Not just TSMC and Intel, Europe attracts all chip makers

Not just TSMC and Intel, Europe attracts all chip makers

Not just TSMC and Intel

Apparently, the European territory is now frowned upon by many manufacturers as a new base for the production of semiconductors. Recently, Intel announced its investment plans in Europe: there is talk of 80 billion euros over the next 10 years, which will be invested throughout the semiconductor supply chain, from research and development departments to the production and packaging of chips. . Seventeen billion will be invested in Germany for the construction of a mega-plant for the production of semiconductors, in France for the creation of a design and research and development hub and in Ireland, Poland, Spain and Italy for the creation of R&D plants. , manufacturing, foundry services and back-end stages of production. Furthermore, TSMC is also considering opening a new plant in Europe, although at the moment there are no concrete plans yet.

Credit: Kristoffer Tripplaar / Sipa USA According to what reported by Bloomberg, a potential joint venture between GlobalFoundries and STMicroelectronics could give light to a new European state-of-the-art semiconductor manufacturing facility by leveraging grants under the European Chips Act. This new plant is likely to be built in France, given that STMicroelectronics, based in Geneva, is an Italian-French conglomerate. STMicroelectronics is involved in the design of various types of chips, from MCUs to microprocessors, sensors, MEMS-based devices and other electronic components for electric vehicles and the space industry. On the contrary, GlobalFoundries has focused on specialized processes and nodes, such as FD-SOI technology.| ); }

Intel Is An Excellent Long-Term Investment

President Biden Speaks On Rebuilding Country"s Supply Chains In U.S.

Chip Somodevilla/Getty Images News

With Intel (NASDAQ:INTC) currently selling below $40, as of this writing, I find myself wondering why this would be so. So, although I strongly recommend buying Intel at this time, it's probably more instructive to look at why people are skittish about this stock, and consider the merits of these reasons. We'll also look at strengths, most are pretty obvious, but given the current price, the negatives are clearly dominating overall thinking.

The current issues Intel has could be summarized as follows: A macroeconomic environment battling inflation and post COVID buying patterns. Lower margins in the low to mid-term. Loss of server market share. Greater competition from AMD (AMD) in the x86 market. I will comment on each below.

Or, try to. I really have no greater insight into the current economy than most people reading this. I don't see a golden opportunity for Intel to avoid the worst of it, especially since Chromebooks are much less popular in the post-COVID economy. Mid-term elections will likely turn over the house, and possibly the senate, but it's difficult to see a major shift in policy from this alone. Also, even policy changes will not stoke demand like the pandemic did. I consider this a real issue, as it has been to almost all companies. From now on, Intel is going to have to compete and win against opposition, not just create products and know they will sell because demand is so much greater than supply. But, a quick look at their Alder Lake line shows they have been very good with pricing before they even needed to be. AMD has buckled from the strain, and lowered their prices. They still aren't very competitive, but as the balance of supply/demand continues to move towards greater supply, AMD will get more aggressive.

So, we'll talk more about that competition. Right now, Intel's Alder Lake seems to be faster than AMD's Zen 3. AMD fans may not be happy to hear this, but it's an obvious truth, according to multiple reviews. Here's another obvious truth - 99 44/100's (sorry Ivory) of the people using computers will be fine with AMD processors. Meaning, the difference is pretty much meaningless in real terms. Yes, there is a halo effect, but almost all of AMD's processor line would be more than adequate for almost everyone reading this. In fact, I am writing this on a Pentium Silver J5040 (the old Atom line), which is plenty fast for what I do with it. AMD's modern processors are much faster than this.

So what really matters? Obviously, Intel is better known, so the average consumer will prefer it, all things being equal. Intel is better performing right now, and people will probably hear that from their IT, or the IT will prefer that. But, as mentioned, it really doesn't matter much for most applications. Zen 4 is coming out, but apparently is going to a marginal improvement. AMD is saying a 15%+ improvement in single-threaded performance, which is horrible since TSMC said the process alone gives 15% better performance, without architectural improvements. AMD showed a demo of the processor running at 5.5 GHz, and this is pre-release, so most of this would be coming from clock speed. My opinion is, AMD is sand-bagging it, and we should expect more than 15%, probably closer to 20%. Which still isn't great, since that won't even catch up to Intel's current Alder Lake in single-threaded, but should be close. AMD then said the IPC of the processor is going to be 8-10% better, which makes more sense to me, and does not equate to the horrible 15% single-threaded number. Again, they're sand-bagging the single-threaded number. But, Zen 4 won't be competing with Alder Lake, it will be competing with Raptor Lake, which seems to be over 10% faster than Alder Lake, although information so far is not conclusive.

So, Intel should be able command something a premium, having a greater reputation (particularly to commercial buyers), having better absolute performance, and being better known. Plus, they make their own processors, whereas AMD has to pay TSMC (TSM) to make theirs, and TSMC keeps increasing their pricing. But, here's where it gets a little tricky for Intel as well. Zen 4 is built on TSMC's 5nm technology, which is hardly new, but still far more dense than Intel's '7' technology. So, AMD will be getting better density, on a process that should have very solid yields. Raptor Lake is still on Intel's 7 process, and will certainly have more transistors than Alder Lake, and thus be bigger (since process density remained the same). Right now, Alder Lake does well against Zen 3 in value, but given that Intel's next generation will be bigger and more expensive to make, whereas it's not at all clear this also applies to AMD, this equation could change a bit during the next round, even if performance continues to favor Intel.

Intel's Meteor Lake will be out in 2023, probably late 2023, and will be on Intel's 4 process. Until then, we're soon going to be looking at a situation where Intel has a higher performing architecture, and makes their own chips, but has to overcome the significantly better density available to AMD. I still like Intel's position, since TSMC makes a nice profit off their customers, but I don't like it as much as I do today.

The best part of the next generation, for Intel, is that AMD is moving their processors to a 'shrink', meaning a more dense process node, whereas Intel is not, and Zen 4 will still likely have a little less single-threaded performance than Alder Lake. Raptor Lake should have significantly more. Intel's '4' shrink will be a much more significant upgrade, having 20% frequency improvement at the same power level. Plus, they have many more transistors to improve IPC, just like AMD did with Zen 4. Meteor Lake, which is the leading processor on Intel's 4 process, is already up and running, but so far, we don't have any indication as to the performance. But, given the above two factors, Intel should easily keep the performance title.

AMD competition makes a nice segue to the server situation, which is currently grotesque for Intel. And it was even worse before. Intel released Ice Lake-SP a little over a year ago, and it was a big improvement over Cascade Lake, but still significantly inferior to AMD's server line, as well as many ARM processors in important workloads. It was Intel's move to the first mainstream 10nm process, which was a big improvement in density, but had a very difficult time with higher clock speeds. Although a big improvement, Intel continues to lose market share in servers, and this makes sense given their weak product line there. Whereas some fuss has been made about market share in desktops and mobile, these numbers are largely irrelevant, because the Chromebook market has collapsed, and Apple is no longer an Intel customer. So, year to year comparisons from Intel must be considered (especially Apple revenue) against this context. However, the server market share losses are very real, and unlikely to change any time soon. Although Sapphire Rapids will be a huge improvement for Intel, it will be competing against Zen 4 from AMD, on TSMC's 5nm process. I believe Intel clearly will have the better single-threaded performance, moving not only two generations to AMD's one, but also will be on a much better process which has the highest clock speeds of any current process any company has. AMD, though, will be moving up one architectural change, which based on their own statements isn't a huge one, and moving up one node. It seems like Intel should really do well here, especially since NVIDIA just dumped AMD for Intel's Sapphire Rapids, due to single-threaded performance being superior. And you can be sure NVIDIA has access to near future products from both companies.

But, it's not really so. Yes, they will certainly close the gap, but it's almost equally certain they will not surpass AMD as a broadly better part. Obviously, single-threaded performance is going to be significantly better, given NVIDIA's switch, but servers aren't really about that. It's a factor, but AMD will be moving to a much denser node, Intel will not be. Although Sapphire Rapids (Intel's next server part) will be made on Intel 7, this is just a newer version of the 10nm process Ice Lake SP uses, with huge improvements in performance, but no improvements in density. TSMC's 5nm will give AMD a lot of levers to play with, not only because it will allow more cores, or larger caches, etc... due to the better density, but also because it is more power efficient. And although single-threaded performance is always important, and will belong to Intel, in many cases performance per watt is more important, and will belong to AMD. Plus, multi-threaded performance per socket is important, and AMD will certainly have more cores per socket. I'd expect this fight to be pretty workload dependent, and although more favorable than the current situation, which is dreadful for Intel, still not decidedly in its favor either. In fact, I'd probably guess AMD will do better in more workloads than Intel, just not nearly to the extent it does now.

In the second half of 2023, Intel will release Emerald Rapids, a server chip based on their '4' process. Or so they currently say. Can you trust their server roadmap after so many delays with Sapphire Rapids?

On the plus side, both Intel and AMD should be more compelling compared to ARM server solutions, which haven't advanced much recently.

So, by now you must think I'm crazy, because I'm telling you to buy Intel, but then come up with the reasons why the stock is low, without destroying the legitimacy of these reasons. This is as far as most people can see, and with this visibility, one can value Intel below $40 per share.

But, this is a transitionary period of Intel, and by the time the rest of the world sees it as a temporary situation, the price is going to be much higher. When will they? Who knows. That's why I wouldn't wait, especially with the attractive dividend.

So, what transition am I talking about? Well, some of this relates to the lower margins mentioned above. We'll address that first.

Intel is spending an enormous amount of money on building more fabs, and moving forward with their fabrication processes. Not just for itself, but so it can build chips for other companies. This takes time not only to build the fabs, but to create relationships with companies and get their designs on a process node. Until then, it just eats up tons of cash. Let's not forget, Intel got into trouble because their 10nm node was delayed repeatedly, for years, and so were their architectures built for this process. Now, their 10nm/Intel 7 process hits the fastest on the planet (but is not the most dense, or power efficient), and they are ahead of their schedule on other nodes. As mentioned Meteor Lake (based on Intel 4) is already booting up Windows, and appears to be roughly a quarter ahead. On top of that, Intel has stated their progress on all revealed nodes is on time or ahead of schedule. That's huge. Remember, one of the problems we mentioned above is Intel is stuck on their 7 process, whereas AMD is moving to 5nm. If Intel can catch TSMC, and it appears right now they can, this type of awkward situation will go away. On top of this, having leading process nodes can only help attract customers to their IFS (foundry business). Even NVIDIA (NVDA) expressed strong interest in using Intel as a fab partner. Of course, Apple, Amazon, Microsoft, Google, etc... already make their own processors, and probably will to a greater extent. Intel could be a very strong partner for these companies, even if they move some or all of their workloads from Intel x86 processors. This is extremely important, and although currently a drag on earnings, could be a huge catalyst in the future. Based on Intel's recent remarks, their execution has been extremely good, and so are their prospects here. Also keep in mind, TSMC's market cap is around $500 billion, and has come fairly close to a trillion. This is a huge opportunity for Intel.

Next, as I mentioned earlier, Intel roadmaps are very aggressive, and so far they have been executing extremely well under Pat Gelsinger. He's more than willing to suffer the pain to make it work longer term, something previous CEOs seemed reluctant to. He's also much more flexible, and willing to use outside fabs if needed to make Intel products more competitive. I still think AMD will become more aggressive with pricing in desktops and mobile, and I still think Intel will be in a somewhat weak situation in servers even with the Zen 4/Sapphire Rapid generations, but both are solvable. Intel makes their own chips, so they can initiate a price war quite easily, plus their architecture is higher performing. When Intel moves to their 4 process node, they can compete much more aggressively against other server part makers. And if they can maintain parity or gain leadership in process tech, which is admittedly uncertain but looking better all the time, they can stop or slow their slide in the server market.

One thing is also clear, the appetite for logic and silicon is getting bigger and bigger as it encompasses more aspects of life, and more types of devices. You will have blips, like the post COVID one we have now, but the overall trend is obvious and probably unrelenting. Intel is really the only company that develops their own processors, and can build them. Unless you count the ones Samsung makes, which I don't believe anyone takes very seriously. On top of this, Intel is moving into the GPU market, although badly so far, which over time will increase revenue, and probably help CPU sales as they create synergy between their CPUs and GPUs, like they are currently doing with mobile chips.

So, yes, there are problems, and pretty serious ones that have people taking a 'wait and see' attitude. But the opportunities are enormous if the company executes, and their unique position in the industry makes this even greater. I wish I had a crystal ball and could say when the stock price turnaround begins, but it will, and where it stops could be several times the current price. Is there really much downside at this point, to such a wildly profitable company? Does it even compare to the upside potential? It's not even close. Intel is as sure a thing there is right now. Don't wait.