In early June, NVIDIA CEO Jensen Huang revealed that Samsung’s High Bandwidth Memory (HBM) is still striving on the certification process, but is one step away from beginning supply. On July 4^th, a report from Korea media outlet Newdaily indicated that Samsung has finally obtained approval from the GPU giant for qualification of its 5th generation HBM, HBM3e. It is expected that Samsung will soon proceed with the subsequent procedures to officially start mass production for HBM supply, the report noted.

Citing sources from the semiconductor industry, the report stated that Samsung recently received the HBM3e quality test PRA (Product Readiness Approval) notification from NVIDIA. It is expected that negotiations for supply will commence afterward.

However, just one hour after the news reported that Samsung’s HBM3e passed NVIDIA’s tests, another media outlet, Hankyung, noted that Samsung has denied the rumor, clarifying it is “not true,” and that the company is consistently undergoing quality assessments.

TrendForce reports that Samsung is still collaborating with NVIDIA and other major customers on the qualifications for both 8-hi and 12-hi HBM3e products. The successful qualification mentioned in the article was only an internal achievement for Samsung. Samsung anticipates that its HBM3e qualification will be partially completed by the end of 3Q24.

Per a previous report from Reuters, Samsung has been attempting to pass NVIDIA’s tests for HBM3 and HBM3e since last year, while a test for Samsung’s 8-layer and 12-layer HBM3e chips was said to fail in April.

According to the latest report from Newdaily, Samsung’s breakthrough came about a month after the memory heavyweight sent executives in charge of HBM and memory development to the U.S. at NVIDIA’s request. Previously, it was reported that Samsung had failed to pass the quality test as scheduled due to issues such as overheating.

The report further stated that though from Samsung’s perspective, supplying HBM to NVIDIA was crucial, NVIDIA is also eager to receive more HBM, with the overwhelming demand for AI semiconductors and the impending mass production of its next-generation GPU Rubin, which significantly increases HBM usage.

According to the report, Samsung is expected to eliminate uncertainties in HBM and start full-scale mass production, giving a significant boost to its memory business. There are also suggestions that its HBM performance could see a quantum leap starting in the second half of the year.

On the other hand, Samsung’s major rival, SK hynix, is the primary supplier of NVIDIA’s HBM3 and HBM3e. According to an earlier TrendForce’s analysis, NVIDIA’s upcoming B100 or H200 models will incorporate advanced HBM3e, while the current HBM3 supply for NVIDIA’s H100 solution is primarily met by SK hynix.

According to a report from the Financial Times in May, SK hynix has successfully reduced the time needed for mass production of HBM3e chips by 50%, while close to achieving the target yield of 80%.

Read more

• [News] Jensen Huang Confirms NVIDIA Approaches Certification of Samsung’s HBM Chips
• HBM3e Production Surge Expected to Make Up 35% of Advanced Process Wafer Input by End of 2024, Says TrendForce

(Photo credit: Samsung)

According to a report from Economic Daily News, TSMC’s advanced packaging platform SoIC is said to have secured another heavyweight customer, with Apple expected to adopt the technology in 2025. If confirmed, Apple will join AMD as a major client expanding its use of TSMC’s SoIC .

TSMC has categorized advanced packaging under its 3D Fabric system integration platform, which comprises three main parts: the SoIC series for 3D silicon stacking technology, and the CoWoS and InFo families for back-end advanced packaging.

Reportedly, it is said that the CoWoS family has been facing capacity constraints recently. To address this, TSMC is not only expanding its own production but also collaborating with testing service providers to increase output.

On the other hand, TSMC’s SoIC platform, which is part of front-end packaging, has fewer bottlenecks and began small-scale production in 2022. TSMC has long-term plans to expand SoIC capacity by more than 20 times by 2026 to meet growing customer demand.

In recent years, NVIDIA and AMD have been aggressively targeting the AI market, setting high growth targets for 2024. Both companies have sought collaboration with TSMC and several Taiwanese supply chain partners. The key advantage is Taiwan’s comprehensive supply chain, which can accelerate innovation. As TSMC’s advanced packaging capacity ramps up, industry analysts are optimistic that this will facilitate the smooth procurement and delivery of critical components.

In the highly anticipated SoIC area, AMD’s MI300 series is a recent story of deepened collaboration with TSMC. According to information from AMD and TSMC’s technology forum, the MI300 series not only uses TSMC’s 5nm process but also integrates multiple technologies from TSMC’s 3DFabric platform. This includes stacking the 5nm GPU and CPU on a base chip using SoIC-X technology and further integrating them into CoWoS packaging.

Beyond AMD’s adoption, the same report has cited rumors that Apple might adopt this technology in 2025.

Although TSMC consistently refrains from commenting on individual client details, industry speculation has long suggested that Apple intends to incorporate related technology into the next-generation M-series chips, and possibly even the A-series processors. This could significantly increase transistor density, driving the next wave of innovation in the mobile device and AI PC markets.

Read more

• [News] TSMC Reportedly Secures 4 Major Clients for SoIC, Including Apple, NVIDIA and Broadcom
• [News] TSMC’s SoIC Demand Heats Up, Reports Suggest Significant Capacity Expansion

(Photo credit: TSMC)

While NVIDIA is likely to face accusations from the French antitrust regulators, the Non-NVIDIA Alliance like the UALink (Ultra Accelerator Link) Alliance and the UXL Foundation are reportedly launching a counterattack, significantly increasing their efforts in developing specialized ASICs.

According to a report from Commercial Times, relevant semiconductor intellectual property (IP) is expected to be widely adopted. The sources cited by the report point out that Taiwanese manufacturers, benefiting from their leading position in wafer foundry and comprehensive ASIC and IP layout, are poised to capitalize on the rise of the Non-NVIDIA Alliance.

The report further cites sources, indicating that major Taiwanese ASIC manufacturers such as Global Unichip, Faraday Technology, and Progate Group Corporation (PGC), along with silicon IP companies M31 Technology Corporation, eMemory, and the Egis Technology Group, are actively expanding in this field.

In order to challenge NVIDIA’s dominance in the market, UALink (Ultra Accelerator Link) Alliance, led by tech giants such Intel and AMD, was formed in May. The alliance aims to establish a new standard for AI accelerator links, aiming to challenge NVIDIA’s  NVLink.

Furthermore, the UXL Foundation’s Open Source Software Project, supported by tech giants Qualcomm, Google, and Intel, is said to be looking to rival NVIDIA’s CUDA software. By providing alternative software solutions, it aims to diminsh NVIDIA’s dominance in the AI field.

Semiconductor industry sources cited in the same report also note that CSPs are accelerating the development of their own chips, with Taiwanese manufacturers actively entering the market.

Although Broadcom and Marvell currently offer diversified design services, Taiwanese manufacturers have an advantage due to the tightly-knit semiconductor supply chain. This enables complete solutions for both chip manufacturing and packaging within Taiwan, giving them a strategic edge over competitors by being close to both the market and factories, thereby enhancing their position in the ASIC sector.

Global Unichip and PGC leverage TSMC as a strong ally. Reportedly, Global Unichip holds AI-related ASIC orders from Microsoft and is gradually finalizing collaborations with major South Korean companies, with business operations expected to improve in the second half of the year.

On the other hand, Faraday Technology closely collaborates with Intel, developing SoCs using Intel’s A18 process. Meanwhile, industry sources cited by the report suggest that Intel’s Gaudi series AI chips might seek collaboration opportunities beyond just working with Alchip.

Read more

• [News] Novatek, a Major Driver IC Manufacturer, Reportedly Eyes Entry into ASIC and Silicon IP Battlefields

(Photo credit: Shutterstock)

On July 1st, according to a report from Reuters, the French antitrust authority plans to file charges against NVIDIA, accusing the company of engaging in anti-competitive practices, making France the first country to take such action against NVIDIA.

The French competition regulator had raided NVIDIA’s local offices in September last year. At the time, they did not disclose the details of the investigation or the company involved, only stating it was related to the graphics card sector.

However, as per a previous report from Bloomberg, NVIDIA claimed that the French agency collected information from them regarding their business and competition in the graphics card and cloud service provider market as part of an ongoing inquiry into competition in those markets.

Sources cited by Reuters’ report indicated that last year’s raid was part of a broader investigation into cloud computing. With the surge in global chip demand following the advent of ChatGPT, NVIDIA, as the world’s largest manufacturer of AI and computer graphics cards, has naturally attracted close scrutiny from antitrust authorities in Europe and the United States.

NVIDIA previously disclosed in regulatory filings that both EU and French regulators had requested information about its graphics card products. The French antitrust authority has been actively investigating to understand NVIDIA’s key role in AI processors, its pricing policies, chip shortages, and the impact on prices.

Last Friday, the French authorities released a report on competition in generative AI, highlighting the risk of chip suppliers abusing their power. The report pointed out concerns about the chip industry’s heavy reliance on NVIDIA’s CUDA software for chip programming. Additionally, NVIDIA’s focus on investing in AI cloud service provider CoreWeave has also raised significant concerns among the authorities.

Reportedly, it is understood that companies violating French antitrust rules could face fines of up to 10% of their global annual revenue, though they can choose to make concessions to avoid penalties.

Moreover, the European Commission is currently gathering informal feedback to determine if NVIDIA has breached its antitrust rules, although it has not yet launched a formal investigation into anti-competitive behavior.

On the other hand, the New York Times reported on June 5th that the U.S. Department of Justice and the Federal Trade Commission (FTC) have reached an agreement, led by senior officials of both agencies, over the past week. The DOJ will investigate whether NVIDIA has violated antitrust laws, while the FTC will examine the conducts of OpenAI and Microsoft.

Read more

• [News] NVIDIA’s H200 Order Delivered from Q3, Boosting Server Supply Chain with Strong Demand
• [News] US Government Reportedly Investigating Microsoft, NVIDIA, and OpenAI for AI Industry Monopolization

(Photo credit: NVIDIA)

In late March, NVIDIA CEO Jensen Huang personally delivered the first high-end DGX H200 AI server to customer OpenAI. According to a report from Commercial Times, following this, the upstream chips for the H200 AI-GPU entered mass production in late Q2, with large-scale deliveries expected from Q3 onwards.

Downstream companies, including Inventec, Quanta (QCT), Hewlett Packard Enterprise (HPE), Supermicro, and Lenovo, have reportedly listed the H200 products as ready for shipment, with deliveries anticipated to begin in the second half of the year.

The same report, citing sources, indicates that current pending orders are still largely focused on the H100 in the HGX architecture, with the H200’s share remaining limited. The H200 shipments expected in Q3 will primarily be NVIDIA’s DGX H200. As for the B100, there is already some visibility, with shipments expected in the first half of next year.

Despite the CoWoS production capacity catching up and a significant easing in the supply of AI GPUs, due to the allocation system, the delivery time for the main H100 GPU shipments from various end-system partners can still reach up to 20 weeks.

However, major Taiwanese manufacturers such as Quanta, Inventec, Wistron, Gigabyte, and ASUS have seen a substantial boost in their overall server operations, driven by AI server business in the first half of the year. They are reportedly optimistic about AI server shipments in the second half of the year, expecting strong demand to continue, making the server business effectively free from any off-season throughout the year.

Among them, Wistron holds a significant advantage in the production and supply of H100 series substrates and the subsequent B100 series GPU modules and substrates. Starting from the second quarter, Wistron’s AI server-related business has shown high visibility, providing strong support for its overall operations.

On the other hand, the H20 series, an AI chip tailored for China due to U.S. chip restrictions, has also seen demand in the Chinese market. As NVIDIA recently secured a rush order for the H20 series, Taiwanese companies, including Wistron and Inventec, are expected to benefit.

Read more

• [News] Taiwanese AI Server Supply Chain to Benefit from Apple’s Potential Collaboration with Baidu and Alibaba
• [News] Rumored Sharp Drop in H100 Server Black Market Prices in China Raises Concerns Over Market Stability

(Photo credit: NVIDIA)