It is not news if it is said that the Core 14 will be a giant leap for Intel due to the use of EUV technology for the first time in its history. In this aspect it is far behind TSMC or Samsung, so to pull itself together and with a very clear message to these rivals, we now know that the blue giant has spent no less than 1,000 million on ASML EUV scanners to manufacture their Core 14.
What has been leaked is a sample of Intel’s seriousness and the fact that it is striving to catch up on such important technical details as EUV technology. It’s true that Samsung and TSMC have acquired all the scanners so far, so Intel has jumped in to buy versions EUV and EUV High-NA.
Between 10 and 12 EUV scanners to manufacture the Core 14: a cost of 1,000 million
It is said soon, but without counting the raw materials, the acids, the various liquids, the labor and of course the wafers, to speak of an initial investment of 1,000 million for the next two generations of CPUs Intel is saying a lot. The problem is precisely the useful life of these, since Intel has switched (this time yes) to TSMC and Samsung for the next generation race has bought ALL the scanners EUV High NAbut that is another story attached to the one we are telling today.
Only 55 units of the latest scanners were made ASML EXE in 2021, so the 10 or 12 that Intel bought in 2022 seem to represent much, if not all, of the production the Dutch company has so far this year. Why buy these scanners now when you’ve already pre-ordered the next generation? Well, by simple steps of the nodes and above all, because the production of these new High-NAs is extremely rare and above all very expensive (+-400 million per scanner according to rumours).
Thus, Intel, which until now did not use EUV and competed with its competitors, will make a qualitative leap with its Intel 4 process for Core 14.
Oregon first, then Ireland, wafers everywhere for Core 14
It may seem 10 or 12 scanners they are few (the number is not very clear) but the reality is that the performance of the latest EXE models is really interesting, not spectacular, but they guarantee a production that, if done over time, means a stock more than covered for Core 14 and possibly part of Core 15.
The service is stipulated between 160 and 220 WPHwhere the mean seems to be stipulated in 200W/hwhich translates to 4,800 platelets per day (the scanners never stop, they work 24/7, except for their maintenance). Logically, this would give us an annual volume of 1.75 million platelets per scanner, where depending on the chip they register there will be more or less.
It is estimated that in the case of larger surface matrices, they could reach 60 or 65 chips per wafer, in the case of the smallest nearly 200 per unit. You don’t have to do the math to figure out that global production would be covered as soon as Intel has enough units produced before the presentation, which it always does.
Therefore, with 10 scanners, you would already have enough, which would be split between your FAB of Oregon first, to then leave the remaining units of Holland at their FAB of Ireland. In any case, the first processors 14 to 4 nm nucleus (Lake of Meteora) they will leave the United States then Europe towards the world.