Intel process transformation is difficult to decide. After the 18A session was not as good as expected in terms of yield and customer expansion, the focus turned to the next generation of 14A processes. However, while this stage promises higher effi...
Intel process transformation is difficult to decide. After the 18A session was not as good as expected in terms of yield and customer expansion, the focus turned to the next generation of 14A processes. However, while this stage promises higher efficiency and energy efficiency, the high cost has put Intel in a difficult situation.
Intel financial manager David Zinsner admitted that the cost of the 14A crystal will be significantly higher than that of the 18A. The main reason is that ASML High-NA EUV is fully introduced in 14A.
Equipment is expensive: High-NA EUV is about US$380 million, 60% higher than Low-NA. Design complex: The exposure scene is reduced by half, and the process and R&D costs are simultaneously increased. Round high price: 14A per Round is estimated at about 30,000 to 35,000 US dollars, higher than the current points.Although Intel promises that the 14A will have an increase in performance-to-power consumption ratio of 15%~20%, and a decrease in energy consumption of 25%~35%, the yield and customers are still uncertain, and the high cost is difficult to explain the market.
The competition strategy is more practical. NTD has admitted that the A14 session "is not necessarily" to use High-NA EUV, while the A16 and N2 are still mainly Low-NA EUV, and only a small amount of introduction is considered when the design needs to be designed to control costs. Currently, the 2-nanometer (N2) crystal round is about 30,000 US dollars. Although it is not low, it is relatively high in market acceptance through mature mass production rules and stable large customers (such as Apple and NVIDIA), and the yield is even close to 90%, indicating a smooth progress.
Samsung has taken cautious steps and has introduced the first High-NA EUV machine, but it has not yet determined whether it will be used for the latest production sessions. Currently, Samsung's main customer is Tesla. Although its scale is not as large as Taiwan Power, its customers' abilities are relatively clear.
Intel could only bet one shot. If 14A lacks external customers and the R&D and investment of the University is difficult to obtain reasonable returns, the company may have to ease or even give up the spot. At the same time, Intel's agreement with the US government requires that at least 51% of the foundry business be held within five years, otherwise the agreement will make Intel's strategic choices more limited.
The market generally maintains a retention attitude about whether Intel can suddenly increase. 14A may represent Intel's technical ambitions, but under the high cost and lack of triple pressure from customers and competitors, Intel will still be full of uncertainty in the future.
Intel CFO confirms that 14A will be more expensive than 18A due to High-NA EUV tool — Intel expects 14A process to offer 15-20% better performance-per-watt or 25-35% lower power consumption than 18A
