it might be a complete rewrite.
not sure why you think it's such a big overhead. Arm32+Arm64 cores were the standard for a decade and still tiny and efficient. Be it A55, A78 or X1, they all supported both.
AMD64 is also incompatible with Arm64.
Exist50@reddit
ARM dropping legacy support is a good thing for the ecosystem moving forward, and Nintendo's almost certainly using A78 because it's cheap, not for ISA of all things.
Tremont is worse than Zen 1. And doesn't have AVX, for that matter. What are you even trying to argue here? That a wider variety of cores is a bad thing?
They do not "often run slower". Pointing at structure sizes instead of performance is absurd.
FloundersEdition@reddit
cost is certainly a consideration for Nintendo, but they had compability friction in the past (GameCube, WII, WII U are PowerPC, DS and 3DS are Arm32) and I don't think they want to repeat that. especially after all the backlash PS3 and PS5 got for it's lackluster backwards compability. if they screw it up, people might change to SteamDeck.
Tremont is Intel. it's bad if a wide variety of new cores comes out (\~5 Arm-cores a year with multiple L1 and L2 cache configurations) as well as custom Arm cores and you can't define a baseline even for a single vendor or line-up. A710 shouldn't be worse than A78, dual A510 shouldn't be worse than dual A55.
it's one thing to define a baseline for 3 x86 chips, it's something different if you can't even do it on an spreadsheet due to a million configurations. big.Little makes it even harder. debugging that is close to impossible, no software company can do that. having a baseline is important for software devs.
Exist50@reddit
Is there any evidence that the Switch is running 32b software?
Yes, I'm pointing out that you can't draw an arbitrary cutoff and assume everything after Zen must be better. That's not something developers want to begin with, speaking at an ISA level. It's literally worse because it restricts the potential markets.
ExeusV@reddit
By which metric?
Exist50@reddit
All else equal, x86 is a harder ISA to develop for.
Dyslexic_Engineer88@reddit
No one really cares about ISA any more except compilers Devs and maybe a few other niche developers.
Exist50@reddit
The hardware developers certainly do. Variable length instructions are a real pain the ass. To say nothing of all the x86 legacy crap.
Strazdas1@reddit
Your another comment in this very thread:
Area51_Spurs@reddit
The fabs aren’t the problem. They’re losing money because intel isn’t selling as many chips and intel isn’t taking advantage of them.
You have this all backwards.
Exist50@reddit
That's not the only reason. Their nodes are not remotely cost competitive with TSMC's equivalents. Intel very explicitly claimed that was their main hurdle to profitability.
And if they don't have demand because Intel's not selling chips, how isn't that also a major problem. Intel Foundry has no other major customers, and has seemingly failed to change that.
JDragon@reddit
IMO an acquirer is probably thinking less about competitiveness, and more about bilking an “America First” government for billions of dollars while that same government strong-arms fabless American companies into using Intel fabs.
Exist50@reddit
That's an extremely unlikely scenario. Forcing the use of Intel fabs would kill the US tech industry. Just not going to happen. If investors thought that plausible, Intel wouldn't be valued so low.
JDragon@reddit
Fully agreed it's unlikely, but not out of the realm of possibility in my opinion. Intel has already planted the idea that its fabs are fundamental to national security. Even in its diminished state it gets preferential treatment in Washington and has shown that it wants to maintain that. (Ex: during CHIPS Act negotiations Intel fought to kill benefits for semiconductor design because it would have benefited fabless companies and not just Intel - even though Intel Product would have benefited also.) At this point, besides the rotting company culture, the reasons why Intel can't return to fab competitiveness are its lack of profitability to fund future node R&D and customers not trusting Intel due to a variety of fair reasons (terrible track record, lack of external foundry experience, competitor of Intel Products). A firehose of government money combined with forced usage solves both problems.
Destroying the tech industry and wiping out his competitors might even be a bonus to a potential oligarch owner. Fully veering into conspiracy theory territory: how does a tech oligarch win the AI game of thrones?
The whole thing probably falls apart because this is Intel after all, but with all these wannabe oligarchs cozying up to an administration known for bombastic and conflict-seeking policies, can't help but wonder...
tset_oitar@reddit
Imo it's still a bit early to judge the foundry. If by next year theres still no sign of a large customer then yes, they are in trouble
nanonan@reddit
The lack of customers for 7, 4, 3 and the cancellation of 20A doesn't exactly bode well. 5 nodes in four years resulting in zero external customers and abandonment by your own design arm isn't very promising.
formervoater2@reddit
oh it totally is, it just doesn't matter.
x86 vs. RISC mattered in 1990 when x86 wasn't superscalar and compilers/interpreters weren't optimized for it because it was easier to optimize a compiler/interpreter for RISC and make a superscalar RISC design.
These days heavily optimized x86 compilers and superscalar x86 cpus are commonplace. Developers and hardware engineers have essentially brute forced through the shortcomings of x86.
therewillbelateness@reddit
X86 isn’t even slightly competitive in anything that requires low energy consumption.
Dyslexic_Engineer88@reddit
That is not true at all.
The new Intel Core Ultra Two chips are competitive with the Qualcomm X Elite chips in power and performance.
AMD Zen chips can achieve better performance per watt than Apple chips in a lot of more demanding tasks.
Performance targets and power targets are different for various applications for which CPUs are designed. So, it is hard to conclude without looking deeper.
ARM started out targeting low power but grew into higher-performance applications.
AMD and Intel have been chasing peak performance for a while, sometimes at the expense of per watt.
A lot of ARM growth is because it can be licensed and customized into any application. It helps to be an ISA design for low power from the get-go, but ISA is not the limiting factor in modern CPUs.
The ARM's success is mainly due to its licensing model, which allows for designs tailored to certain applications.
I won't deny that the ARM ISA has benefits for programmers. Still, again, ISA doesn't improve performance or power consumption much anymore when it is all interpreted into microcode before it is executed.
Exist50@reddit
It's one or the other with LNL. Qualcomm has the better cores.
FloundersEdition@reddit
but not even Intel has mask sets ready to produce something good on them. designing new chips/chiplets will take 4-5 or 3-4 years without the Fabs being utilized.
14nm (where they expanded the most to capitalize on the server demand after Spectre and Meltdown!) and Intel 7 fabs are basically dead now. Intel 7 is frying Raptor Lake and still should be quite expensive (at least if chips are pushed for performance), their margins weren't to hot and even Altera had troubles breaking even after launching most of their lineup on this node. Intel 4 only has MTL. Intel 3 has no client product whatsoever, only a server chip. Intel 20A has zero products and 18A will only ramp in H2 - noone knows cost/wafer, PPA and yield.
I agree, x86 will not be replaced by Arm. Arm raising prices and bitching around with Qualcomm killed any incentive to break out of x86 and enter a new ISA prison. RISC-V + some extensions the industry agrees on? maybe, but 2030+x.
Vushivushi@reddit
They're also what keeps Intel afloat.
They're an IDM, the company falls apart when you remove the M.
Exist50@reddit
No, it's not. Look at their financials. You have a reasonably healthy design business coupled to a grossly unprofitable foundry.
Worked out the exact opposite for AMD.
Automatic_Beyond2194@reddit
Not just the fabs. Intel has decades of R&D. Stuff like glass substrates, both from the foundry and design side. And just knowing what stuff Doesn’t work both from foundry and design side.
They may not be able to properly use it all right now, but it is certainly valuable R&D they have accumulated over the years.
juicenx@reddit
And IP
gnollywow@reddit
Article put broadcom as one of the tags.
Dyslexic_Engineer88@reddit
Intel has more cash in the bank, and less debt than Broadcom does right now.
Broadcom would need to sell a lot of stock or get some outside help to make that deal happen.
Exist50@reddit
Broadcom's worth a lot more than Intel now. If they really wanted to buy it, they could. Hock Tan's done much more ambitious acquisitions before.
abbzug@reddit
Lina Khan is out in a few days. The regulatory hurdles after that will be "How many money trucks do we need to send to Mar-a-lago?"
Top_Independence5434@reddit
Anything to pump up $INTC for the remaining bagholders.
signed7@reddit
It's up 9.5% as per Bloomberg lol https://www.bloomberg.com/news/articles/2025-01-17/intel-shares-jump-on-report-it-s-an-acquisition-target
monocasa@reddit
Doesn't Intel lose it's x86 license if it gets acquired?
RealThanny@reddit
Yes, anyone who acquired Intel would automatically lose the legal right to manufacture x86 processors. So would AMD, which means the acquisition would never be allowed by regulators without an agreement being made with AMD first.
It would have to be nothing short of the current agreement, which is that Intel and AMD each have rights to the x86-related IP of the other, without restrictions, and without money changing hands. Anything short of that would be refused by AMD, making the acquisition impossible.
COMPUTER1313@reddit
Time for Intel to completely pivot to ARM.
Now if Broadcom bought Intel, I could see them immediately suing AMD over the x86 licenses and burn the entire x86 ecosystem to the ground in the aftermath (e.g. Intel barred from manufacturing AMD's x64 and thus need to make their own incompatible x64 or bring Itanium back alive).
Vb_33@reddit
What profits would they make off of that? Seems like leaving a lot money on the table.
COMPUTER1313@reddit
Take a look at how Broadcom is handling VMWare.
Last time I checked, AT&T is suing Broadcom after the billing charge went up by something like 300%-1000%.
RealThanny@reddit
They'd never be permitted to do that. The x86 ecosystem is far too important to the economies of many nations, and certainly has a ton of national security implications as well.
Helpdesk_Guy@reddit
That's sounds exactly to be to Broadcom's liking. Itanium was 100% proprietary!
iBoMbY@reddit
No, the lose access to AMD patents, and IPs, and vice versa:
https://www.sec.gov/Archives/edgar/data/2488/000119312509236705/dex102.htm
monocasa@reddit
Which is half of (and required for) what is collectively called the x86 license.
iBoMbY@reddit
Only Intel owns the x86 part of the deal. What Intel calls x86-64, is a different story, because it is originally called AMD64.
jaaval@reddit
Both of those have patents expired. Intel is the primary holder of the remaining patents, related to AVX and later extensions.
Veastli@reddit
Both AMD and Intel continually add new IP.
To build a modern X86 processor, all that new IP is necessary, and it's protected by patents or copyright. Each use the other's IP, and each have a non-transferable license to the other's IP.
If either firm is sold, they lose the rights to build modern X86 processors.
jaaval@reddit
We are talking about the ISA not about the CPU itself. They don’t share most of their IP with each other.
Veastli@reddit
Yes. And the expired patents from 2004 are not enough to build a modern X86 processor.
If a firm like Qualcomm felt they could build a competitive X86 processor using the expired patents, they would.
Never said they did. Said they share. And they do.
jaaval@reddit
As I think I said, intel is the primary patent holder for the remaining patents, which is basically AVX and later.
There are some security and virtualization features that are specific to intel or AMD and AMD has added some small things over the years but a lot has fell out of use. I’m pretty sure you can build a modern x86 processor without recent contributions from AMD.
Veastli@reddit
Then why is Qualcomm spending massive resources to break into the PC business with ARM?
If your suggestion were accurate, they could have done the same with X86, without having to pay license fees to ARM, or potentially, any group.
Likely because the modern additions to X86 are still protected, and without those, Qualcomm could not make a competitive processor.
ProperCollar-@reddit
Cause they think ARM is the future and have a ton of experience with it??
It's a huge mistake to assume that Qualcomm not making x86 is because of the patents.
jaaval@reddit
Umm… there is still intel? What are you even talking about?
And that aside Qualcomm naturally wanted to make arm cpu which they had experience with and already a huge market for.
Veastli@reddit
They is absolutely room for more competition in X86 processors.
If the x86 IP is as unencumbered as you suggest, certainly some firm, somewhere would be building them.
But they're not doing that. With the exception of the legacy provider VIA, no firm anywhere on the planet is building X86 processors based on the expired IP. And VIA processors aren't in any real demand.
Logically, the ever moving IP added by AMD and Intel is the reason.
iBoMbY@reddit
This is not about patents, this is about copyright/intellectual property, which is also covered by the agreement.
jaaval@reddit
Copyright over what?
ProperCollar-@reddit
Ok. So they do in fact lose the license to what we colloquially refer to as x86.
ProperCollar-@reddit
That guy is being frustratingly pedantic and entirely missing the point. Ignore them
Ok_Suggestion_431@reddit
...they just lose access to the things needed to make an x86 CPU...lol
richburattino@reddit
No, they are owners of this license
monocasa@reddit
It is jointly owned with AMD (who invented the 64 bit extensions), and their cross licensing agreement has either company losing access to the joint body of IP on acquisition.
Skensis@reddit
That can always be renegotiated upon a deal, likely part of any acquisition would stipulate that it would only go through if AMD/Intel can renegotiate that license.
imaginary_num6er@reddit
Stacey Raskin on MSNBC who had for the longest time said Intel has hit bottom each quarter stated that he can’t recommend shorting the stock because it is too volatile. Must be real dire times when a speculator like him essentially gave up on Intel
6950@reddit
Bruh everyone forgets Regulators it's not some small company it's the only US Leading Edge Manufacturer and Designer
Exist50@reddit
I think the assumption, for better or worse, is that regulation will either be cut entirely or be sufficiently easy to bribe.
6950@reddit
China would say No lol
Independent-Fragrant@reddit
Wait why does china have a say? Because intel has operations there? And why would china oppose?
k0ug0usei@reddit
For any company of intel's size, you will need to get approval from every major country solely due to market presence in said country.
6950@reddit
Yeah and China and US are in a Semi Trade war
Exist50@reddit
That's probably true.
6950@reddit
And that's 20-25% of Intel revenue so yup and especially in such a scenario of tension
6950@reddit
News: I will be acquiring Intel Source: Random website article made by me that is paywalled
CammKelly@reddit
The x86-64 license isn't transferable so gonna love seeing how this works, lol.
Frexxia@reddit
Even if the rumor happened to be true (which I doubt), I don't see any possibility that an acquisition of Intel would pass antitrust hurdles.
SherbertExisting3509@reddit
under the next administration all of of those antitrust hurtles could be bypassed by a sufficiently large donation to Mar A Lago
martylardy@reddit
Buy out target is $69 dollars per share
Helpdesk_Guy@reddit
Thanks for the laugh! For who?
AbsoluteAmerica@reddit
Qualcomm? If not them, any ideas on who else is positioned for taking over Intel?
bushwickhero@reddit
Maybe some hedge fund that will load it up with debt and run it into the ground?
Helpdesk_Guy@reddit
The last part was basically almost finished by just the last clown at their helm. So only load it with debts and—Wait, that too Intel already managed by themselves with +$45Bn in debts!
Gosh… Does anyone ever thinks of these needy hedge-fonds to get rid of their debts or make a dime every once in a while?!
DYMAXIONman@reddit
Nvidia and Apple are the other two that make sense.
Auautheawesome@reddit
Broadcom is listed as a tag, doesn't mean it's that, but odd that it's listed on a tag for an article that didn't mention them, and is the only other company listed as a tag other than intel
Helpdesk_Guy@reddit
It's most definitely Broadcom, thus Avago. And their current evaluation of +$1T would allow for that.
They are insane enough to pull it off, and Broadcom will most defintely capitulate upon their current valuation of over 1 Trillion (!) one way or another, rest assured. They won't let that chance of a massive take-over using a stock-swap go wasted!
Them buying Intel as a whole using a stock-for-stock buy-up, would absolutely allow them to pocket the Intel-disaster, only to divest the most valuable assets afterwards.
SlamedCards@reddit
Amazon?
ValVenjk@reddit
So how accurate is the claim?
Exist50@reddit
Semiaccurate has a long history of blatantly making shit up. Sometimes maliciously so. So I'd treat it as less reputable that a random reddit comment.
Automatic_Beyond2194@reddit
I mean Intel stock is up like 8% so people with actual money seem to believe it.
Exist50@reddit
Plenty of people with more money than sense.
Geddagod@reddit
You mean Charlie "Intel kills off the 10nm process" Demerjian might be wrong?
anival024@reddit
Charlie's entire streak about Intel's 10nm process was correct.
The 10nm process Intel ultimately delivered was not what they had been promising, and lying to investors about, for years. Charlie cataloged it quite thoroughly, just like he cataloged Intel's claims about 3D XPoint / Optane.
Charlie even posted a tongue-in-cheek "admission" that he was "wrong" when Intel eventually trotted out their first pathetic 10nm parts that everyone laughed at. Remember Canon Lake? It only existed so Intel could claim they did in fact release "10nm" parts and not get sued by investors. It was a joke.
If you don't want to take Charlie's word for it, you can look at Intel's own roadmaps and briefings. If you don't want to do that, check out this Chips and Cheese article, which touches on some of it.
Exist50@reddit
Charlie claimed they outright cancelled the node. Instead, it was "merely" late and without the densest original library. He was simply wrong.
And of course, that's hardly the only such instance. It's hilarious to see people defending him after once again caught bullshitting with the Qualcomm case.
Geddagod@reddit
Idk about the entire streak, but his final conclusion of Intel 10nm being entirely cancelled was definitely wrong. Hell, Charlie even admits it in a later article- one which he entirely dedicates to why he was wrong about 10nm being canned.
The funnier part was that the day Charlie wrote the article on 10nm being canned, Intel denied it, which Charlie edited in, but also added he still stood by his claim of 10nm being canned.
It was, at least eventually. The biggest point of contention here would prob be about density, since that was arguably the most ambitious part about 10nm.
The 2nd gen Intel 10nm products, (ICL) had the same density as gen 1 (CNL) according to techinsights.
Charlie speculates that their 10nm in CNL would be different than what Intel claimed in their earlier press:
However the density techinsight found was in line with Intel's claims
The perf/watt gains for Intel 10nm didn't really start showing up until arguably TGL, however yea they didn't completely cancel 10nm like Charlie claims. TGL did have obvious perf/watt advantages over SKL even at ULP.
constantlymat@reddit
I mean, Intel is an acquisition target and I would be surprised if companies weren't doing their due diligence on the potential of a takeover.
Frankly, rivaling CEOs aren't doing their jobs if they don't take a very close look at this opportunity. Companies like Intel becoming even a potential acquisition target is a once in a decade opportunity.
A journalist's job in this scenario is to determine if there's even the slightest chance a possible acquisition attempt is maturing beyond the "due diligence" phase.
scbundy@reddit
He had Nvidia declaring bankruptcy years ago.
audaciousmonk@reddit
Semi
GunsouBono@reddit
Semi... duh
NewKitchenFixtures@reddit
If it is then it is a coincide.
positivcheg@reddit
Since price jumped only 8% up means good to move it but not really certain to make it jump 15-20%.
Aelsworn@reddit
Semi.
battler624@reddit
👈(ﾟヮﾟ👈)
symmetry81@reddit (OP)
No unpaywalled disclosure of who but the fact of acquisition talks by themselves are very interesting, if they are accurate.
richburattino@reddit
They are semi-accurate
ExtendedDeadline@reddit
It's in the website name, even!
abruzzi92@reddit
im a stocks novice but why did share price jump on acquisition and if it did theoretically get acquired what would price become?
Kurbalaganta@reddit
Richtig. Es ist entscheidend, dass sie nicht überlebt. Mit anderem Personal wäre evtl. eine andere Einschätzung möglich, aber wer an Lindner als Boss festhält, kann in der Politik nicht ernst genommen werden.
TheAgentOfTheNine@reddit
Charlie demerjian has seen a mail? that's the source? jfk, those that subscribe to that moron's bulletin really deserve to part with their money.
YakPuzzleheaded1957@reddit
Why do people keep pushing the "intel getting acquired" narrative? Intel hasn't shown any interest in selling the whole company, especially now at their lowest price and right before 18A is launched. Excluding foundry their products business still does more revenue than QCOM, and just behind Broadcom. Not like they're weeks away from bankruptcy or something.
Rocketman7@reddit
The market seems to be reacting very positively (insiders making moves), so there might be some truth to this rumor
constantlymat@reddit
The biggest reason to be skeptical of this report:
Considering he is relying on "unnamed sources" an "insiders", what benefit does it have for him to withhold the name of the company? He must know the company's name if he has the first party confirmation by his source.
x3nics@reddit
Charlie Demerjian is a name I haven't heard in a while.
scbundy@reddit
Last i heard from him, he was about 100% certain that Nvidia's demise was anyway now.
f3n2x@reddit
That narrows it down to a window of... about 20 years.
Exist50@reddit
Then you missed his recent meltdowns about Qualcomm, including claiming to have insider knowledge that ARM was a shoo-in to win their lawsuit.
that_70_show_fan@reddit
I am glad he keeps most of his stuff behind a paywall.
blueredscreen@reddit
Never liked and still do not like this guy. Valuable info but atrocious writing style.
Exist50@reddit
Not valuable info either. Everyone knows he flagrantly lies. Remember his Qualcomm stuff?
blueredscreen@reddit
The one about the PMIC? He does get things more often right than wrong, but then he acts like he's the Einstein of his generation...
Exist50@reddit
There was that, which seems to have been sensationalized to the point of falsehood, but it was far from the only instance. He also claimed Qualcomm was outright cheating on their benchmarks, and that he had insider info that ARM's lawsuit was a shoo-in. Neither of those claims aged well.
And of course, it's not just Qualcomm. He famously claimed that Intel cancelled 10nm.
gihty123@reddit
Anybody has a subscription that can share the summary behind the paywall ?
FalseAgent@reddit
isn't this the site that was basically crashing out over the snapdragon X elite launch lol....
Exist50@reddit
He also claimed to have insider info that ARM has an ironclad case, lol.
Inevitable_Bee_9830@reddit
on tradingview the "fact of the day" is that Intel wants to enhance their quanutm chips
Exist50@reddit
They should really not be investing in that.
tset_oitar@reddit
Not happening, even if the plans are legit. Especially not the whole conpany
SlamedCards@reddit
Just a random guess it's Amazon. We'll see if anything comes to fruition.
FloundersEdition@reddit
They have zero synergy tho. Besides Graviton (which is basically only Arm-IP), they have zero experience in hardware designs. No experience in developing software stacks like GPU drivers and math libraries. No experience in providing others support to build their own chips - they even rely on TSMC helping them. No client products to fill the fabs (okay, okay. FirePhone, FireStick and that ebook). No control over OEM products and a pretty low reputation for DIY sales and electronic sales in general. I don't think any major company buys there PCs over Amazon.
SlamedCards@reddit
Well I was thinking about it more. If Charlie isn't lying. He's said it's not any of the ones talked about in the press.
So Qualcomm and Broadcom are out
Nvidia is out for anti competitive reasons easily, AMD was teased in media so they are out
Leaves us with big tech companies.
Apple- They normally don't buy companies and a strong TSMC relationship. Good synergies but anti trust might be an issue.
Google- TSMC relationship is weak. But anti trust issues with being owner of chrome and in PC space.
Meta- Doesn't make sense at all
Microsoft- Lol, never be approved they are giant in PC space
Amazon- Weak TSMC relationship, already have a deal with Intel. Aren't in PC space. Could get approved but business case is harder unless 'AI' etc they believe it's helpful
Best synergies are Apple, but they are tight with TSMC, an antitrust might be an issue. Google less synergies, but antitrust is an issue. Or Amazon, little synergies but almost no anti trust issues
FloundersEdition@reddit
I think Tesla would have the best shot, but I think it's unlikely there will be an aquisition at all.
close relationship to the new admin. mobileye synergy. manufacturing background. SpaceX, Starlink, infotainment, Self driving chips (which they are willing to license out). Some server synergy (twitter and supercomputers for self driving features). Low additional anti consumer/client potential for PC, server, GPU, FPGA, networking, manufacturing. + Crazy guy who would buy Intel. I don't think admin should allow Musk to buy anything tho.
Wyvz@reddit
This is very cool and all, a lot of players are interested and are able to aquire them, partially or fully, this is no secret and not really new either.
It all comes down to whether the regulators will allow it, which I personally doubt they will.
imaginary_num6er@reddit
Note: The following is analysis for professional level subscribers only.
SlamedCards@reddit
Broadcom would be blocked. IMO if Intel is being bought. It's one of the big tech companies
AutoModerator@reddit
Hey symmetry81, unfortunately your submission has been reported because the link address indicates that the content may be locked behind a pay wall. Please consider resubmitting from a different source that everyone can view. If the content is not behind a paywall, please message the moderators and they will be by shortly to take a look. Thanks!
I am a bot, and this action was performed automatically. Please contact the moderators of this subreddit if you have any questions or concerns.