Intel isn’t exactly having a successful season at the moment. The financial results are not looking good, and the company is losing money and has to save at a time when it would need to invest heavily in factories if it is to have a chance to compete again in advanced chip-making technologies, in which it has fallen behind TSMC and Samsung. Confidence in the stock has fallen significantly and there is even talk of selling the company. It seems that this will also be reflected in the products – Intel has now canceled one planned generation of processors.
According to information from Chinese sources, which was shared on the Chiphell forum by a well-known leaker with the nickname Panzerlied, Intel will reduce investment in desktop processors, which will now use the same generation of processors for two years, although the company has always stuck to a one-year cycle of releasing new generations .
Intel is now preparing to release a new generation of Arrow Lake desktop processors, which will bring a new architecture and a new LGA 1851 platform. Fortunately, this is unchanged, and the first part of the processors (125W enthusiast models) are due to be released next month. But Intel canceled their successor on the LGA 1851 platform, the so-called Arrow Lake refresh generation, which was supposed to be a product released in the second half of 2025 and again serve the desktop market for roughly the next 12 months. This means that the now released Arrow Lake processors will remain Intel’s main offering for two years.
This is probably part of the austerity measures that Intel took after poor financial results. The preparation of the refreshed models probably needed certain investments, and Intel decided to save them and rather accept the possibility that without the release of a new refreshed generation, it will lose some sales and possibly market share.
Gradual reduction of post-Arrow Lake desktop products
This isn’t the first strike in the desktop roadmap. Originally, Arrow Lake processors were supposed to be followed by Panther Lake processors, which would have a new production process (Intel 18A) and quite possibly an improved architecture. But just like Intel failed or it was not worth it to get the new generation of Meteor Lake on the 4nm process into the desktop and instead released the Raptor Lake processors and their refresh, the decision was made some time ago to release Panther Lake only in notebooks, and in the desktop instead it should have been released a refresh of Arrow Lake processors, which, as it turned out, uses TSMC’s 3nm process.
But this refresh also went through several stages of cutting. A long time ago, it was planned that it would get more cores, a new Compute chiplet should have been designed for it, in which there would be 8 large Lion Cove cores (so no change), but instead of 16 E-Core cores with the Skymont architecture, 32 of these cores, so it would be 40 cores with 40 threads. This would have the potential to massively lift multi-threaded performance, though probably not without degrading power consumption.
However, these larger chiplets were canceled some time ago and Intel decided that the refreshed version of Arrow Lake will also remain with only 8+16 cores, i.e. no change in the number of cores compared to the original Arrow Lake. Apparently, the frequencies should have been increased, but in addition, a more powerful unit for accelerating artificial intelligence applications NPU should have been installed in the processor (perhaps on its own chiplet, or as part of the redesign of one of the chiplets). This would have performance as high as Lunar Lake, and the Arrow Lake Refresh processors could therefore go into the so-called Copilot+ PC with artificial intelligence capabilities. The original generation of Arrow Lake processors has a less powerful NPU with a performance of only 13 TOPS, while Copilot+ PCs require over 40 TOPS.
This was probably supposed to be the main differentiating element of that Arrow Lake refresh generation after the CPU core count boost was abandoned. However, when Intel cut the budget, it was precisely the funds that were supposed to go into designing or integrating the new NPU that led to the cancellation of the entire refresh generation.
The generation of processors that was supposed to have the name Core Ultra 300 in the desktop was basically degraded or canceled three times, and in the end it may not be anything at all. It’s possible that after being released now in autumn and in January 2025, the Core Ultra 200 desktop processors will be sold unchanged for two full years until autumn 2026, when they will be replaced by a new generation of processors, which should have a new architecture and the Nova Lake designation – perhaps it should be manufactured by TSMC’s 2nm process (which is interesting because it’s like Intel admitting that its 1.8nm 18A process is inferior to TSMC’s 2nm). It is possible that only enthusiast models will be released again in the fall of 2026, and the mainstream nominally 65W models (as usual, but we must point out that their real consumption can be considerably higher) would not be released again until January 2027.
While the comments on this report say that the cancellation of the Arrow Lake Refresh generation may mean that Intel will rush the release of this new generation Nova Lake, this may be overly optimistic. Lately, Intel has been unable to get new processors ready ahead of time and has been struggling with delays. Therefore, skepticism is probably in order and we would not bet on this.
What could theoretically happen is that the cancellation of the Arrow Lake refresh version will not affect notebook processors, for which Intel could theoretically still produce an updated version with a more powerful NPU and compatibility with the Copilot+ PC program. But there is no information about that yet. Nowadays, the notebook market is significantly more important than the desktop market, and Copilot+ PC support will probably be more important to it, so such a move would make sense.
Resources: Chiphell, VideoCardz
Source: www.cnews.cz
