According to a report from Nikkei citing sources, TSMC is developing a new advanced chip packaging technology that uses rectangular panel-like substrates, rather than the traditional circular wafers currently in use.

This technology allows for more chips to be placed on a single substrate, addressing the future demand trends driven by AI. Although the research is still in its early stages and may take several years to reach mass production, it represents a significant technological shift for TSMC, the report notes.

Reportedly, TSMC is currently experimenting with rectangular substrates measuring 515 mm by 510 mm, providing more than three times the usable area compared to the current 12-inch wafers, and therefore can better suit the demand for AI chipsets.

In response to Nikkei’s inquiry, TSMC stated that the company is closely monitoring advancements and developments in advanced packaging technologies, including panel-level packaging.

Read more

• [News] TSMC Advanced Packaging Crucial for AI Computing Power

(Photo credit: TSMC)

While TSMC faces overwhelming demand for its 3nm technology, with orders from major clients like Apple and NVIDIA pouring in, Intel has now announced its progress on the 3nm technology. According to the latest report by Tom’s Hardware, Intel 3 has entered high-volume production at two sites, the Oregon and Ireland factories, with datacenter-related applications being the node’s primary focus.

However, the capacity seems to be mainly allocated to in-house chips for now. Citing Walid Hafez, Vice President of Foundry Technology Development at Intel, the report notes that Intel’s recently launched Xeon 6 “Sierra Forest” and “Granite Rapids” processors are being manufactured with the company’s 3nm node. Eventually, Intel will utilize this production node to produce datacenter-grade processors for its customers, the report states.

According to Tom’s Hardware, in addition to the standard Intel 3, Intel will also provide various versions for its 3nm node. The semiconductor giant plans to offer Intel 3T, which supports through silicon vias and can serve as a base die. Looking ahead, Intel aims to introduce the feature-enhanced Intel 3-E for chipsets and storage applications, as well as the performance-enhanced Intel 3-PT, designed for various workloads such as AI, HPC, and general-purpose PCs.

According to the report, the Intel 3 process offers both higher performance and increased transistor density, and it supports 1.2V for ultra-high-performance applications. In terms of performance, Intel claims that the new node will deliver an 18% improvement compared to Intel 4.

Regarding major competitors’ development on the 3nm node, TSMC is reported to receive strong demand from clients like Apple and NVIDIA, booking its capacity through 2026, and possibly leading to a price increase by over 5% in the node.

Like Intel, TSMC also offers various choices for its 3nm process. The members of TSMC’s 3nm family include N3, N3E, N3P, as well as N3X and N3A.

As the existing N3 technology continues to be upgraded, N3E, which began mass production in the fourth quarter of last year, targets applications such as AI accelerators, high-end smartphones, and data centers.

N3P is scheduled for mass production in the second half of this year and is expected to become mainstream for applications in mobile devices, consumer products, base stations, and networking through 2026. N3X and N3A are customized for high-performance computing and automotive clients.

On the other hand, Samsung’s second-generation 3nm production line in South Korea will reportedly commence operations in the latter half of this year  The first product to be manufactured on this line will reportedly be the application processor (AP) for the upcoming Galaxy Watch7, tentatively named “Exynos W1000,” which is expected to be unveiled in July.

Read more

• [News] Intel CEO Gelsinger Unveils Lunar Lake Processors, Giving Credit to TSMC
• [News] CoWoS Booming, TSMC Price Hikes Reportedly Imminent

(Photo credit: Intel)

As tech heavyweights eagerly pursue more market share in the AI sector, the battle for talents in the semiconductor industry has also heated up. According to the latest report by The Chosun Daily, citing LinkedIn data as of June 18, NVIDIA has become the hottest tech talent hub, not only drawing talents from semiconductor giants, but recruiting from memory companies in full swing.

According to the report, NVIDIA employs 89 former TSMC employees, while only 12 former NVIDIA employees have joined TSMC. Moreover, comparing with the number of former NVIDIA employees now at Samsung (278), there are 515 NVIDIA employees coming from Samsung Electronics, which indicates a significant talent migration to the GPU giant.

Regarding the talent war between NVIDIA and Intel, the former has attracted as many as 2,848 employees from Intel, whereas only 544 former NVIDIA employees have joined Intel.

NVIDIA also attracts talents with its charisma in AI from other memory giants. The LinkedIn data cited by the report shows that there are 38 NVIDIA employees previously with SK hynix, with none moving in the opposite direction. In addition, NVIDIA has attracted 159 employees from Micron, whereas only 38 former NVIDIA employees have joined Micron.

Interestingly enough, though Samsung lags behind in the competition with TSMC regarding the advanced nodes, it seems to be attractive to talents from the world’s largest foundry. Data show that there are 195 TSMC employees transitioning to Samsung, while only 24 former employees have joined TSMC.

An industry insider cited by the report observed that Korean semiconductor companies are vigorously recruiting for high-performance memory positions, such as those involving HBM. Moreover, a growing number of master’s and doctoral-level semiconductor experts in South Korea are joining the industry, showing a trend where talent moves from academia to local companies and then to international firms.

(Photo credit: NVIDIA)

Samsung is reportedly mulling to update its fabrication capabilities in the Taylor fab near Austin from 4nm to 2nm, and delaying the related equipment orders, according to reports from Tom’s Hardware and Korea media outlet ETnews. The decision is said to be made in Q3 2024.

Samsung’s initial projection, announced in 2021 when the investment plan was disclosed, had anticipated the Taylor plant to start its mass production of 4nm in the second half of 2024. But afterwards, the tech giant has reportedly postponed the schedule to 2025.

Now, according to the latest reports, with the AI hardware race intensifying and Nvidia, one of Samsung’s biggest clients, keeps advancing in its GPU platforms, 4nm production may no longer suffice. Therefore, Samsung has postponed its equipment orders for the Taylor fab as it considers the upgrade from 4nm to 2nm. The plant may not begin operations until 2026, US local media MySA noted.

According to information from Samsung’s press release, the company plans to provide one-stop solution for clients to expedite their production of AI chips, while its updated 2nm node with backside power delivery is expected to enter the market in 2027. SF4U, its high-value 4nm variant that offers PPA improvements by incorporating optical shrink, plans to enter mass production in 2025.

Earlier in April, the U.S. government announced that it will provide up to USD 6.4 billion in subsidies to Samsung for expanding advanced chip production capacity at its Texas plant, while it previously approved subsidies of up to USD 8.5 billion for US chip giant Intel and USD 6.6 billion for TSMC, respectively.

Regarding major semiconductor companies’ capacity expansion plans in the U.S., TSMC’s new fab in Arizona aims to complete all preparations for mass production by the end of this year, and commences mass production using the 4-nanometer process in the first half of 2025. On the other hand, TSMC also plans to build a third fab in Arizona, which will use 2nm process or even more advanced technologies to manufacture wafers for customers, though the date for construction has yet to be disclosed.

Meanwhile, Intel plans to mass-produce its 20A and 18A nodes at its Arizona and Ohio plants in 2024 and 2025, respectively.

(Photo credit: Samsung)

As the market demand for AI chips keeps booming, two South Korean AI chip companies, Sapeon and Rebellions, are pursuing a merger to enhance their competitiveness. According to a report by the Korea Economic Daily, if everything proceeds as planned, the merger will result in the formation of Korea’s first semiconductor unicorn, with the new company’s corporate value amounting to 2 trillion won (USD 1.5 billion).

According to a report by Korean media outlet TheElec, the merger will be finalized within this year, following the signing of the deal in the third quarter.

TheElec stated that Sapeon Inc, headquartered in the US and the sole owner of Sapeon Korea, will be the largest shareholder of the merged entity. Post-merger, Sapeon Inc is expected to hold about 30% of the shares, with Rebellions’ CEO, Sunghyun Park, leading the new company.

Based on the reports, the alliance has been formed in order to win in the global AI chip market, while the two companies identify the next two to three years as a crucial window of opportunity.

Sources cited by TheElec indicated that Sapeon Korea is valued at 500 billion won, while Rebellion is valued at 880 billion won. Regarding details of the merger, during its Series A funding round, Sapeon issued convertible bonds to its investors instead of equity. Afterwards, these investors will receive shares in the merged entity.

It is worth noting that both Sapeon and Rebellion have been backed up by tech heavyweights. According to The Elec, South Korean telco SK Telecom, as Sapeon Inc’s largest shareholder, holds 62.5% of the company. On the other hand, memory giant SK hynix owns 25%, while SK Square holds 12.5%.

Meanwhile, according to the Korea Economic Daily, Rebellions is part of the telecom giant KT Corp.-led Korean “AI full stack” service providers, offering AI infrastructure including AI chips, cloud computing, and various applications.

The report by TheElec, citing industry sources, noted that since SK Group will be the majority shareholder of the merged entity, it may prefer TSMC over Samsung Foundry, given that SK hynix and Samsung are rivals in memory chips. For now, Sapeon uses TSMC as its foundry, whereas Rebellions collaborates with Samsung Foundry.

(Photo credit: Rebellions)