Intel was the most advanced semiconductor brother in the chip process on the earth, in the process of leading friends for three and a half years without pressure, Samsung, TSMC was completely incomparable, but after the 14nm node, TSMC, Samsung rose rapidly, Intel's 14nm process lasted for more than 6 years, last year was replaced by the 10nm process.
Last year, Intel replaced its new CEO, Henry Kissinger, who has been at Intel for more than 30 years, and one of the goals after re-emerging is to revive Intel's leadership in the semiconductor market and surpass TSMC in the next few years.
To this end, Kissinger launched the IDM 2.0 strategy, in the next few years to invest hundreds of billions of dollars in the United States and Europe to build factories, not only for their own use, but also to enter the wafer foundry market, before 2025 to master at least 5 generations of CPU processes, and take the lead in mass production of Amy-level 20A and 18A processes.
So the question is, assuming that Intel's progress in the next few years is not a problem, can it surpass TSMC in 2025? The semiwiki website really analyzed and calculated this problem.
Measuring the chip process is a complex problem, but their calculations mainly focus on process performance and density, in 2025 Intel should be mass production of 20A and improved version of the 18A process, Samsung is also a 2nm process, TSMC although not clear, but then the main force should also be 2nm process, the three will be on the new GAA transistor structure.
The final result is that in 2025, Intel's 20A and 18A processes will have advantages in performance, higher than the 2nm-level processes of TSMC and Samsung.
However, while the performance returns to the first, Intel's process is still unlikely to have the upper hand in the transistor density at the same time, and TSMC's 2nm will still lead in this regard.
Of course, the above results are still a reference, because the manufacturer's published data is not comprehensive enough, semiwiki is a senior expert can not be completely accurate, but in 2025 Intel in the 2nm node king return should be no problem, at least the process performance advantage can be guaranteed.