Taiwanese media: Intel's 3nm chip plan has changed

According to Taiwanese media Digitimes, Intel has reportedly urgently revised its platform blueprint and its own process capacity planning for the next year. Industry rumors have it that CEO Pat Gelsinger will visit Taiwan for the third time in August and meet with TSMC Chairman Liu Deyin and President Wei Zhejia.

Over the past year or so, the market has been skeptical about whether TSMC's 3nm process can be launched as scheduled. However, TSMC's research and development progress is still proceeding step by step. A few days ago, it was officially announced that 3nm will be launched in the second half of the year. The family members are designed based on customer needs in terms of performance, power consumption and cost, and will launch N3, N3E, N3P, N3S and N3X.

As for customers, not only are Apple and Intel expected to be the first to adopt it, but other companies such as NVIDIA, Qualcomm, AMD and MediaTek have also lined up to reserve production capacity.

However, contrary to expectations, the supply chain has recently reported that due to the sudden downturn in the global PC market and the unexpected collapse in demand, the semiconductor and electronics industries are shrouded in dark clouds. The protection of long-term contracts is no longer a talisman. Many companies would rather regret the contract and pay fines in the hope of avoiding this storm of falling prices and volumes.

Generally, small and medium-sized companies can quickly evaluate the pros and cons and make a choice, but the situation is difficult for international IC design giants and foundry companies with annual capital expenditures of tens of billions of dollars. Recently, the market has been rumored that Intel, AMD and NVIDIA, which affect the global PC and server platform landscape, have to revise their shipment and operating targets in the face of a rapid reversal in demand.

Due to the large scale of the supply chain involved, the chain effect of high inventory and low demand has even extended to the upstream wafer foundry. According to semiconductor equipment manufacturers, it has been reported recently that Intel's CEO, who took office in 2021, will visit Taiwan for the third time in August to meet with TSMC executives to discuss the readjustment of process capacity planning.

The new IDM 2.0 strategy previously released by Pat Gelsinger hopes to deepen and expand Intel's partnership with wafer foundries. In addition, Intel pointed out that the modular approach to architecture will drive the next round of evolution, allowing Intel to mix and match individual chips or chip blocks at different process nodes and connect them through Intel's advanced packaging technology. For example, the "Meteor Lake" used for PC client computing will use the "Intel 4" process to manufacture computing chip blocks, and other supporting chip blocks will also be produced by TSMC.

While most of Intel's products will be manufactured internally, chip blocks from external foundries are expected to play a larger role in modular products in the coming years, including core computing functions at advanced nodes to serve emerging workloads in PC clients, data centers and other areas.

However, plans cannot keep up with changes. Recently, it was reported that Intel has internally readjusted its platform blueprint and its own process capacity planning for the next year. The 14th-generation Meteor Lake, which was originally scheduled for mass production at the end of 2022 and released in the first half of 2023, has been delayed to the end of 2023. Since the computing chip block uses Intel 4 process and the graphics chip block is outsourced to TSMC's 3nm foundry, the delayed launch will disrupt TSMC's 3nm plan.

Semiconductor equipment manufacturers said that if Meteor Lake is delayed, the price Intel has to pay will be very high. The contract signed with TSMC is already at the level of a top-tier customer. It was previously rumored that after TSMC confirmed that it would take over Intel's large order, in order to differentiate itself from Apple's 5/3nm production line in Nansho, it would adjust the P8-P9 R&D center of the expanded Fab 12 in Baoshan, Hsinchu Science Park, which was originally planned to be mainly for R&D and mini line, to the second important base for formal 3nm production. In other words, according to the contract, TSMC will produce 3nm graphics chips as scheduled.

However, if Intel's own Intel 4 computing chip blocks are unable to be produced in time due to poor market conditions and internal process technology problems, and TSMC is expected to delay production, Intel may absorb all losses.

However, it is also rumored that Intel was forced to come up with another plan, that is, the 3nm graphics chip block will still be produced as scheduled, and the computing chip block will be re-evaluated and handed over to TSMC for 5nm or even 3nm production. Although this plan loses face, it can allow Intel to take a temporary breather and readjust the pace of process advancement, which will also help reduce costs and improve profits.

For TSMC, the relationship with Intel is both an enemy and a friend. All cooperation is carried out in accordance with the contract, and TSMC will not suffer any loss no matter what decision Intel makes. If TSMC can get more Meteor Lake cooperation orders, it will also be a great help to capacity utilization and operating performance.

On the other hand, TSMC will hold an earnings conference in mid-July, and the outlook for the second half of the year and the next few years will remain unchanged, which has attracted global attention. The current market expects that since orders have reached the end of the third quarter, TSMC will set a record high in the second quarter and continue to rise in the third quarter. With Apple's full sprint, the fourth quarter's performance is still expected to reach a record high, and 2023 will depend on the inventory clearance of terminal consumer electronics.

Neither Intel nor TSMC commented on market rumors.

*Disclaimer: This article is originally written by the author. The content of the article is the author's personal opinion. Semiconductor Industry Observer reprints it only to convey a different point of view. It does not mean that Semiconductor Industry Observer agrees or supports this point of view. If you have any objections, please contact Semiconductor Industry Observer.

Today is the 3096th content shared by "Semiconductor Industry Observer" for you, welcome to follow.