Intel officially confirmed on February 15 that it will acquire Israeli foundry Tower Semiconductor for nearly US$6 billion, and the deal will likely contribute to the growth of Intel’s foundry business if it reaches a successful conclusion, according to TrendForce’s latest investigations. Tower was 9th place in the global ranking of foundries by revenue for 4Q21 and operates a total of seven production sites across Israel, the US, and Japan. Tower’s foundry capacity in 12-inch wafer equivalents accounts for about 3% of the global total. The majority share of Tower’s foundry capacity is for 8-inch wafers, and Tower’s share of the global 8-inch wafer foundry capacity is around 6.2%. Regarding manufacturing process platforms, Tower offers nodes ranging from 0.8µm to 65nm. It has a diverse range of specialty process technologies for manufacturing products in relatively small quantities. Products that Tower has been contracted to manufacture are mostly RF-SOI components, PMICs, CMOS sensors, discretes, etc. As such, the Tower acquisition is expected to help Intel expand its presence in the smartphone, industrial equipment, and automotive electronics markets.

Although Intel undertook a series of business strategies to compete with TSMC and Samsung, IFS (Intel Foundry Services) has historically manufactured with platform technologies for processors such as CPUs and GPUs. Furthermore, competition still persists between Intel and certain foundry clients that require advanced processes below the 10nm node, such as AMD and Nvidia, which have long histories of developing server products, PC CPUs, GPUs, or other HPC-related chips. Intel’s preexisting competitive relationship with these companies may become a barrier to IFS’ future expansion because IFS will be relatively unlikely to attract them as customers.

Taking the aforementioned factors into account, TrendForce believes that the Tower acquisition will likely expand IFS’ business presence in the foundry industry through two considerations. First of all, the acquisition will help Intel both diversify its mature process technologies and expand its clientele. Thanks to advancements in communication technologies and an increase in demand for new energy vehicles, there has been a recent surge in demand for RF-SOI components and PMICs. Tower’s long-term focus on the diverse mature process technologies used to manufacture these products means it also possesses a long-term collaborative relationship with clients in such markets. By acquiring Tower, Intel is therefore able to address IFS’ limited foundry capabilities and limited clientele. The second consideration pertains to the indigenization of semiconductor manufacturing and supply allocations, which have become increasingly important issues in light of current geopolitical situations. As Tower operates fabs in Asia, EMEA, and North America, the acquisition is in line with Intel’s current strategic aim to reduce the disproportionate concentration of the foundry industry’s supply chain in Asia. As well, Intel holds long-term investments and operates fabs in both the US and Israel, so the Tower acquisition will give Intel more flexibility in allocating production capacities, thereby further mitigating risks of potential supply chain disruptions arising from geopolitical conflicts.

In addition to the aforementioned synergy derived from acquiring Tower, it should also be pointed out that Intel is set to welcome an upcoming partnership with Nuvoton. Tower’s three Japan-based fabs were previously operated under TowerJazz Panasonic Semiconductor, a joint venture created by Tower and Panasonic in 2014, with Tower and Panasonic each possessing 51% and 49% ownership, respectively. After Nuvoton acquired PSCS (Panasonic Semiconductor Solutions Co.) in 2020, Panasonic’s 49% ownership of the three fabs was subsequently transferred to Nuvoton. Following Intel’s Tower acquisition, Intel will now possess the 51% majority ownership of the fabs and jointly operate their production lines for industrial MCUs, automotive MCUs, and PMICs along with Nuvoton. Notably, these production lines also span the range of CIS, MCU, and MOSFET technologies previously developed by Panasonic.

For more information on reports and market data from TrendForce’s Department of Semiconductor Research, please click here, or email Ms. Latte Chung from the Sales Department at lattechung@trendforce.com

Intel has long dominated the x86 architecture based server and PC processor market through the IDM model. At the same time, it acts as a pioneer in the semiconductor industry’s process miniaturization according to Moore’s Law. Yet, in recent years, Intel has seen continued delays in the development of 10nm and 7nm technologies. At the same time, in the ARM architecture based SoC processor market, customers can continuously and steadily obtain higher performance, lower power consumption, and more cost efficient IC design and manufacturing services through the continuous technological breakthroughs of TSMC at 10/7/5nm or even 3nm, integrated with the TSMC-led Open Innovation Platform (OIP), process and design-technology co-optimization (DTCO), and 3DFabric advanced packaging services. In addition to Apple leading the world in releasing the most advanced AP-SoC mobile processors, AMD’s PC processor market share on the client side is gradually threatening Intel. At the same time, the supply stability of the AMD Graphic and Data Center also trumps Intel and Nvidia. Furthermore, Apple’s self-developed M1/M1 pro/M1 max processors built with TSMC’s 5nm technology have been a reason for Intel’s lost Macbook series orders in the past two years which, in turn, has encouraged more brand-named manufacturers to initiate Fabless development strategies.  Companies such as Microsoft, Amazon, Google, Facebook, and Alibaba have all put forward self-developed ARM architecture solutions.

In 2020, Intel continuously spoke publicly stating that the company’s long-term core development strategy is gradually shifting from the old CPU processor business to xPU data computing services and, after revealing plans to outsource a portion of their CPU business, discussed plans to partner with TSMC. According to TrendForce’s investigations, Intel’s earlier non-CPU products such as FPGA, ASIC, RFIC, PMIC and Wi-Fi have already been outsourced to TSMC, UMC or Samsung.  As of today, Intel has officially released orders for CPU products to TSMC. Discounting cooperation in existing product lines, the division of labor between Fabless and Foundry combined with TSMC-led OIP, DTCO and 3D Fabric services will provide Intel with a multitude of choices. In addition to maintaining their original IDM model, Intel can maintain a high-margin self-developed production line and appropriate capital investment while flexibly using TSMC’s production line to create additional diversified value and maintain a competitive advantage against competitors such as AMD.

（Image credit: Google）

Demand for NAND Flash products will undergo a noticeable and cyclical downward correction in 1Q22 as major smartphone brands wind down their procurement activities for the peak season and ODMs prepare for the New Year holidays, according to TrendForce’s latest investigations. As such, the NAND Flash market will remain in an oversupply situation, with prices continuing to undergo downward corrections accordingly. However, PC OEMs have been reinstating certain orders for client SSDs since early November in response to improvements in the supply of upstream semiconductor materials. By fulfilling these orders, suppliers are able to keep their inventory level relatively low, meaning they are not under as much pressure as previously expected to reduce inventory by lowering prices. Taking these factors into account, TrendForce expects NAND Flash ASP to undergo a 10-15% QoQ decline in 1Q22, during which NAND Flash prices will experience the most noticeable declines compared to the other quarters in 2022.

Regarding the price trend of NAND Flash products across the whole 2021, TrendForce further indicates that suppliers have actively transitioned their output to higher-layer technologies, resulting in a bit supply growth that noticeably outpaces demand, though the tight supply of components such as controller ICs and PMICs has constrained the production of NAND Flash end-products. Hence, the decline in contract prices of NAND Flash products has not been as severe as previously expected. Moving ahead to 2022, however, the supply of relevant components is expected to gradually improve, so the market for various NAND Flash products will also likely shift towards a noticeable oversupply. As a result, prices of NAND Flash products will steadily decline before the arrival of the peak season in 3Q22.

Client SSD prices will maintain a downward trajectory in 1Q22, by about 5-10% QoQ

While PC OEMs aggressively push out shipments in 4Q21, demand also remains strong for commercial notebooks, in turn propelling the overall volume of notebook production for 4Q21 to 3Q21 levels, surpassing prior expectations. Moving into 1Q22, however, notebook demand from the consumer segment and education segment is expected to moderate, and client SSD buyers’ procurement activities for the quarter will therefore become more conservative. Suppliers, on the other hand, continue to shift the bulk of their client SSD output to 128L and higher layer products as they release the next generation of client SSDs to both capture market share and increase the consumption of these higher-layer products. The average storage capacity of client SSDs will expand to 567GB next year, with WD releasing QLC products alongside existing QLC manufacturers Intel and Micron, in turn intensifying suppliers’ pricing competition. TrendForce therefore expects contract prices of client SSDs to maintain their existing downward trajectory and undergo a 5-10% QoQ decline in 1Q22.

Prices will decrease by about 3-8% QoQ for PCIe enterprise SSDs but hold flat for SATA enterprise SSDs

North American hyperscalers saw their inventory levels rising throughout the fourth quarter as their production capacities for servers were negatively affected by component gaps. In addition, some of these issues are expected to persist in 1Q22, so server shipment for the 4Q21-1Q22 period will experience continued declines, thus putting downward pressure on the growth of enterprise SSD bit demand. As for the supply side, not only has the issue of insufficient PMIC production capacity become gradually alleviated, but hyperscalers have also cut down on their enterprise SSD orders somewhat due to their focus on inventory reduction. Hence, the production capacities for enterprise SSDs with PCIe interface have slowly returned to normal, and room for price negotiations with suppliers is also beginning to surface. Regarding enterprise SSDs with SATA interface, their supply has become relatively tight because manufacturers prioritize the production of high-density PCIe SSDs over SATA SSDs, which feature an older interface and lower density. As such, contract prices of SATA SSDs are unlikely to drop. For 1Q22, TrendForce forecasts an overall 3-8% QoQ decline in enterprise SSD prices, with contract prices of SATA enterprise SSDs mostly holding flat and prices of PCIe products declining by 3-8% QoQ.

eMMC prices will decrease by 5-10% QoQ

TVs, Chromebooks, and other categories of consumer products that carry eMMC solutions have been experiencing sluggish demand in the second half of this year as related subsidies and tenders in the US wind down. The seasonal fluctuations of the demand for consumer products will return to the pre-pandemic pattern next year. Chromebook production is forecasted to show a small rebound in 1Q22 and climb to the year’s peak in 2Q22 in accordance with the traditional seasonal pattern. Even so, the annual total Chromebook production for 2022 will still register a significant decline from the previous year. Turning to TV production, a QoQ decline is projected for 1Q22. Taking account of these demand-related projections, TrendForce expects the demand for eMMC solutions to be fairly weak in 1Q22. The overall production capacity for low-density 2D NAND Flash products has remained relatively constant. Some suppliers continue to scale back 2D NAND Flash production capacity, but they have slowed down the pace of reduction. Regarding the price trend of eMMC solutions, it is now adjusting downward to a stable level after the surge in 2Q21. TrendForce forecasts that contract prices of eMMC solutions will drop again by 5-10% QoQ for 1Q22.

UFS prices will decrease by 8-13% QoQ due to rising supply and falling demand

Component gaps in the upstream sections of the supply chain are still a serious issue affecting smartphone brands’ device production. Despite the contribution from the traditional peak shipment season in the second half of the year, the YoY growth rate of the total smartphone production in 2021 is expected to once again fall short of earlier projections. Looking ahead to 1Q22, Apple is expected to scale back its smartphone-related demand due to seasonality. This, in turn, will negatively affect NAND Flash suppliers’ bit shipments and further weaken mobile storage demand as a whole. The latest examination of product shipments from NAND Flash suppliers indicates that 1XX-L technologies are now mainstream, and 1YY-L technologies will gradually be adopted during 1H22. Micron has skipped 128L in its stacking technology migration and thereby advanced from 96L directly to 176L. To lower production cost and raise bit output, suppliers continue to increase the layer number of their 3D NAND technologies. This means that supply growth will further outstrip demand growth in 1Q22 as the off-season sets in. Hence, TrendForce forecasts that prices of UFS solutions will also register steeper QoQ declines of 8-13% for 1Q22.

NAND Flash wafer prices will decrease by 10-15% QoQ as oversupply becomes more severe

Sales of retail storage products such as UFDs and memory cards have been weak through this entire year. The promotional activities initiated by e-commerce companies for the special events and festivals near the end of the year have generated only a marginal amount of demand. Looking ahead to the early part of 2022, the demand for retail storage products is not expected to gain noticeable momentum before the arrival of Lunar New Year holiday. Additionally, the cryptocurrency market has been energetic, so the demand for graphics cards from cryptocurrency miners has been outpacing supply for the most part. This development has been impacting shipments of DIY PCs and thereby suppressing the demand for retail client SSDs during 2021. In sum, the aforementioned factors have significantly impeded the consumption of NAND Flash wafers. In view of the demand situation in the different application segments, NAND Flash suppliers will likely ramp up wafer shipments to prevent excess inventory. Moving into 1Q22, even if there is sustained demand for storage components in the PC and server segments, smartphone-related demand will shrink further and exacerbate the oversupply situation of the NAND Flash wafer market. TrendForce forecasts that contract prices of 3D NAND Flash wafers will fall by 10-15% QoQ. Among the various NAND Flash products, 3D NAND Flash wafers will suffer the sharpest price drop. It is worth noting that the growing gap between supply and demand is already exerting considerable pressure on some suppliers, so there is a possibility that suppliers could begin dumping products earlier than expected at the end of this year. Such development could help moderate the magnitude of the price downtrend in 1Q22.

For more information on reports and market data from TrendForce’s Department of Semiconductor Research, please click here, or email Ms. Latte Chung from the Sales Department at lattechung@trendforce.com

The growth of the NAND Flash market in 3Q21 was primarily driven by strong demand from the data center and smartphone industries, according to TrendForce’s latest investigations. More specifically, NAND Flash suppliers’ hyperscaler and enterprise clients kept up their procurement activities that began in 2Q21 in order to deploy products based on new processor platforms. Major smartphone brands, on the other hand, likewise expanded their NAND Flash procurement activities during the quarter as they prepared to release their new flagship models. As such, clients in both server and smartphone industries made significant contributions to the revenue growth of the NAND Flash industry for 3Q21. At the same time, however, suppliers also warned that orders from PC OEMs began showing signs of decline. On the whole, the industry’s quarterly total NAND Flash bit shipment increased by nearly 11% QoQ for 3Q21, and the overall NAND Flash ASP rose by nearly 4% QoQ for the same quarter. Thanks to rising prices and expanding shipments, the quarterly total NAND Flash revenue increased by 15% QoQ to a new record high of US$18.8 billion in 3Q21.

Moving into 4Q21, the impact of the ongoing component gaps has widened to numerous application segments of the NAND Flash market as the capacity crunch in the foundry market remains unresolved. Currently, NAND Flash components are in abundance relative to other kinds of key components. For OEMs and ODMs, the differences between the NAND Flash inventory level and the inventory levels of other components have been growing over the past several months. Therefore, they have to scale back orders and reduce stock for NAND Flash. As inventory adjustments are happening, NAND Flash contract prices will start to drop and thus bring about an end to the several quarters of strong revenue growth enjoyed by suppliers.

Samsung

Owing to procurement demand from hyperscalers and smartphone brands, the NAND Flash market generally remained in shortage in 3Q21, thereby driving up Samsung’s ASP by 10% QoQ. Even so, Samsung’s NAND Flash bit shipment increased by only about 5% QoQ due to weakening demand from PC OEMs and low inventory levels of certain other components carried by Samsung’s clients. Samsung’s NAND Flash revenue for 3Q21 reached US$6.51 billion, a 16.5% QoQ increase.

Kioxia

Although orders from PC OEMs began to wane, Kioxia still benefitted from orders from its major smartphone and data center clients in 3Q21, during which Kioxia’s NAND Flash bit shipment underwent a major QoQ increase exceeding 15%. As the NAND Flash market remained in a shortage situation, Kioxia’s ASP increased by about 4% QoQ, resulting in a revenue of US$3.64 billion, which represents a 20.8% QoQ increase and the highest single-quarter revenue in Kioxia’s history.

SK hynix

Among all NAND Flash suppliers in 3Q21, SK hynix registered the highest growth in bit shipment at more than 20% QoQ. This performance can be attributed to several reasons: the cyclical upturn in procurement activities from smartphone brands, persistently strong demand from the data center segment, and inventory-clearing by SK hynix in anticipation of weak demand in the upcoming off-season. Thanks to an ASP increase of about 5% QoQ, SK hynix’s NAND Flash revenue for 3Q21 reached US$2.54 billion, a 25.6% QoQ increase.

Western Digital

Although Western Digital’s PC OEM clients reduced their SSD orders due to supply chain disruptions, and demand from the retail end also remained weak, Western Digital was able to increase its NAND Flash bit shipment by 8% QoQ in 3Q21 due to enterprise SSD demand from the data center segment and NAND Flash demand from smartphone brands for the release of new smartphone models. Nevertheless, Western Digital’s ASP fell by 3% QoQ because the company increasingly focused on major clients and high-density products. Western Digital’s NAND Flash revenue for 3Q21 reached US$2.49 billion, a 2.9% QoQ increase.

Micron

Demand from the data center segment remained strong, and clients continued to adopt Micron’s 176L products. However, Micron’s shipment share in the smartphone market lagged behind that of other NAND Flash suppliers. Furthermore, its PC OEM clients were starting to be affected by the uneven supply of semiconductor chips. In light of these factors, Micron’s NAND Flash bit shipment increased by a modest 4% QoQ. On the other hand, the NAND Flash market remained in a severe shortage in 3Q21, thereby driving up Micron’s ASP by about 5% QoQ. Hence, Micron’s NAND Flash revenue for 3Q21 reached US$1.97 billion, an 8.8% QoQ increase.

Intel

Although persistently strong demand from the data center segment led to a massive price hike for enterprise SSDs and a nearly 6% increase in Intel’s ASP in 3Q21, the company was unable to fully meet its client demand since it could not procure sufficient upstream components. This lack of upstream components resulted in a severe decline of about 5% QoQ in Intel’s NAND Flash bit shipments and offset the upward momentum generated by an increase in Intel’s ASP. Intel’s NAND Flash revenue for 3Q21 reached a mere US$1.11 billion, a slight 0.6% QoQ increase.

For more information on reports and market data from TrendForce’s Department of Semiconductor Research, please click here, or email Ms. Latte Chung from the Sales Department at lattechung@trendforce.com

According to TrendForce’s latest report on the server industry, not only have emerging applications in recent years accelerated the pace of AI and HPC development, but the complexity of models built from machine learning applications and inferences that involve increasingly sophisticated calculations has also undergone a corresponding growth as well, resulting in more data to be processed. While users are confronted with an ever-growing volume of data along with constraints placed by existing hardware, they must make tradeoffs among performance, memory capacity, latency, and cost. HBM (High Bandwidth Memory) and CXL (Compute Express Link) have thus emerged in response to the aforementioned conundrum. In terms of functionality, HBM is a new type of DRAM that addresses more diverse and complex computational needs via its high I/O speeds, whereas CXL is an interconnect standard that allows different processors, or xPUs, to more easily share the same memory resources.

HBM breaks through bandwidth limitations of traditional DRAM solutions through vertical stacking of DRAM dies

Memory suppliers developed HBM in order to be free from the previous bandwidth constraints posed by traditional memory solutions. Regarding memory architecture, HBM consists of a base logic die with DRAM dies vertically stacked on top of the logic die. The 3D-stacked DRAM dies are interconnected with TSV and microbumps, thereby enabling HBM’s high-bandwidth design. The mainstream HBM memory stacks involve four or eight DRAM die layers, which are referred to as “4-hi” or “8-hi”, respectively. Notably, the latest HBM product currently in mass production is HBM2e. This generation of HBM contains four or eight layers of 16Gb DRAM dies, resulting in a memory capacity of 8GB or 16GB per single HBM stack, respectively, with a bandwidth of 410-460GB/s. Samples of the next generation of HBM products, named HBM3, have already been submitted to relevant organizations for validation, and these products will likely enter mass production in 2022.

TrendForce’s investigations indicate that HBM comprises less than 1% of total DRAM bit demand for 2021 primarily because of two reasons. First, the vast majority of consumer applications have yet to adopt HBM due to cost considerations. Second, the server industry allocates less than 1% of its hardware to AI applications; more specifically, servers that are equipped with AI accelerators account for less than 1% of all servers currently in use, not to mention the fact that most AI accelerators still use GDDR5(x) and GDDR6 memories, as opposed to HBM, to support their data processing needs.

Although HBM currently remains in the developmental phase, as applications become increasingly reliant on AI usage (more precise AI needs to be supported by more complex models), computing hardware will then require the integration of HBM to operate these applications effectively. In particular, FPGA and ASIC represent the two hardware categories that are most closely related to AI development, with Intel’s Stratix and Agilex-M as well as Xilinx’s Versal HBM being examples of FPGA with onboard HBM. Regarding ASIC, on the other hand, most CSPs are gradually adopting their own self-designed ASICs, such Google’s TPU, Tencent’s Enflame DTU, and Baidu’s Kunlun – all of which are equipped with HBM – for AI deployments. In addition, Intel will also release a high-end version of its Sapphire Rapids server CPU equipped with HBM by the end of 2022. Taking these developments into account, TrendForce believes that an increasing number of HBM applications will emerge going forward due to HBM’s critical role in overcoming hardware-related bottlenecks in AI development.

A new memory standard born out of demand from high-speed computing, CXL will be more effective in integrating resources of whole system

Evolved from PCIe Gen5, CXL is a memory standard that provides high-speed and low-latency interconnections between the CPU and other accelerators such as the GPU and FPGA. It enables memory virtualization so that different devices can share the same memory pool, thereby raising the performance of a whole computer system while reducing its cost. Hence, CXL can effectively deal with the heavy workloads related to AI and HPC applications.

CXL is just one of several interconnection technologies that feature memory sharing. Other examples that are also in the market include NVLink from NVIDIA and Gen-Z from AMD and Xilinx. Their existence is an indication that the major ICT vendors are increasingly attentive to the integration of various resources within a computer system. TrendForce currently believes that CXL will come out on top in the competition mainly because it is introduced and promoted by Intel, which has an enormous advantage with respect to the market share for CPUs. With Intel’s support in the area of processors, CXL advocates and hardware providers that back the standard will be effective in organizing themselves into a supply chain for the related solutions. The major ICT companies that have in turn joined the CXL Consortium include AMD, ARM, NVIDIA, Google, Microsoft, Facebook (Meta), Alibaba, and Dell. All in all, CXL appears to be the most favored among memory protocols.

The consolidation of memory resources among the CPU and other devices can reduce communication latency and boost the computing performance needed for AI and HPC applications. For this reason, Intel will provide CXL support for its next-generation server CPU Sapphire Rapids. Likewise, memory suppliers have also incorporated CXL support into their respective product roadmaps. Samsung has announced that it will be launching CXL-supported DDR5 DRAM modules that will further expand server memory capacity so as to meet the enormous resource demand of AI computing. There is also a chance that CXL support will be extended to NAND Flash solutions in the future, thus benefiting the development of both types of memory products.

Synergy between HBM and CXL will contribute significantly to AI development; their visibility will increase across different applications starting in 2023

TrendForce believes that the market penetration rate of CXL will rise going forward as this interface standard is built into more and more CPUs. Also, the combination of HBM and CXL will be increasingly visible in the future hardware designs of AI servers. In the case of HBM, it will contribute to a further ramp-up of data processing speed by increasing the memory bandwidth of the CPU or the accelerator. As for CXL, it will enable high-speed interconnections among CPU and other devices. By working together, HBM and CXL will raise computing power and thereby expedite the development of AI applications.

The latest advances in memory pooling and sharing will help overcome the current hardware bottlenecks in the designs of different AI models and continue the trend of more sophisticated architectures. TrendForce anticipates that the adoption rate of CXL-supported Sapphire Rapids processors will reach a certain level, and memory suppliers will also have put their HBM3 products and their CXL-supported DRAM and SSD products into mass production. Hence, examples of HBM-CXL synergy in different applications will become increasingly visible from 2023 onward.

For more information on reports and market data from TrendForce’s Department of Semiconductor Research, please click here, or email Ms. Latte Chung from the Sales Department at lattechung@trendforce.com