In the ongoing global race for advanced semiconductor technology, TSMC, the leader in semiconductor manufacturing services, continues its strides towards 2nm project. The Hsinchu’s Baoshan plant is set to commence equipment installation in Q2 2024, with mass production scheduled for Q4 2025, starting with a monthly output of around 30,000 wafers. Meanwhile, TSMC fab in Kaohsiung is organizing for N2P mass production, featuring backside power supply tech, a year after N2’s debut.

According to a report by Taiwan’s Money DJ, as previously shared by TSMC, the N2 process introduces a backside power rail solution, ideal for high-performance computing (HPC) applications. The backside power rail promises a 10% to 12% speed boost and a 10% to 15% logic density improvement. The aim is to introduce backside power rail to customers in H2 2025, aligning with supply chain reports.

Notably, Intel led the transition from planar transistors to FinFET, and now, with evolving technologies like MBCFET, BSPDN (Backside Power Delivery Network) based on Gate-All-Around (GAA) FET. Major players such as TSMC, Samsung, and Intel actively compete for leadership in the next-gen GAA technology, and have further presented promising and proactive technology roadmaps.

According to Samsung Semiconductor’s plans, they target to implement the 2nm process into mass production by 2025, with 1.4nm scheduled for 2027. Intel, adopting RibbonFET transistor architecture based on GAA technology, anticipates pilot production of the 20A version in H1 2024 and mass production of the 18A in 2025.

Amidst the AI boom, HBM technology steps into the spotlight as market demand continues to surge. Global market research firm TrendForce anticipates a 58% year-on-year increase in HBM demand in 2023, with a potential additional growth of approximately 30% in 2024.

Compared to traditional DRAM, HBM (High Bandwidth Memory) boasts advantages such as high bandwidth, high capacity, low latency, and low power consumption. These attributes accelerate AI data processing and make it particularly well-suited for high-performance computing scenarios like ChatGPT. As a result, it has gained popularity, and major storage manufacturers are actively driving HBM technology upgrades.

Leading memory manufacturers are intensifying their efforts, with Samsung set to introduce HBM4.

Since the inception of the first HBM products utilizing TSV packaging technology in 2014, HBM technology has seen multiple upgrades, including HBM, HBM2, HBM2E, HBM3, and HBM3e.

Regarding the SK Hynix and Samsung, two major South Korean companies, have been at the forefront of HBM3 development. NVIDIA’s H100/H800 and AMD’s MI300 series, represent HBM3’s progress. Both SK Hynix and Samsung expected to offer HBM3e samples by the first quarter of 2024. On the other hand, Micron, a U.S.-based memory company, is bypassing HBM3 and directly pursuing HBM3e.

HBM3e will feature 24Gb mono die stacks, and with an 8-layer (8Hi) configuration, a single HBM3e chip’s capacity will soar to 24GB. This advancement is expected to be incorporated into NVIDIA’s GB100 in 2025, leading the three major OEMs to plan HBM3e sample releases in the first quarter of 2024 and enter mass production in the latter half of the year.

In addition to HBM3 and HBM3e, the latest updates indicate that storage giants are planning the launch of the next generation of HBM—HBM4.

Samsung recently announced that it has developed 9.8Gbps HBM3E and is planning to provide samples to customers. Furthermore, Samsung is actively working on HBM4 with a goal to begin supply in 2025. It’s reported that Samsung Electronics is developing technologies such as non-conductive adhesive film (NCF) assembly for optimizing high-temperature thermal characteristics, as well as hybrid bonding (HCB), for HBM4 products.

In September, Korean media reported that Samsung is gearing up to revamp its production process and launch HBM4 products to capture the rapidly growing HBM market. HBM4 memory stacks will feature a 2048-bit memory interface, a significant departure from the previous 1024-bit interface for all HBM stacks. This enhanced interface width holds great significance for the evolution of HBM4.

While HBM4 promises a major breakthrough, it is still a ways off, making it too early to discuss its practical applications and widespread adoption. Industry experts emphasize that the current HBM market is dominated by HBM2e. However, HBM3 and HBM3e are poised to take the lead in the near future.

According to TrendForce’s research, HBM2e currently accounts for the mainstream market share, being used in various products like NVIDIA A100/A800, AMD MI200, and many AI accelerators developed by CSPs. To keep pace with the evolving demands of AI accelerator chips, OEMs are planning to introduce new HBM3e products in 2024, with HBM3 and HBM3e expected to become the market’s primary players next year.

In terms of the demand transition between different HBM generations, TrendForce estimates that in 2023, mainstream demand will shift from HBM2e to HBM3, with estimated demand shares of approximately 50% and 39%, respectively. As more HBM3-based accelerator chips enter the market, demand will substantially shift toward HBM3 in 2024, surpassing HBM2e and accounting for an estimated 60% of the market. This transition, coupled with higher average selling prices (ASP), is poised to significantly drive HBM revenue growth next year.

(Photo credit: Samsung)

Source to UDN, the DRAM market has been buzzing with positive developments lately, and may get a chance to see an upturn by the end of the year. Among the key factors driving this optimism is the DDR5 specification DRAM, which is poised to capitalize on opportunities in AI servers and laptops next year, gradually increasing demand.

After more than a year of corrections, the DRAM market is finally showing signs of improvement. Major DRAM manufacturers like Samsung and SK Hynix are still reducing production capacity, but their focus is primarily on DDR4 specification DRAM. Industry sources suggest that Samsung, in response to the growing demand for DDR5 DRAM, is set to significantly ramp up DDR5 production in the fourth quarter of this year, anticipating strong order demand next year.

In fact, such as Intel and AMD are planning to introduce new platforms next year that will support DDR5 specification DRAM, indicating a gradual decline in DDR4 demand. Beyond the consumer market, the server market is expected to experience a substantial surge in DDR5 demand, driven by the imminent launch of Intel’s fifth-generation server platform, Emerald Rapids, which fully supports DDR5. As AI server demand gains momentum, DDR5 demand is poised to enter a high-growth phase.

(Source: https://money.udn.com/money/story/5612/7488629 ; Image credit: Micron)

As reported by China’s Jiwei on October 2nd, Samsung has revealed that its chip manufacturing division has secured an order from AI chip client Tenstorrent to produce chips utilizing its cutting-edge 4nm process.

…

According to a report by Taiwan’s TechNews, Samsung Electronics’ semiconductor foundry division, Samsung Foundry, has been operating at less than 50% utilization rate as of the second quarter due to weak demand for 8-inch wafer foundry services. Industry sources have revealed that Samsung Foundry has already halted operations on 30% of its equipment, but with further inventory reduction expected, there is a possibility of restarting these machines by the end of the year and resuming production in the first quarter of the following year.

Previously, South Korean media outlet The Elec reported that the IT industry’s demand is currently low, leading South Korean wafer foundries to reduce prices for 8-inch wafer services by 10%. As of the second quarter, both Samsung Foundry and other South Korean wafer foundry companies like Key Foundry and SK Hynix System IC, a subsidiary of SK Hynix engaged in foundry operations, have been operating at capacity utilization rates ranging from 40% to 50%.

8-inch wafer services primarily manufacture components such as power management ICs, panel driver ICs, and microcontrollers. Given the uncertain demand for consumer electronics products, Samsung Foundry has decided to halt operations on 30% of its equipment as a cost-saving measure. However, market expectations suggest that the overall manufacturing and semiconductor industries have undergone more than a year and a half of inventory adjustment, and there is hope for inventory replenishment by the end of the year in the supply chains of three major sectors: smartphones, PCs, and consumer electronics.

According to industry sources, Samsung Foundry plans to restart the halted 8-inch wafer equipment by the end of the year and aims to resume full production in the first quarter of the next year. However, the revival of consumer product demand has not yet shown clear signs, and whether the restart will proceed as expected remains to be observed.

(Photo credit: Samsung)