True.

You have to consider that AMD competes on price in the datacenter, especially at the high end. This means that Intel had to lower prices but AMD stll wins designs because of the better performance and efficiency. You can't just substract AMD's gain from Intel's losses. Intel feels the competition double, by losing units sold but also by having to compete on price. Especially painful in the high margin server business. This can become an avalanche quickly.

Disclosure: I went short INTC (small position) now because of exactly this.

Yes, let's look at some numbers instead of chart babble .

AMD did well. Here's the snippet that Intel investors should maybe worry about the most:



If Intel had a different tone on their call, now you know why. To me it seems a bit odd that Intel claims that market conditions made for their decline in the data center and not competition. I think, at a certain point, outright lying to your shareholders does more harm than just stating the truth.

Intel's cash cow is faltering. Be aware of that (no matter what the chart "tells" you).

No comments on the lastest report? Especially the DCG numbers? Only trader babble here ...

That article is very interesting. It basically states that Swan doesn't expect Intel to compete with AMD for the foreseeable future. That basically means they are giving in, leaving their side open for AMD to attack. They want to concentrate on the foundry side of things, at least that's how I interpret it (not possible to capture 30% of the global semiconductor market without foundry and memory business - they sold their memory business to Micron lately, again at the most stupid timing one could imagine.

For AMD, this is very good news (ok, it may have been meant as a distraction, but Intel staff got the message too, so ...). This is very, very dangerous for Intel. They leave their core cash cow business open for competition to prioritize seeking for new business in an area where Intel has failed in the past. Foundry business was meant to be an important source of revenue for Intel for many years already, but Krzanich didn't succeed with it really. To compete with the big foundries, you also have to be flexible and cost competitive, which is very hard to do. Intel doesn't seem to have the skills for it, otherwise they would have succeeded while they had a competitive edge in semiconductor process tech (at 22nm and 14nm). Now they are even behind in this field (foundries at 7nm and transitioning to EUV and 5nm soon). In such a situation and stating to compromise one's core business (which is raking in all those dollars needed for new process tech) is absolutely nuts. If Intel loses more to AMD in market share, they won't have the money to compete with TSMC and Samsung, falling behind further and fully leaving their core business to AMD.

Stay away!

The Versal is shipped to customers since summer. How many 10nm FPGAS has Intel/Altera shipped so far?

The Radeon Instinct is a high performance server part, not something you buy from ebay. AMD has many consumer cards that are 7nm based that you can find online easily if you want to buy them. Those were examples anyway, since you wrote that there wouldn't be larger chips made on TSMC 7nm. That is simply not true.


Yeah, 40% versus your projected 10% (which I would doubt) from Intel. TSMC has around 50% of the foundry market, which is huge. 40% of that business is a lot of wafer starts. Definitely more than 10 or whatever percent from Intel - and those are only rather low clock, low yield parts.


Well, I wouldn't call the Samsung Note 10 with its Exynos a low volume part. This is also only an example. Fact is that Samsung is ahead of the crowd, TSMC follows close and Intel fell back at least a year.


Well, that's all you need to know. Bad yields are a showstopper. It affects economics and throughput. The fact that they are not in full control of the process also tells you why they have performance/clock issues and that certainly is a problem for Intel with its 4+ GHZ designs.

AMD has 4+ GHZ designs in the market, based on 7nm TSMC, shipping in volumes. That's the problem and this is what Intel can't compete against at the moment. AMD just released Threadripper with 24 and 32 cores, going against Intel 18 core and crushing it in terms of performance and power/performance. They are even going to release a 64 core Threadripper early next year and Intel has nothing to go against it. Same on the server front against EPYC Rome.

No matter how you twist and turn this, Intel fell behind and is in trouble. They are consistently lowering prices of their cash cow products. Desktop business has already taken beating, server will be next.

AMD has 7nm GPUs in the market for quite some time already. The Radeon Instinct is a 330mm2 die and has been released about a year ago. Xilinx has 7nm FPGA products available.


No I don't but Intel sells very high ASP products versus TSMC selling wafers. INTC still, as we know, has, by far, its biggest chunk of products/chips still on 14nm (+++ or whatever). No way they have remotely near as many wafer starts at 10nm as does TSMC have at 7nm. Face it: TSMC is way ahead and this is because Intel failed at 10nm. They wanted to be in mass production years ago.


Don't know (though I'd say most phones Apple sells these days have 7nm based chipsets). Apple is not the only one producing 7nm based mobile chips. Huawei is another one, just to name another big one. They even have a middle class 7nm based chipset.

Again: No way Intel processes as many wafers in its 10nm as does TSMC process in its 7nm. Never!

Not just that, but Samsung is in 7nm EUV (!) volume production already. Neither TSMC nor Intel are in EUV volume production yet.


Ok, now you're referring to L1 cache instead of just cache in general. I doubt it has anything to do with Intel's issues on 10nm. AMD gets a nice speed for its 7nm L1 cache and the EPYC has a huge total cache size. This certainly isn't the reason why Intel has issues on 10nm. It is not the design side of things, which they would have already fixed, but it is the process itself which they just can't get under control. They will, eventually, no doubt, but in the meantime, AMD release a whole bunch of products at 7nm and is about to release its next generation processors based on Zen 3 cores next year. Intel has a lot of catch up to do, which costs them high ASP products and high margin business in servers.

As to performance per watt, AMD has very good numbers there with its 7nm products. Intel can't compete there at the moment. They probably will at 10nm, but as I said, AMD is not standing still. If Zen 3 really provides more than 10% higher IPC, Intel has a big problem.

I don't think you should conclude from two products on the entire process range. There are also FPGAs and GPUs made on TSMC 7nm, which are tiypically a lot bigger.

It should be clear, in general, that with increasing amounts of transistors, the likelihood of defects increases and therefore larger chips will face yields that are worse compared to a mature larger process node with the same chip size but half or less of transistors squeezed into it. Therefore the art of getting high end products from small as possible dies is what makes the difference. If Intel failed on that (I think there is more to it at 10nm though), it is still a fail. AMD saw this coming much earlier and their chiplet approach proves to be very successful. Also helps with yields and keeping different parts (with extra costs from foundries) low.


Certainly not. Wafer capacity from TSMC is way higher than from Intel. Intel sells products with much higher ASPs than what TSMC gets from its customers. No way Intel processes more wafers than TSMC. Not even remotely.


How is an increased cache limiting clock speed? Have you had a look at how much cache AMD integrates on 7nm EPYC? It is huge and clocks are just fine. Intel has an issue with the maturity of its process, therefore they can't get enough yield for the fast corner, whatever the reasons are.

I think AMD already announced that they will ship 7nm mobile parts beginning of next year. Don't forget that AMD needs to integrate the GPU so this is a completely new design compared to the current chiplets with its 12nm interface chip. Also the platform integration for mobile takes longer and the notebook makers need some integration and ramp up time up front. The first mobile Ryzen parts also took much longer to come into the market than the desktop parts.

As to capacity of TSMC 7nm, rest assured that it is way higher than Intel 10nm. TSMC is producing 7nm chips for many customers for quite some time already and AMD is just one of many. I doubt that Intel will ever come even close to the volume TSMC makes on its high end processes in terms of wafers sold. One of the reasons why fabless producers can still have a good cost position compared to big fab owning ones. Economics of scale.

The current problem with Intel's 10nm process doesn't seem to be availability but its speed. All 10nm parts available so far seem to clock pretty low and that is something Intel needs to get fixed asap.

AMD's stock had a very nice run up in the past days and weeks, by the way. I am really glad I changed ship from Intel to AMD. AMD will have a resurgent game console business by second half of next year and some nice high margin server design wins with EPYC Rome. Valuation seems a bit lofty at the moment, though.

Back to Intel: Their announcement of Ponte Vecchio looks quite ambitious. Could help them to protect their server business better and get a share of the server GPU business. We'll have to wait for real life performance and prices though. They have to do many things right this time and not fail in almost everything like they have in the past few years.

Why do you think this is the case? EUV steppers can only be had from ASML and from what I have read, they require very big clean rooms which are expensive and this essentially means that, for EUV, the industry has to build new fabs instead of upgrading existing ones.

To put it differently: Without new fabs they won't get the required production capacity.

You can be sure though that there is a lot more needed than just placing EUV steppers in your clean room to make 7nm production work.

I think this part of the report could be essential for Rome's success:

Customers can use less sockets or increase performance with the same amount of sockets while actually saving power and money. There isn't really much left for Intel in terms of arguments, other than some AI and AVX512 niche maybe. Intel will answer this with lower pricing, official or unofficial, but it's going to hurt their cash cow margins in the server business. Money they desperately need to finance 7nm and beyond. How are they supposed to compete with the foundries, TSMC and Samsung to name them? Those have a much larger business to pay for their investments. Intel failed as a foundry, Intel failed in mobile (and FPGA) and now they have to pay the price. Very bad situation they put themselves into.

No words on EPYC Rome benchmarks and announcements? All major server maker announced multiple new offers based on it. Google is officially a customer now as well, with a full custom version of it. All major cloud provider now offer AMD based server instances. PCI Express 4 datacenter SSDs with speed up to 8GB/s ready available from Samsung (and others) now. PCIE 4 provides very fast datalink between AMD datacenter GPUs and CPUs. Total cost of ownership according to the presentation about half of what Intel offers today (one time but also running costs, like energy). Benchmarks beat Intel to the ground. Looks like a very good lineup to me.

Where is Intel? What about its cash cow? Do you really want to continue your chat about drawing chart lines?

Looks like Intel starts feeling the heat. Funny they tried hard making it look like their weak guidance had nothing to do with AMD. Well, the stockmarket didn't seem to believe it anyway (flummy did, though).

First benchmarks from Ryzen 3000 leaked. Only the mid range part Ryzen 5 with six cores but the single thread performance already shows that Intel just has lost its single thread lead. So, what lead is there left? No semi process lead, no performance lead, no lead on price, no lead on cost either. Not a position you want to be in as a monopolist.

Nice read: Xeon vs. Rome. Some benchmarks, tricks and price indication. Rome is at least on par with Intel's fastest server processor but likely much cheaper (about half maybe) and consumes significantly less power. It clearly is the performance/power/price winner. Hard times ahead for Intel's server business.

Yes, but if I recall, Intel guided to cut on Capex and revenue guidance was week as well. Some of the reasons why the stock price tanked. We'll see how that turns out with increased competition from AMD. AMD just announced a Zen 2 based 16 core CPU to be released in a few months which will clock at up to 4.7 GHz with improved IPC from Zen 2. That's basically taking away every bit of the tiny performance lead (in single threads mainly) Intel had so far. The power envelope also seems to be way better than what Intel can offer today. Not much left for Intel to keep prices(margins) high.

If you look at the cash flow statement from Intel, you can see that they are spending 15 billion $ on Capex, it's true. That's not necessarily only for fabs, but the most part will be, I agree. Intel has a free cashflow of about 15 billion, not exactly starving either, but they are paying out a lot for dividends and buybacks. I am really not sure if that is the right strategy taking into account the issues this company is facing. Be prepared to have much lower pay out in the future, which would be a reasonable thing in my view.

Intel is a slow ship, clearly, so they will try to fight back AMD as hard as they can. Intel, without a doubt, has the capability to do so, but I did think that also in the past when Intel wanted to conquer the mobile and foundry market. None of it happened, big fails, so the big ship doesn't just seem to be moving slow but also to be short on armour.

We'll see if Intel finally finds a way to use its deprecated fabs for external customers while they move to EUV for their own products. Interest in their production services, so far, seems to be low and the issues with making Altera products in Intel fabs, with years of delays, don't really seem convincing to me.

We'll see if the beancounter and this board of directors will be capable of turning the ship. I definitely wouldn't bet on it.

Definitely. TSMC is losing market share to Samsung. Just recently, NVIDIA announced they would be moving to Samsung for 7nm EUV. Looks like Samsung is about to overtake TSMC there (first EUV based process).

There is only Samsung and TSMC left at the leading edge foundry race. Intel is a bit behind and tries to be in the foundry business without success so far. That leaves plenty of capacity to AMD and the likes to compete with Intel at reasonable prices. Intel has to pay plenty for its fabs. Samsung recently announced it would be spending 120$ billion (!) until 2030 into new fabs. That is definitely too much for TSMC and also not possible for Intel with a stagnating or even shrinking business. AMD can easily use Samsung as a foundry partner, or TSMC, whoever is ahead.

Yes, AMD for instance!

Good value for money are Samsung and Micron at the moment. In my opinion.

No problem with that statement, though we'll still have to wait for actual performance (3rd party benchmarks), power and prices until release.

What you should note, though, is how sentiment has changed. In the past, AMD was no real competition for Intel. Then, Zen 1 was released and suddenly, they were close, real close to what Intel had. Many Intel bulls thought, this would be a one time thing and Intel will soon be back on track and crush AMD. That didn't happen. On the contrary, AMD now is first on the process side (with help of the foundries), often excels on performance, e.g. on APU side clearly, multithreading side mainly, core count for servers being the first, power efficiency (think: server) and now seems to be even on par with single threading performance to Intel. There really is no domain left where Intel is leading, but they still have to charge their sky high prices while AMD can compete in price aggressively, since they don't have to pay for owning expensive fabs.

A lot has changed in only about two years time. It's not that Intel has been beaten to death, but their lack of leadership and drive is shown exactly when they can't beat a small company like AMD anymore. This can get ugly, eventually.

AMD announced 3rd Gen Ryzen and EPYC CPUs. Bottom line: Matches Intel Cores in single thread performance now and surpasses them in multithreading.

Also notable to me seems to be the improvements on the graphics side of things. A redesigned compute unit delivers much better performance/watt. That is certainly tweaked towards game consoles, where it is likely going to be included as integrated graphics, just like it was with PS4 and Xbox One, but it also brings AMD close to NVIDIAs performance. NVIDIA is a much tougher competitor than Intel, though.

Intel wants to compete in the graphics business, as you all know. I suppose this is mainly for the server business, where Intel failed (once more) with Xeon Phi. Let's see what they can do in short time in a business where AMD and NVIDIA fight with their teeth for decades already. Hiring some experts from the competition won't be sufficient I suppose.

One big advantage AMD has is the combination of CPU and GPU. They can tune their platforms toward specific high level applications, where CPU and GPU must work together at massive performance and scale. That's something Intel + NVIDIA will have a hard time at to match. Cloud gaming is an example, where AMD already had some high profile design wins.

Intel will need substantial architectural and semi process improvements soon to get back to the top. Server without high end graphics is going to be tough for them in the future. AMD has proven now that they can deliver highest performance cores and graphics for data centers across multiple generations. I bet a lot of data center companies are considering AMD these days. They'll get the highest performance CPU+GPU solution from a single provider.

AMD faced serious headwinds from the gaming consoles end of cycle and from the crypto crash. Consoles are going to be back by end of next year, with the release of the next console generation. That will also fuel demand on the PC side, where people will need to upgrade in order to play latest blockbuster titles at high settings (same happened with every new console generation). Crypto may still be weak but stuffed channels will recover and demand for gaming graphics cards will likely improve then as well. A lot to be gained for AMD.

Oil certainly doesn't seem to be the worst bet at the moment, at least if you don't think that Musk will save the world (I don't).

Agree on the dictator and the only policy to agree with. Hope this won't end like it did in Turkey or Russia.

AMD's PE is meaningless, since the company operates at the border of profitability. There is a lot of headwind they are facing currently from crypto and end of the console cycle.

The console cycle will restart by end of next year (Sony already announced PS5), where sales will improve significantly. The crypto issue is tougher. I don't believe in that hype, but the stuffed channels should soon be depleted and the GPU business at least be back to normal. The icing is market share gains from Intel, where AMD is doing very well (and the reason why the didn't fall back into losses like they did in the past). Economic cycles don't matter much to AMD. It is much more about market share for them.

Ah, and the trade war: AMD has this joint venture with China in place, where they are paid license fees for their processor IP. That's big business for AMD when servers in China will replace Intel cores with those homemade ones based on AMD's IP.

Well, I've been an Intel bull (I try hard to never fall in love with an investment) since end of 2012. I didn't believe the PC is dead and that it will be replaced by tablets and phones bullshit (where I was right), knew that ARM wouldn't have a chance in datacenter (right so far). But I was especially bullish about Intel's semiconductor lead (gone by now), its prospect in mobile (big fail), its high potential as high end foundry (big fail again) and at least some IoT business (zero). I have seen Krzanich making a stupid acquisition one after another (drone startups, Mobileye, Altera and the likes) and not fixing any of its weaknesses. Then I have seen them screwing up on 14nm and even more so on 10nm, while I was worried about AMD overtaking Intel on the right lane (which just happened). That's where I thought that AMD is the stock with much higher potential and even shorting Intel would be worse than putting it straight into AMD's stock (where I was right).

That's where I sold my Intel shares (at around 100% gain excluding dividends) and put that money into AMD, plus adding some levered calls. AMD's share price about doubled from there, as you may recall (a flyer is a flyer) and sold the levered calls in the 30$ range. After the dip, I bought (too early, I admit) levered calls again. They are at about +150% at the moment, if you like to know.

In the meantime, Krzanich was fired because he got this diversity thing plain wrong and has been replaced by the beancounter, after months of unsuccessful "search" for a decent CEO. Really bullish!

This is not about being right or wrong, this is about having actual reasons for your actions. Your reasoning seems to be about as deep as Weedtrader's. Actually this site should be called TradersHub instead of InvestorsHub.


That's fanboy, without the "s" and with different spelling.

I am actually more into total return, but that may be too reasonable for you. How many years of dividends do you need to make up for those "dips", uh - I mean "buying opportunities", of the past three weeks?

Well, simply holding AMD with additional levered calls. Doing really good, and you? Keep enjoying your dips!

Now that's a dip that Flum and Weedie can buy, isn't it? Enjoy!

Wow, your stuff seems to be really strong! Not sure if it helps with your investment decisions though.

While nobody here seems to care about actually reading the financial report of the company they are invested in, I had a look into it and try to summarize it.

Revenue flat from last year (guess you heard that already).
Gross margin down 4% from last year. Now at 56%. That's a bad one. Costs increase, maybe some pricing pressure (AMD?).
R&D spend down 7% (not sure if that's a good idea when you're losing your semi lead)
Net income down 11%, EPS down 6% (due to buybacks I suppose)
PC client group revenue is up 4% (what was the overall market growth in comparison?). PCG contains the modem business revenue, which was just shut down - don't expect better numbers next quarter!
Data center group revenue was down 6% (demand or competition from AMD?)
Even the Altera business is down 2%. Overpaid for it and then delayed 14nm projects for years - great addition to its business!

First volume 10nm processor (Ice Lake) shifted to end of the year. Originally their 10nm process was meant to be ready in 2016! That's how they lost their lead. TSMC and Samsung already moving to EUV processes.

Uh, by the way: Samsung announced to be investing 115 billion $ into next gen semiconductor fabs . That's about 10 billion per year, half of Intel's current yearly operating cash flow. If they want to compete with that level of investment, they should stop paying a dividend and buying back shares asap. That would require some guts from the CEO, though.

Units shipped in the notebook segment were down -7% QoQ, -8% QoQ for the desktop segment. That means that the revenue increase was achieved by price increases, higher ASPs. I suppose this is related to their 14nm capacity limitations, so they chose to sell higher ASP parts, but also due to more competition from AMD, which seems to jump in and sell in the lower ASP segment (while AMD competes with lower prices at comparable/better performance anyway).

The big question is: Once 10nm is up and running and Intel's capacity limitations are gone, will the OEMs, which now have partially switched their product line to AMD platforms, switch back to 100% Intel? I doubt it. AMD has its foot in the door now and Intel supported it with its 10nm failure.

But who cares? Flum will rather listen to weedtrader than to anything I have written here.

Hell, were you right! On the next dip, you can buy even more for the same money. Isn't that awesome? But don't get too paranoid from your skunk!

Yes, go and buy the dip! Would you mind to share some with your mate flum?

@weedtrader
Really, you should trade more and smoke less of it.

The problem with the 5G modems from Intel is that they aren't integrated with mobile APUs. Since Intel gave up on mobile phone APUs, there is basically only Apple left which uses a descrete modem in the volume market. I just don't see a big opportunity for Intel here. They screwed this up a long time ago. Only chance I would see is selling it as IP, but then they need to support foundry processes, not just their own.

Depends on how you define leadership. If its by revenue, maybe, but that really isn't important going forward. Technologically, Intel lost the leadership in semis already, and that's to TSMC mostly, which generates a lot less revenue than Intel. Intel's revenue doesn't only have to pay for semiconductor process technology, but also for designing the processors, marketing and all the other stuff that distracts Intel from it's core business (things like drones, auto pilots, virus scanners - put in here all the failed acquisitions by Intel of the past years).

What matters is if Intel can make up for the design fails (i.e. insufficient improvements) of their processors with superior process tech, like they did in the mid of the first decade of the 2000s. AMD obviously is on par regarding processor design and they have access to at least the same advanced process tech as Intel does (even a bit ahead). There is absolutely nothing on the horizon that tells me that Intel is going to solve this issue. Don't know about you. Sheer hope won't do it.

The numbers in the table are actually from Mercury Research but I get what you are saying: If it doesn't support your position, discredit the source. It belongs to this self-confirmation bias.

In case you are interested in actual numbers about market share changes Intel/AMD. Or you keep throwing out arbitrary price targets and follow them blindly (no names mentioned this time).

Self confirmation bias at work

Ok, that explains a lot. I suppose that's why your price targets make a lot of sense for you.

Yeah, keep throwing in arbitrary target prices without any reasoning. Are you trading your weed or smoking it?

In my opinion, Intel is one of those companies that should hire its CEO from internal. I know, Krzanich was a bad choice but that doesn't mean there are no good/better candidates available internally. I mean, Intel employs how many workers? 100k? You'd suspect there is at least one decent choice available.

Intel is very specific, due to its unique position in a very specific technology field (highest performance semiconductors/processors). They have their own chip manufacturing and are run very differently to all sorts of fabless or foundry companies. I think someone who really understands Intel from the inside and its market would be the best choice. They certainly would have to choose someone with good capabilities and a feel for the business, not like Krzanich.

Not sure if this was the case with Krzanich, but from my experience, often bad leaders are chosen by the former CEO, mainly due to personal preference, which never is a good way of selection. An internal application/assessment open to everyone is the best way in my opinion.

Wow, that guy can read P/E ratios. What a genius analyst!

When I first heard about the logic stacking, I first thought is was true 3D, like flash, but it just seems to be chip stacking with TSVs (through silicon vias). I wonder how they can call this something new for logic chips, since that technology is a few years old and already in broad use to my knowledge. Still don't understand what the actual twist should be.

Stacking chips is an issue for thermal design, so I would expect that only for low TDP and area constrained designs, not high end ones like desktops and servers. SDRAM stacked on top of the processor will likely make cooling more difficult, so I don't see a benefit performance wise. It may improve area, power and also performance for lower TDP packages though. Memory performance can be increased with shorter connections, wich is also the reason why some LPDDR4 in mobile devices are actually faster than desktop DDR4 DIMMs. For the same reason, GPU memory can be made faster.

AMD could still use dies with defective cores/cache and combine them to a lower overall core count part, e.g. a 48 core variant or something, with the same amount of 7nm dies on the carrier. Whether this makes sense depends on many things, like yields (you need enough defective parts that are still usable) and platform costs for the Rome socket. Since it seems to be downward compatible to Naples, that shouldn't be an issue. The 48 core variant would be in between the current high end Naples with 32 cores and would also benefit from the 7nm improvements (and also the Zen 2 cores). They can further improve effective yield by separating the high and lower clocking parts for different SKUs, as well as for different power consuming parts (process corners). The Rome design gives a lot of options to AMD to really make use of the scrap.

I recently read an article about wafer costs between 28nm and 14nm Finfets. The wafer costs are said to be around 2000$ for 28nm compared to 7000$ for 14nm Finfets. This means that the costs are not compensated by the shrink, I think this gives you the trend. I would expect 7nm to be a lot more expensive than 14nm, maybe also not compensating for the gains by the shrink either. This means that the only reason to shrink further is performance gains, not economics, leaving 7nm designs for the high end for a while. Will be interesting how Intel's economics compare at 10nm, but I believe they'll have the same issue, therefore concentrating on the high margin/high end designs first, once their process yield is sufficient (for now, they have to experiment with smaller and lower clocked designs like Core M until they get the yields for performance parts).

A lot of speculation involved, but I wouldn't expect AMD to introduce Zen 2 based mainstream cores on 7nm first. Instead, they would stay at 12nm for mainstream Zen2 and introduce 7nm designs (Ryzen 7 and below) only when Intel forces them to do. We'll see if I am right with that assumption. A 7nm Ryzen 9 could be an option to fight back Intel's Core i9.

It's also interesting that AMD once mentioned that the 10nm (foundry) node is not of interest for them. It provides for better density, I know that from a comparison, so there must be other reasons for it. Could be performance or economics. 7nm seems to be a different animal though and that's most certainly not because of economics.

Wasn't me!

By the way: A Rome carrier shot has been shown by Lisa Su. It really uses 8 8-core 7nm dies together with a (huge) interconnect. That is, again, a pretty unexpected/unorthodox design. AMD is really thinking out of the box. The interconnect die, which is huge, is rumored to still be a 14nm design. This may be to use the high performance, but still expensive, 7nm process only for the parts where it matters most - the CPU cores, and leave the rest of the peripherals on a cheaper and lower performance process. Smart move it seems. It makes me wonder how AMD can reuse those 7nm dies for other designs, though. I expect those 7nm chiplets to always need an interconnect die of some sort, which likely leaves 7nm to the highest end many core designs. Maybe some more lower bin EPYCs and Threadrippers, but no Ryzen 3/5/7, I would expect.

Could be that AMD will wait at least a year or two for mainstream designs in 7nm. We'll see about that.

Performance wise, Rome is beating a high end dual socket Intel server platform in a rendering application. It was also compared to a EPYC 1 dual socket system, where Rome also came in first (with the same overall core count). That looks like a compute beast indeed.

Intel will be reacting with a 48 core, dual chip(let) CPU by mid of next year. Total cost of ownership will be crucial here, taking into account power consumption. Will be interesting how Rome on 7nm does there compared to Intel.

Different story: Just today I was told that Intel is offering wafer slots for foundry business in 22nm Finfet. Those slots are pretty large, though, so a customer needs quite some volume to get in. Still interesting that, slowly and quietly, Intel seems to get into the foundry business. Took them quite a while, but I would see this as positive for Intel.

Mercury Research released market data for PC CPUs, indicating that AMD is well above 10% unit market share in desktops and notebooks. Numbers increasing from quarter to quarter, almost 50% unit increase for AMD within a year. Also growing from last quarter sequentially, so the story seems intact. Server numbers not publically available. I'd consider server adoption to be slower than PCs, since qualification is needed by operators, which takes time.

Ah, and Intel announced a 48 core server CPU residing on two dies, similar to AMD's chiplet approach. This is to conquer AMD's Rome server CPU in 7nm it seems, which provides 64 cores. There are rumors that AMD's Rome will be distributed across 8 dies with a dedicated interconnect IC. If this is true, it would be a good way for AMD to improve yields of its 7nm dies, which can be reused for other designs, saving a lot of those expensive mask costs (a lot of multi patterning needed on 7nm TSMC, which still uses 193nm immersion lithography). Will be tough to get the performance right for AMD, with all that wiring.

Who would have thought that this chiplet approach could be crucial for competitiveness some years ago? Nice foresight from Lisa Su it seems

Well, I am holding the stock from the lower teens but I was selling my levered longs when the stock was at around 30$ (bought when base price was in the lower teens), which proved as good timing (lucky or not, leave that to you). Spent part of the gains for levered longs at around 24$. Those are currently under water, I admit that. Over all, I am fine, thanks for asking!

The soft result was due to headwinds from GPUs because of crypto winter and very soft semicustom business due to end of cycle in consoles. GPUs will recover once inventories have been depleted and new consoles are to be released in one or two years, likely with AMD chips inside.

What is important, also to Intel investors, is that AMD is in the single digits of server market share and nicely growing. Think it was in the upper double digits of sequential growth. Also client and notebook unit share are growing nicely according to the call. 7nm EPYC rome stays on track for beginning of next year.

It's a rather slow development than what the sudden rush of the share price suggested. Needs time.

Where is Intel's growth in the coming years coming from? Want to elaborate?