Intel revealed that last summer development of 7nm production process encountered a serious problem that causes at least a 6-month delay, and at that time it was already said that this technology would not be ready until about 2023. This is now a very real result and we know very well that Intel’s first 7nm processors (Meteor Lake, Granite Rapids) to board just that year and that us after 14nm Rocket Lake-S two more 10nm desktop and mobile generations are waiting. But that also means that Alder Lake-S really has to come very soon after Rocket Lake-S for Intel to make it. –
–
Pat Gelsinger has now confirmed in a speech to Intel Unleashed: Engineering the Future that the problem with the 7nm manufacturing process has been resolved and development can continue at full speed. Of course, this will not eliminate the delay, which will mean that Intel will have to compete with AMD, which will have a technological advantage on its side. This is, of course, TSMC’s 5nm process, against which Intel will be able to build its 10nm Enhanced SuperFin (10nm ++) technology at best. On his side, however, there will be production capacities with which the possibilities of AMD cannot be compared.
For the time being, we can only estimate further development, but it can be concluded that in 2023, AMD could already look for production using the 3nm process of the TSMC company, although in the case of processors it may not be across all series. We will probably know this this year, as it is possible that after the Ryzen 5000 5nm processors will not come on the market yet, but another generation of 7nm models (Warhol), while the next EPYC generation Genoa should already be made of 5nm technology and we expect them on the market in next year.
If that happens, it will mean at least partial relief for Intel, whose 7nm process should be at least equal to TSMC’s 5nm process. But it will not be such a simple comparison, because Meteor Lake and logically also the Granite Rapids server will be made up of more chips produced by various technologies and most likely also in the TSMC company.
Pat Gelsinger also stated that he intends to dust off the Tick-Tock strategy again, which meant alternating a new production process and a new or sharply improved processor architecture, with approximately one-year delays. This must have seemed to Intel, because there were problems with the development of its production processes, and also the modified version of Tick-Tock known as Process-Architecture-Optimization, ie extended to three years, did not last long. So we’ll see how Intel’s new boss can handle it.
Prices of related products:
–
