Sources Say Intel Is An Acquisition Target

[hist78]

I see zero possibility of that happening. Was Jensen's jet at Mar-a-Lago?

GTC 2025 should be very entertaining this year. Nvidia really is crushing it.

GTC AI Conference 2025

Experience In Person and Online.

www.nvidia.com

I think Nvidia, Qualcomm, Broadcom, ARM, Tesla, and others may be eyeing Intel's product design division only. The Intel Foundry is a big headache for Intel today and foreseeable future. I don't see those potential buyers being too fond of owning this headache at all.

 

[blueone]

The other problem I have is with GlobalFoundries being involved is that they are not a US owned company. Would that really fly?

GF is a laughably complicated company. Headquartered in Malta, New York, but incorporated in the Cayman Islands, listed on the US NASDAQ exchange, and 81% owned by the UAE Sovereign Wealth Fund. I'm thinking that set-up will make Trump's head explode.

 

[yanfeng]

It may be an option selling Intel foundry to China. It keeps Intel foundry still be US and China get what they needs

 

[DaRAGingLunatic]

It may be an option selling Intel foundry to China. It keeps Intel foundry still be US and China get what they needs

China getting what they need is what the sanctions are all about.

China is preparing for a Taiwan invasion.
China is preparing for war.

So is the west.

I'm quite certain that Australia is preparing for war.
The UK preparing for war.
Canada preparing for war.
New Zealand preparing for war.

Countries are preparing for war.
It's clearly about China and their Taiwanese intentions.
God help us all.

 

[count]

I could see a transaction where Qualcomm buys the design arm and GF buys the foundry business. That could also work.

 

[XYang2023]

Another Suitor Talking To Intel

Since SemiAccurate broke the story about Intel being an acquisition target last Friday, there has been a lot of speculation and rumor. While researching the topic we came across, and verified, anot…

www.semiaccurate.com

 

[sit]

In short Products will succeed and Foundry will fail.
Intel's product and foundry are interlinked for foreseeable future imo the thing is improve Foundry efficiency by filling it with products as much as possible and than selling the rest of capacity for external customer on guaranteed basis meanwhile Products needs to get their act together

Foundry may fail, but products is facing extinction. The world has moved beyond x86 so product is playing in a shrinking market. Absent inventing and then producing what comes after ARM, products is nearing the end. If Intel is to survive, it will be as a Foundry not a Products company.

 

[dkr1986]

I could see a transaction where Qualcomm buys the design arm and GF buys the foundry business. That could also work.

Intel's Products team is worth more than Qualcomm entirely if you look at their financials. Even if I accounted for about 50% unallocated corporate expenses, Intel Products team Market cap would be nearly same as QCOM. If BoD okayed this acquisition (not really in best interest of Intel shareholders - so it may actually happen with Intel's current BoD!), I am pretty sure they will be sued for this.

GFS has too much stake from Saudi money and chose not to pursue leading edge node developments. They are not in a position to do any better with Intel's Foundry assets. Only thing they can bring is expertise with external customer PDKs and Efficient operation of fabs. I think bringing in Naga Chandrasekhar from Micron and already laid out cost cutting plans by management at Intel are already helping with these issues.

Intel just needs time to execute the current plan. If CHIPS act remains intact under Trump admin (which I am pretty confident it will be, at least for Intel), Intel is not in a dire situation as market thinks it is.

Whatever people think about 5N4Y plan by Pat Gelsinger (3N4Y or 2N4Y), it has enabled Intel to catch up with TSM in some aspects. Intel has already showcased Panther Lake laptops powered by 18A. This node should at least bring parity with what AMD can get from TSM at that time. So their competitiveness in core market will be intact.

CCG team is not doing too bad (Notebook revenue is growing, Desktop revenue is lagging). And all their products are improving with new products in metrics that are important (more efficient, less cooling requirements - only lost in gaming but perf/watt is improving there too). Window EOL is a tailwind here (more so for Intel than AMD due to enterprise focus).

DCAI is still doing okay, roughly $3B in CPU sales and new products are more competitive than old gens and have core count parity with competition. AMD is still doing half of that in DC CPU sales (growing very fast though). They missed out on AI GPU sales. Not all workloads can be accelerated and head node is important for AI GPUs too. If a traditional CPU refresh happens, even better. If AI is as big as everyone says, then they still have time to come out with products and capture some crumbs of a market share like AMD.

NEX may have bottomed or stabilized a bit. so probably revenue accretive in near future quarters.

Intel Foundry is making loss because utilization is low and cost is high, once the products from TSM come back for future Intel products, this business will improve. I remember the CFO said that the 18A has higher ASP (3x) based on comparable node offerings in market and low cost structure to produce on than Intel 7. So that is good. Now whether it breaks even with just Intel volume alone is the big question. I don't think so and it has never been the case imo. And they have some external customers signed up too! They definitely need to work on reducing cost and slow down the expansion here to match external customer demand especially the tool/equipment purchases. Note the depreciation cost of these things are very high compared to empty fab shell space. This will be a grind for the next 5-10 years. With Samsung having their own problems, I think Intel Foundry will do well if they just execute well to get the not TSM crowd as others mentioned here. Some of the action to reduce cost and slow down expansion is already in motion.

I think Intel will be just fine in the next couple of years with focus on executing current roadmaps and planned cost savings. Only caveat is 18A and nodes after that need to deliver at least process parity with TSM in HVM.

 

[count]

Foundry may fail, but products is facing extinction. The world has moved beyond x86 so product is playing in a shrinking market. Absent inventing and then producing what comes after ARM, products is nearing the end. If Intel is to survive, it will be as a Foundry not a Products company.

I don't think this is true. x86 may not be a growth market, but if IBM can still profitably sell mainframes in 2025 with their own z processors I am sure x86 will also remain for a long time to come.

 

[dkr1986]

New stuff from SemiAccurate about someone looking to acquire part of Intel's fab or just equipment (This makes more sense as Intel wrote off lof of equipment associated with Intel 7 in Q3'2024 earnings release).

Another Suitor Talking To Intel

Since SemiAccurate broke the story about Intel being an acquisition target last Friday, there has been a lot of speculation and rumor. While researching the topic we came across, and verified, anot…

www.semiaccurate.com

Since SemiAccurate broke the story about Intel being an acquisition target last Friday, there has been a lot of speculation and rumor. While researching the topic we came across, and verified, another current suitor for Intel fabs.

Now we realize that that in Friday’s piece we said Intel was an acquisition target whole and the above paragraph appears to contradict this. While pulling on threads, SemiAccurate heard quite a few rumors and obvious speculation. Topping this list are people claiming that tech CEOs going to Mar-A-Lago en masse are obviously doing so to talk about Intel and purchasing it. Some of those names correspond to earlier rumors, since disproven, so what else could it possibly be?

Perhaps, you know, tech leaders going to meet the president elect to talk policy? I know it is a stretch but, stay with me here, there was a meeting on this topic to express views and hopefully influence policy. Unfortunately such things do not lend themselves well to the self-aggrandizing and questionable business models of clickbait sites so ‘INTEL GETS THE BUYZ!!!!’ it has to be, right?

Back to sanity, SemiAccurate heard another name that was said to be in active talks with Intel over the fabs. This isn’t to say the company in question is interested in buying them all, it could just be a specific one or unloading unneeded equipment. We were able to verify that this company and Intel are talking about the fabs but not specifically what parts. So who is this mystery company and what does this new twist portend?

 

[Daniel Nenni]

IMO the 5N4Y is just marketing in reality we got 3 Process or 4 if you count Intel 3/4 as different I7/I4/I18A 20A was canned so I wouldn't count it as a process.We can count 4+Power Via as a process it is quite a complex thing in it self it's vauge but they definitely covered Major ground vs competition.

They Will be 15% behind SRAM vs TSMC N2 (ISSCC 25) and around 10% behind Density vs N2 and 5% in PPW behind N2 PPA Estimation from Dan's comment of
N3X<18A<N2 which is very nice
Reference
View attachment 2702

TSMC also has DTCO for N3 and N2 which customers are now using so the PPA deltas will be more:

At the 3nm node, DTCO improvements from 5nm can result in ~25% area reduction for logic circuits, as seen with TSMC's N3 process.

At 2nm and beyond, advanced DTCO techniques (e.g., CFETs and hybrid bonding) may push area savings further, potentially exceeding 30% in certain applications.

These are big numbers and I do know some customers who are using it at N3 and they are impressed. PDKs don't lie but PPT slides do.

 

[XYang2023]

Foundry may fail, but products is facing extinction. The world has moved beyond x86 so product is playing in a shrinking market. Absent inventing and then producing what comes after ARM, products is nearing the end. If Intel is to survive, it will be as a Foundry not a Products company.

I don't think the debate between x86 and ARM matters much, at least not to me.

At the moment, there are at least 5–10 applications on my PC that I know only work on x86 (ranging from networking tools to CAD software).

As a user, I just want good products. I don’t care about the underlying architecture.

Lunar Lake is impressive. I am considering the following product to replace both my laptop and iPad Pro:

I don’t need to worry about compatibility issues, and I know my software will just work. I can use it for Davinci Resolve to edit 4K videos and play games on the go. The battery life is excellent too.

For heavy compute tasks (running large LLMs or training), I remote into my workstation, which also runs x86. When I’m on the go, I use my phone as a hotspot to connect to other machines. If I need additional compute resources, I rely on the cloud, where x86 is pervasive.

For Intel, it just needs to focus on producing better products. For example, I’m also interested in Panther Lake with Xe3 iGPU (laptops and other small form factors) that could support large CAMM memory configurations (e.g., 64GB or larger). This would allow me to run large LLMs directly on such devices.

This highlights another important point: graphics are very important. I would prefer Intel to allocate more die area to the GPU. However, having an NPU is acceptable for running inference tasks at low power. Ultimately, it’s not about a single component—it’s about multiple component working together to make the device truly useful.

 

[blueone]

GFS has too much stake from Saudi money and chose not to pursue leading edge node developments.

Actually, it's UAE money. Also, DCAI is not doing "okay", DCAI is in trouble and getting worse.

 

[dkr1986]

Actually, it's UAE money. Also, DCAI is not doing "okay", DCAI is in trouble and getting worse.

Oops on the confusion about UAE/Saudi

I understand the situation with share loss to AMD but like I said Intel is still selling $3.3B in DC CPU with inferior product compared to AMD offering (AMD still sells about half in revenue but with better margins and growing fast).

GNR (-~18% perf/watt) vs Turin is much better than EMR/SPR (-~40+% perf/watt) vs Genoa (based on GeoMean from Phoronix reviews). Also GNR has core count parity with Turin now.
One thing I recently learned was Phoronix reviews does not use any of the accelerators in GNR which Turin lacks. I am not sure how that affects these numbers probably to Intel's benefit (just a guess).

If Intel Sales and Marketing manage to hold the fort with EMR/SPR in DC market share in Cloud & Enterprise, I am just thinking they will do a better job with GNR. We will see, the Q4'24 earnings and next couple of quarter financial results will give us an indication if this segment is doing okay or in real trouble!

 

[MKWVentures]

I understand the situation with share loss to AMD but like I said Intel is still selling $3.3B in DC CPU with inferior product compared to AMD offering (AMD still sells about half in revenue but with better margins and growing fast).

GNR (-~18% perf/watt) vs Turin is much better than EMR/SPR (-~40+% perf/watt) vs Genoa (based on GeoMean from Phoronix reviews). Also GNR has core count parity with Turin now.
One thing I recently learned was Phoronix reviews does not use any of the accelerators in GNR which Turin lacks. I am not sure how that affects these numbers probably to Intel's benefit (just a guess).



If Intel Sales and Marketing manage to hold the fort with EMR/SPR in DC market share in Cloud & Enterprise, I am just thinking they will do a better job with GNR. We will see, the Q4'24 earnings and next couple of quarter financial results will give us an indication if this segment is doing okay or in real trouble!

How does the Xeon 6 Sales volume compare to the Xeon 5 (and earlier) Sales Volume today? Any numbers? Datacenter processors are notoriously slow to ramp

 

[dkr1986]

How does the Xeon 6 Sales volume compare to the Xeon 5 (and earlier) Sales Volume today? Any numbers? Datacenter processors are notoriously slow to ramp

I don't have access to that kind of insights into sales volume in units of each Xeons etc.

This following article says Intel did well in Q3'24 in terms of volume shipments but ASP is still lacking (explains the low margin on DCAI).

The Server Recession Ends, And Both Intel And AMD Won

Anyone who thinks that Intel is easy to kill need look no further than the historical trends of the Mercury Research market share statistics that we see

www.nextplatform.com

In the third quarter, Intel shipped 4.09 million X86 server CPUs, which was up 15.3 percent year on year and up 9.8 percent sequentially from the 3.55 million chips that the company got out the door in Q3 2023. As you can see from the chart above, which shows a kind of Mandelbrot fractal line chart of Intel and AMD market share overlaid on top of a bar chart of aggregate X86 server CPU shipments from Q1 2001 through Q3 2024, it looks like Intel bottomed out in Q3 2023 and again in Q1 2024, when it sold 3.46 million CPUs. Clearly, the “Granite Rapids” and “Sierra Forest” Xeon 6 processors are being better received than their “Sapphire Rapids” predecessors – at least from the early data. We will know more for sure when we see Q4 2024 and Q1 2025 data.

In terms of shipments, AMD actually grew slower than Intel in the third quarter of this year, but kept pace just fine. According to the Mercury Research data, AMD peddled 1.39 million Epyc processors in Q3 2024, up 14.4 percent year on year from 1.22 million CPUs in Q3 2023 and up 7.1 percent sequentially from Q2 2024. Total X86 server CPU sales in the period rose by 15.1 percent to 5.48 million units, which was also a 9.1 percent sequential bump.

Now, when you start talking about money, Intel is not doing so well, and our financial analysis for Q3 2024 for Intel (see here) and for AMD (see there) showed this very clearly. Take a gander at the revenue streams from those X86 server CPUs over time:

Regarding ramp up of Xeon 6s. It does look like a very slow ramp.
What I heard (anecdotally) is that GNR CPUs are hard to come by compared to AMD's Turin but both launched 1 or 2 weeks apart in end of September 2024 timeframe. GNR was always out of stock online as well. It looks like GNR was a paper launch.

Even SuperMicro said they are shipping Xeon 6900P (GNR) servers in volume only recently in first week of January.

https://twitter.com/x/status/1877483113195929693

So I think volume mix of Xeon 6900P (GNR) is going to be still low in Q4'24 compared to older Xeons.

I have no idea what market, SRF (Xeon 6700E) is targeted at and how it is doing. I only read online that was created mainly for Meta but that business has gone heavily to AMD's compact cores server CPUs (Bergamo & Turin Dense now). If that is true, no idea if Intel can switch them back with either upcoming 288 core/ already launched 144 core SRFs.

 

[MKWVentures]

"" ramp up of Xeon 6s. It does look like a very slow ramp.
What I heard (anecdotally) is that GNR CPUs are hard to come by compared to AMD's Turin but both launched 1 or 2 weeks apart in end of September 2024 timeframe. GNR was always out of stock online as well. It looks like GNR was a paper launch.

Even SuperMicro said they are shipping Xeon 6900P (GNR) servers in volume only recently in first week of January.""



@dkr1986

Good info, thanks!

 

[XYang2023]

The tone is slightly better than neural.

"cash reserves look stable for now, thanks to capex and opex cuts, the dividend suspension, government funding, and private equity injections via SCIP deals in Arizona and Ireland."

‘Don’t Get Too Excited,’ Says Top Analyst About Intel Stock Amid Buyout Speculations - TipRanks.com

Intel (NASDAQ:INTC) has seen its newsflow go into overdrive recently. A report from industry blog SemiAccurate speculates that Elon Musk might be a potential M...

www.tipranks.com

 

[siliconbruh999]

How does the Xeon 6 Sales volume compare to the Xeon 5 (and earlier) Sales Volume today? Any numbers? Datacenter processors are notoriously slow to ramp

Same Applies to Turin as well DC is slow we will know with Q4 Earnings

 

[siliconbruh999]

Foundry may fail, but products is facing extinction. The world has moved beyond x86 so product is playing in a shrinking market. Absent inventing and then producing what comes after ARM, products is nearing the end. If Intel is to survive, it will be as a Foundry not a Products company.

GPU diversification is a long term play x86 is not dead for foreseeable future it is shrinking yes but they need to take the fight back to ARM if ARM can invade desktop than why x86 can't invade mobile and stuff