According to a report from Commercial Times citing sources, it’s revealed that NVIDIA has executed changes to the Blackwell series’ 6-layer GPU mask. Therefore, the process can now proceed without re-taping out, as production delays being minimized.

The report noted that NVIDIA’s updated version of B200 is expected to be completed by late October, allowing the GB200 to enter mass production in December, with large-scale deliveries to ODMs expected in the first quarter of next year.

Previously, as per a report from The Information, NVIDIA’s GB200 was said to be experiencing a one-quarter delay in mass shipments. Another report from the Economic Daily News further suggested that the problem likely lies in the yield rates of advanced packaging, which mainly affected the non-reference-designed GB200 chips.

Industry sources cited by Commercial Times addressed that NVIDIA’s Blackwell chip used to be facing instability in metal layers during the HV process, which was then resolved by July.

In addition, since the issue reportedly occurred in the back-end-of-line process, a new tape-out was deemed unnecessary. Still, as CoWoS-L capacity remains a bottleneck, the advanced packaging for GB200 this year is expected to adopt  CoWoS-S.

Read more

• [News] Foxconn Confirms Shipment of NVIDIA’s GB200 to Start in Q4 as Planned
• [News] Foxconn Remains Unfazed Despite NVIDIA’s GB200 Delay, with Customers Aim to Secure Their Orders

(Photo credit: NVIDIA)

After its 8-Hi HBM3e entered mass production in February, Micron officially introduced the 12-Hi HBM3e memory stacks on Monday, which features a 36 GB capacity, according to a report by Tom’s Hardware. The new products are designed for cutting-edge processors used in AI and high-performance computing (HPC) workloads, including NVIDIA’s H200 and B100/B200 GPUs.

It is worth noting that the achievement has made the US memory chip giant almost on the same page with the current HBM leader, SK hynix. Citing Justin Kim, president and head of the company’s AI Infra division at SEMICON Taiwan last week, another report by Reuters notes that SK hynix is set to begin mass production of its 12-Hi HBM3e chips by the end of this month.

Samsung, on the other hand, is said to have completed NVIDIA’s quality test for the shipment of 8-Hi HBM3e memory, while the company is still working on the verification of its 12-Hi HBM3e.

Micron’s 12-Hi HBM3e memory stacks, according to Tom’s Hardware, feature a 36GB capacity, a 50% increase over the previous 8-Hi models, which had 24GB. This expanded capacity enables data centers to handle larger AI models, such as Meta AI’s Llama 2, with up to 70 billion parameters on a single processor. In addition, this capability reduces the need for frequent CPU offloading and minimizes communication delays between GPUs, resulting in faster data processing.

According to Tom’s Hardware, in terms of performance, Micron’s 12-Hi HBM3e stacks deliver over 1.2 TB/s. Despite offering 50% more memory capacity than competing products, Micron’s HBM3e consumes less power than the 8-Hi HBM3e stacks.

Regarding the future roadmap of HBM, Micron is said to be working on its next-generation memory solutions, including HBM4 and HBM4e. These upcoming memory technologies are set to further enhance performance, solidifying Micron’s position as a leader in addressing the increasing demand for advanced memory in AI processors, such as NVIDIA’s GPUs built on the Blackwell and Rubin architectures, the report states.

Read more

• [News] SK hynix to Begin 12-Stack HBM3e Mass Production, Marking Key Moment in HBM Battlefield
• [News] Samsung Likely Emerges as the Pacemaker for the AI Market if It Secures HBM3e Supply to NVIDIA

(Photo credit: Micron)

According to a report by the Economic Daily News, TSMC’s first high numerical aperture extreme ultraviolet lithography (High-NA EUV) equipment will arrive this month, aiding the company in progressing its advanced process technology.

Regarding these rumors, ASML stated on September 9 that it does not comment on individual customers. TSMC has also declined to respond to market speculation.

The report has cited industry rumors, pointing out that TSMC’s first High-NA EUV machine is expected to be moved to its Global R&D Center for research purposes, addressing the development needs of future advanced processes like A14.

As for pricing, it’s reportedly rumored that TSMC President C.C. Wei personally negotiated a favorable deal, reducing the overall price by nearly 20% through a combination of purchasing the new equipment while also selling older models.

The same report further suggests that the High-NA EUV is priced at over EUR 400 million. Due to the inability to disassemble the optical lens components, the equipment is taller than a conference room and significantly longer than the previous generation. 

The report, quoting statistics, claiming that TSMC is currently the world’s largest holder of extreme ultraviolet (EUV) lithography systems, estimated to account for 65% of the global EUV wafer equipment output. It was also said to be the first manufacturer to introduce EUV equipment into the 7nm process.

ASML has already received orders for next-generation products from all of its EUV equipment customers.

Greet Storms, ASML’s Vice President of High NA EUV Product Management, recently stated that ASML is advancing new technologies and has gained the support of all EUV customers during the R&D phase.

She addressed that, these customers, who have also placed orders for High-NA equipment, are expected to move toward mass production by 2026, although the timeline will ultimately depend on customer process costs and other factors.

ASML had previously confirmed that it will deliver its latest High-NA EUV to TSMC by the end of this year. ASML emphasized that the next-generation EUV equipment began shipping at the end of last year, with a capacity to expose over 185 wafers per hour.

This will support mass production of sub-2nm logic chips and memories with similar transistor densities.

Read more

• [News] ASML’s Second High-NA Equipment to be Installed in Intel’s Oregon Fab Soon
• [News] Price for ASML’s Hyper-NA EUV Rumored to Double, Causing TSMC, Samsung and Intel to Hesitate

(Photo credit: ASML)

Grappling with a series of operational challenges, Intel has been weighing options to stem losses, with cost reduction being one of its major focus. Under this scenario, the decisions the semiconductor giant has made include slashing 15% of its workforce and the cancellation of its 20A process, which can help avoiding the significant capital expenditures needed to scale the node to full production.

In addition to the aforementioned moves, rumors have been circulating regarding the progress of Intel’s global expansion, hinting that the projects at Germany and Penang might be put on hold. Moreover, though granted with USD 8.5 billion in grants and USD 11 billion in loans under the Chips and Science Act, the company has not yet received any money from the U.S. authority, raising concerns on the feasibility of its ongoing Arizona and Ohio projects.

Here is a roundup of Intel’s major expansion projects which have been reportedly delayed, scaling down, or put on hold recently, as cost considerations may be the primary reason.

Accompanied by construction crews on Oct. 12, 2023, Intel CEO Pat Gelsinger toured the company’s two new fabrication plants being built in Chandler, Arizona. (Credit: Intel)

US: Ohio Delayed for 2 Years, with Mass Production Expected in 2027-28

Intel plans to invest USD 100 billion over the next five years in new fabs and expansions across Arizona, New Mexico, Ohio, and Oregon, creating 10,000 manufacturing jobs and 20,000 construction jobs.

However, according to a previous report by The Register, the construction of Intel’s two fabs in Arizona has progressed more slowly than anticipated. Rising costs for materials and labor, combined with a surge in investments in the state, have resulted in a shortage of workers.

Citing Nikkei, the report notes that the cost of constructing a plant in Arizona is now four to five times higher than in other regions, such as Asia, and several times more than the initial budget Intel had anticipated.

The semiconductor giant’s Fab 52 and Fab 62 in Arizona are previously scheduled to be completed in 2024. However, The Register notes that the schedule may be delayed a bit, as the fabs are likely to begin operations later this year or in early 2025, targeting to manufacture chips using Intel’s next-generation Angstrom-era process technology, including the 18A node.

The 20 billion project in Ohio, on the other hand, may be facing larger obstacles as Intel has delayed the plan due to market downturns and delays in U.S. subsidies.

With an initial plan to begin chip manufacturing in Ohio in 2025, the company has now postponed the pipeline, aiming to complete the two fabs in 2026–2027, with operations expected to commence around 2027–2028.

EMEA: Hard-hit Area as Investments on Hold in Israel, France and Italy

The situation is not too optimistic in Europe, neither. According to a report from global media outlet Volksstimme, the construction of Intel’s Fab 29.1 and Fab 29.2 near Magdeburg, Germany, which may totaling 30 billion euros, has been postponed due to pending approval of EU subsidies and the need to remove and reuse black soil. The date of commencement has been pushed from summer 2024 to May 2025.

It is reported that Intel’s fabs in Germany were originally scheduled to start operations by late 2027 and were expected to employ advanced manufacturing processes, potentially Intel 14A (1.4nm) and Intel 10A (1nm) nodes. However, Intel now estimates that it will take four to five years to build these plants, and production is expected to commence between 2029 and 2030.

In addition, earlier in June, the struggling semiconductor giant also put a halt to the expansion of a major factory project in Israel, which was set to invest an additional USD 15 billion in a new chip plant, according to The Times of Israel.

In December, Intel had announced plans to increase its investment in the chip manufacturing plant in Kiryat Gat, from USD 10 billion to USD 25 billion, in order to secure a USD 3.2 billion grant from the Israeli government. However, it has now suspended the plan, suggesting that “managing large-scale projects, especially in our industry, often involves adapting to changing timelines,” the report states.

It is also worth noting that Intel tends to scale down on its production in the EMEA area, as in June, the company announced the sale of a 49% stake in its plant in Leixlip, Ireland, to Apollo Global Management for USD 11 billion, securing more external funding for usage.

Another report by POLITICO in July also notes that the U.S. semiconductor giant has discreetly suspended several investment plans in Europe after incurring significant losses, dealing a setback to Europe’s push to produce more microchips.

In 2022, Intel announced an initial investment of over €33 billion for R&D and Manufacturing in EU, targeting countries including France, Germany, Ireland, Italy, Poland and Spain. However, the report suggests that the promised investments for France and Italy, valued at billions of euros and potentially creating thousands of jobs, will be put on hold for now.

Asia: New Packaging Project in Malaysia Suspended

The latest of Intel’s moves on halting its global expansion would probably be the cancellation of its new chip packaging and testing project in Penang as part of cost-cutting efforts, while the operations of existing facilities will remain unaffected.

The U.S. chip giant had announced three years ago that it would invest approximately USD 7 billion to build new chip packaging and testing facilities in Malaysia, looking to make the site its largest overseas packaging and testing base. Per a report by Malaysia media outlet The Star, Intel employs around 14,000 people in Malaysia, meaning over 2,000 local employees may face the risk of job loss.

Would the series of emergency measures taken by the struggling giant stem the bleeding? Or would these moves more like a cornered beast fighting out of its way? The only thing for sure now may be that Intel would have a pretty different look after the upcoming board meeting in mid-September.

Read more

• [News] Qualcomm Rumored to Consider Acquiring Parts of Intel’s Chip Design Business, with a Focus on PC Segment
• [News] Amidst Rumors of Germany Plant Delay, Intel’s Penang Project Also Reportedly on Hold

Despite U.S. export controls aimed at preventing Chinese companies from acquiring advanced AI chips, small cloud service providers in China have reportedly found ways to obtain NVIDIA’s A100 and H100 chips. The cost of renting cloud services in China is even lower than in the U.S.

According to a report from the Financial Times, four small-scale Chinese cloud providers are offering servers equipped with eight A100 chips each, charging around USD 6 per hour. In comparison, similar services from U.S. cloud providers cost approximately USD 10 per hour.

As Chinese companies are reportedly bypassing U.S. export controls, industry sources cited by the Financial Times have further noted that the lower prices in China may hint at a robust local supply of NVIDIA chips.

Since the fall of 2022, the U.S. has banned NVIDIA from supplying A100 chips to China, and the more powerful H100 chips have not been approved for sale there.

However, industry sources and startups have revealed that these chips are still available in China. Ads for A100 and H100 have appeared on social media platforms like Xiaohongshu and e-commerce sites such as Taobao, with prices higher than those abroad.

At the Huaqiangbei electronics market in Shenzhen, reportedly, industry sources have revealed that the price of NVIDIA’s H100 is quoted at USD 23,000 to USD 30,000, while Chinese online sellers list it at USD 31,000 to USD 33,000.

Meanwhile, larger Chinese cloud providers such as Alibaba and ByteDance emphasize service stability and security in the local market. For servers equipped with A100 chips, they charge two to four times more than smaller cloud providers.

According to another source cited by Financial Times, large companies must consider regulatory compliance, which puts them at a disadvantage because they are reluctant to use smuggled chips. In contrast, smaller providers are less concerned.

The same report also indicate that after the US government tightened export controls in October last year, servers from Supermicro equipped with eight H100 chips were priced as high as approximately CNY 3.2 million. However, as supply constraints eased, the price has dropped to around CNY 2.5 million.

Several sources cited by the report claim that merchants from Malaysia, Japan, and Indonesia frequently ship Supermicro servers or NVIDIA chips to Hong Kong, from where they are then transported to Shenzhen.

In response to these issues, NVIDIA reportedly stated that it primarily sells chips to well-known partners, ensuring that all sales comply with U.S. export regulations.

NVIDIA also mentioned that its used products can be obtained through various channels and, although they cannot track products after sale, they will take appropriate action if they determine a customer is violating U.S. export controls.

Read more

• [News] China’s 1H24 Chip Equipment Purchases Exceed Taiwan, Korea, and US Combined, Reaching USD 25 Billion
• [News] China Contributes Over 40% of U.S. Chip Equipment Makers’ Revenue despite Export Controls

(Photo credit: iStock)