Because even though AMD products were good on paper they had large issues. It wasn't until Zen2 mobile that PPW became compelling, and zen1 server chips were terribly buggy/had poor customer support. Now that they have many of their issues sorted out they can actually move units. That 14nm node also had the benefit of not being that much more complicated a process flow from foundry "14nm". As opposed to intel 7 competing with mature EUV simplified nodes. Heck on their earnings they literally place part of the blame for their margins on the "increased intel 7 product mix". While it seems like some structural issues may be solved, I am still concerned. SPR came out over a year after ADL despite using the same arch, same DDR, same PCIE spec, fabed on the same mature node, and not needing to worry about BIG.little optimizations.
What does that even mean? Is Samsung suddenly a better company than Toyota purely off the back of Toyota "only" makes cars? TSMC plays in the fields it needs for it's business model. Intel plays in the fields it needs for it's business model. In TSMC's case the only way they don't see growth is if the only TAM growth is coming from intel products using 100% intel wafers. For intel the only way they do see growth is if people buy intel products in the growing sectors or if people choose to fab at intel as opposed to TSMC or Samsung.
People have speculated they want to focus all capacity into the segment that matters/has the best margins per mm^2, some think intel isn't EUV ready, some think this is a cannon lake situation, others an icelake situation where the clocks are really low, and the last idea I have seen thrown around was that the larger than 6+8 dies had design issues. Time will tell when we see the TTM, volumes, clock speeds, and skus launched what the issues were.
I agree, but the flood gates have been opened. The competition wasn't sitting still either and they are taking MS. Another important factor is scale. Fabs have a large fixed cost and smaller operating cost. If intel losses MS it hits them on the revenue front. But it also hits them because their per unit cost then goes up.
The loss of market share is the biggest issue. AMD is getting to the point where software will be optimized for both. With that intel's protective moat will be gone. Same deal with the extra money allowing AMD to actually support more server customers and client OEMs. Desktop is a rounding error for both companies. For intel the breadwinner is laptop and historically server (of course now their margins are like 2% so not so much anymore). Ever since the 2000s AMD always has been an HPC company first who would just scale down their architectures for client, so desktop is even more irrelevant for them than it is for intel.
Wrong. That was all GPU shipments (not sales) in dollars vs only AMD's client GPU business in dollars. This "story" had tons of bad reporting.
With regards to the one trick pony remark, I never said one company was better than the other. But, one company has better opportunities than the other. When that company is worth less than 1/4 of the other, the valuations are something to consider. And that is important. The remark about TSMC is obviously wrong, since Intel is not the only fab out there, and companies do switch from TSMC to Samsung (and back
). Also, growth isn't the only thing that matters, how quickly are they growing? Intel will almost certainly grow market share with IFS, if only because they have essentially none. They can also grow by selling their products, and margins there (particularly servers) are quite high. So, it's relevant and important that Intel has more routes to growth, because they may (or may not) have large upsides on their stock price.
Sapphire Rapids was problematic, no argument. So many delays. They still have a ways to go.
You are incorrect about market share, except in servers. Intel has over 80% market share in mobile and desktops. YtY they were up against AMD in mobile, down in desktop. So, no, outside of servers. Servers though, big trouble last year. Let's see if Sapphire Rapids and Emerald Rapids slows or stops the damage.
You completely under represent desktop sales. Not even close to a rounding error. If so, how do you explain the 3D cache processors from AMD? And Intel releasing Raptor Lake on desktop first? Desktop is still 20% market share, that includes tablets, slates, laptops, etc.. Over 78 million desktop processors shipped in 2022, over 70m are expected to ship this year. So, not as big as laptop, by any means, but still pretty substantial.
The GPU statement was correct, because it took into account the correction Jon Peddie made. I did not say Intel had more market share, as their earlier report indicated, but almost as much, which the corrected report does. They are only 3% below, corrected, despite being new and have a limited product portfolio. It's a good start.
