At its previous earnings call in July, Samsung has announced the ambitious goal that its HBM sales would increase three to five times in 2H24. However, as it is still struggling to pass the verification of 12-Hi HBM3e products, the company’s prospects for returning to glory in the near term seems to be rather dim.

According to a report by Korean media outlet ZDNet, the main issue may lie in the core die of HBM, while the adoption of 1a DRAM is hindering Samsung’s recent efforts to supply HBM3e for NVIDIA.

Notably, an insider cited by the report notes that Samsung’s Vice Chairman Jun Young-hyun, the new Head of Device Solutions (DS) Division, is aware of these issues, so the decision for whether to redesign the 1a DRAM or not may be made soon.

According to the report, Samsung began the mass production of 1a (4th generation) DRAM, which has a linewidth of approximately 14 nm, as early as in the second half of 2021. It is worth noting that the company tries to enhance the product’s competitiveness by actively adopting advanced technologies such as EUV (extreme ultraviolet lithography).

ZDNet notes that Samsung applied five EUV layers to its 1a DRAM, which is significantly more than the one layer used by its major competitor, SK hynix.

However, though EUV is advantageous for reducing linewidths compared to the existing ArF (argon fluoride) lithography process, which is supposed to enhance efficiency and thus lowering manufacturing costs, EUV’s high technical difficulty has negatively affected the stability of the process, according to the report.

As a result, the cost of Samsung’s 1a DRAM has not decreased as initially anticipated, according to the report, with the yield issue occurring reportedly hinders Samsung’s HBM3e verification progress with NVIDIA.

Previous reports indicate that Samsung had conducted an on-site inspection with NVIDIA regarding the 8-layer HBM3e products at its Pyeongtaek campus. While the inspection itself concluded without any issues, concerns have reportedly been raised as the data processing speed (Gbps) of Samsung’s 8-layer HBM3e is about 10% lower compared to its rivals, according to sources cited by ZDNet.

Both SK hynix and Micron utilize 1b DRAM for their HBM3e core dies, the report notes.

Therefore, industry insiders cited by ZDNet reveal that Samsung has been internally discussing the possibility of redesigning some of the circuits in its 1a DRAM.

However, If Samsung does proceed with the redesign, it is expected to take at least six months for the product to be completed, the report suggests, which means the mass production could only begin by the second quarter of next year, and it will be challenging to supply the product in a timely manner.

Read more

• [News] Lagging behind SK hynix in HBM, Samsung Plans Significant Year-end Cuts to Chip Executives
• [News] Samsung Rumored to Cut HBM Production Target by 10% by End of 2025 amid HBM3e Delays

(Photo credit: Samsung)

TSMC held its earnings call earlier today, October 17th. According to a report in Economic Daily, based on an exchange rate of USD 1 to TWD 32, the company estimates the fourth-quarter revenue to be between USD 26.1 billion and USD 26.9 billion, representing an average quarter-over-quarter increase of 13%.

For the third quarter, TSMC reported consolidated revenue of USD 23.5 billion. The gross margin was 57.8%, while the operating profit margin was 47.5%.

Looking ahead to the fourth quarter, TSMC estimates that the revenue will be between USD 26.1 billion and USD 26.9 billion, estimated an average quarter-over-quarter increase of 13%. The gross margin of the fourth quarter will range between 57% and 59%, with the operating profit margin expected to be between 46.5% and 48.5%.

As TSMC has announced its estimated revenue for the fourth quarter, the company anticipates a year-over-year increase of 30% for 2024, revised up from the previous estimate of 24%-26%. TSMC Chairman and CEO C.C. Wei noted that this projected growth of nearly 30% is driven by advanced technology and AI.

According to a report from Commercial Times, driven by strong AI demand, TSMC will continue its investments in this area. Regarding the capital expenditure, TSMC estimated that the capital expenditure of 2024 will be slightly above USD 30 billion, aligning closely with the projected range of USD 30 billion to USD 32 billion estimation from the Q2 earnings call in July 2024.

(Photo credit: TSMC)

At TSMC’s third-quarter earnings call on the 17th, Chairman C.C. Wei was asked about potential interest in Intel’s spin-off of its IDM business. According to a report from TechNews, Wei responded with two decisive “No’s,” indicating that TSMC has no interest in acquiring the businesses.

C.C. Wei noted that a California-based IDM has been a strong customer for TSMC, consistently placing sizable orders. While Wei did not explicitly name the company, industry observers widely believe the client to be Intel.

During the call, analysts also raised questions about whether the current AI surge might be a bubble and how TSMC views the AI boom.

Wei confidently stated, “AI is real,” and noted that many large-scale cloud customers and AI innovators are collaborating with TSMC. He also mentioned that TSMC utilizes AI and machine learning (ML) within its own facilities to enhance capacity, add value, and improve yield rates.

Wei explained that for TSMC, even a 1 percent improvement in efficiency through AI could result in a revenue increase of NT$1 billion. He also noted that TSMC is not the only company benefiting from AI, and many other businesses are already leveraging AI to boost productivity.

Wei further stated that the demand for AI is just beginning. He referenced a key customer who described the current demand as crazy and noted that it is only starting, with expectations that this trend will continue for several years. This was widely seen as aligning with earlier remarks from NVIDIA CEO Jensen Huang regarding the high demand for chips based on the Blackwell architecture.

(Photo credit: TSMC)

According to a report by Commercial Times, as NVIDIA’s GB200 is set to officially ship by the end of this year and gradually ramping up in 2025, it is anticipated to usher in a new era of liquid cooling.

TrendForce predicts that the liquid cooling penetration will rise from 11% in 2024 to 24% in 2025. This growth is expected to be fueled by advancements in AI, which we may witness in 2025.

The expansion of liquid cooling will benefit several Taiwanese companies, including Delta Electronics, Asia Vital Components, Auras Technology, LOTES, and Fositek.

The growing global awareness of ESG principles, coupled with cloud service providers (CSPs) accelerating the deployment of AI servers and the upcoming official launch of NVIDIA’s Blackwell platform all contribute to the growth of liquid cooling technology, as the report in Commercial Times indicated.

According to the report, the market is optimistic that the significant growth in NVIDIA Blackwell’s penetration will drive a shift from air cooling to liquid cooling solutions.

Currently, the Hopper architecture remains dominant, commanding 90% of the market share this year. On the other hand, Blackwell is projected to capture only 4% of the market by the end of 2024. However, as Blackwell ramps up sales next year, it is expected to replace Hopper and establish itself as the new mainstream.

From the perspective of thermal design power (TDP), NVIDIA’s GB200 NVL72 racks, with a TDP of approximately 140 kW, will require liquid cooling solutions to manage heat dissipation, with Liquid-to-Air (L2A) technology anticipated to become the predominant method.

While immersion cooling is an option, it faces technical challenges that may prevent it from becoming mainstream in the next 3 to 5 years, as the report stated.

The report highlighted that the market price of a single NVL72 rack is 28 times that of a traditional server. As a result, an increasing number of companies are entering and competing in the liquid cooling market.

Currently, for those Taiwanese companies in the supply chain which may benefit from the market opportunities, Asia Vital Components and Cooler Master lead in providing cold plates, while Cooler Master and Auras supply manifolds, and Vertiv and Delta Electronics provide coolant distribution units.

On the other hand, quick disconnect (QD) components are critical for preventing leakage and have frequently been reported as out of stock. Taiwanese companies like LOTES and Fositek are reportedly in the validation phase, and by the first half of 2025, they are expected to join the list of QD suppliers, to compete with major companies such as CPC, Parker Hannifin, Denfoss, and Staubli.

Read more

• Blackwell Platform and ASIC Chip Upgrades to Boost Liquid Cooling Penetration to Over 20% in 2025, Says TrendForce

(Photo credit: NVIDIA)

TSMC today announced consolidated revenue of NT$759.69 billion, net income of NT$325.26 billion, and diluted earnings per share of NT$12.54 (US$1.94 per ADR unit) for the third quarter ended September 30, 2024.

According to the press release issued by TSMC, year-over-year, third-quarter revenue increased by 39.0%, while net income and diluted EPS both rose by 54.2%. Compared to the second quarter of 2024, third-quarter results showed a 12.8% increase in revenue and a 31.2% increase in net income. All figures were prepared in accordance with TIFRS on a consolidated basis.

In US dollars, third quarter revenue was $23.50 billion, which increased 36.0% year-over-year and increased 12.9% from the previous quarter.

Gross margin for the quarter was 57.8%, operating margin was 47.5%, and net profit margin was 42.8%.

In the third quarter, shipments of 3-nanometer accounted for 20% of total wafer revenue; 5-nanometer accounted for 32%; 7-nanometer accounted for 17%. Advanced technologies, defined as 7-nanometer and more advanced technologies, accounted for 69% of total wafer revenue.

(Photo credit: TSMC)