Intel representatives at the quarterly already explained that the company managed to speed up the production cycle for the manufacture of 10-nm products, the level of yield of suitable products is encouraging, all this allows not only to start deliveries of second-generation serial 10-nm processors from the third quarter, but also to deploy their full-scale deliveries by the fourth quarter. In addition, Intel this year will be able to release more 10-nm processors than originally expected.
Intel has to spend money not only on mastering the 10nm process technology and preparing for the transition to the 7nm process technology, but also on expanding the capacity to produce 14nm processors. The latter expense is important to Intel, as incumbent CEO Robert Swan almost vowed that during his tenure as head of the company, customers would never again suffer from product shortages.
Meanwhile, a quarterly form appeared on the Intel website. , which pays slightly more attention to the cost structure than the documents released on Friday. It is this form that allows us to estimate the magnitude of the negative impact of the costs of developing 10-nm technology on the company's profit margin in the first quarter. Compared to the same period last year, Intel's profit margin fell by four percentage points at once.
As the processor manufacturer explains, it had to spend about $530 million last quarter on the release of engineering samples and preparation for mass production of 10-nm products. We emphasize that we are talking not only about processors, since the structure of these costs leaves room for other Intel products manufactured using 10nm technology.
In the client segment, $275 million was spent on related needs. In the server segment, similar expenses amounted to $235 million. The sum of these values ββdoes not add up to $530 million, and about $20 million remains for other divisions. Of the well-known Intel products that should be produced using 10-nm technology, in addition to central processors, one can only recall programmable matrices, highly integrated solutions from the Snow Ridge family for base stations in 5G networks, as well as Nervana neural network accelerators. Obviously, their production volumes are modest enough to meet the $20 million limit. Unfortunately, Intel's documentation of the cost structure in these areas of activity does not reveal to the same extent as for client and server products.
Source: 3dnews.ru