This article was reprinted with permission from Superpower.com.
Intel today announced that it is investing US$5 billion to upgrade its fab in Israel. This means that the 10nm process has accelerated, but Intel’s investment plan will take two years, and the current 10nm process yield problem has not yet been completely resolved.
We already know that 14nm will also be the main force of Intel processors. Now the question is how long will the 14nm still last?
Intel’s vice president mentioned in the meeting today that in the next 12-18 months, the 14nm process will continue to improve, and its performance has been improved by 70%.
Murthy Renduchintala, Intel's senior vice president, recently participated in JP Morgan's 46th Global Technology, Media and Communications Conference. At this meeting, analysts consulted a lot about Intel's extension of the 10nm process.
He first explained why the Inte l10nm process is so difficult.
As early as 2014, when Intel defined the 10nm process for the first time, they came up with very demanding goals. Compared with the 14nm process, the conversion factor of the 10nm process was 2.7 times. (Scaling factor, an important indicator to measure the progress of the process is how many times to reduce the previous generation).
In the 22nm process upgrade to 14nm, the conversion factor of the new process is only 2.4 times , so Intel's team has a very ambitious ambition on the 10nm node.
To achieve this goal requires a lot of innovation, so Intel has faced more challenges in this process, Renduchintala stressed that they encountered The problem is mainly on yield, not fundamental. Intel knows how to solve the problem and is currently focusing on solving these problems.
At the same time, they also discovered that there is still potential for 14nm processing. Renduchintala said that From the first generation of 14nm process to the latest generation of 14nm process, its performance has been improved by 70%, which also gives Intel more room to solve the 10nm process yield problem.
As for the future, Renduchintala stressed that they will continue to work hard to solve the 10nm process yield problems, but also satisfied with the 14nm processor roadmap, will let them continue to lead the next 12-18 months.
