$Intel(INTC.US) is really despicable. I read the article and found its credibility to be low. For example, the article claims that 18A uses high EUV, which is wrong—14A was the first to use it. The article also states that 18A's efficiency is worse than N3, but the Geekbench single-core/multi-core results are worse than Lunar Lake's. That's pure nonsense—Lunar Lake is just a low-power optimized version. If Lunar Lake is worse than Intel, then Intel might as well go bankrupt. To measure process efficiency, you should look at the single-core curve; multi-core is meaningless because dynamic power consumption is related to frequency cubed (CPUs can be simplified as large capacitors charging and discharging). So even with the same process and architecture, different core counts will result in completely different multi-core power curves. Next month, Panther Lake will be released. Testing its real single-core efficiency curve against Arrow Lake's will reveal the true performance of 18A compared to N3B.