While all eyes are on Intel’s restructuring plan, which features the foundry unit’s spin-off and the delay of Germany and Poland factories, there is another critical decision regarding its photonics business.

According to Intel’s announcement, the tech giant is moving Integrated Photonics Solutions (IPS) into its Data Center and Artificial Intelligence division (DCAI), as it tries to drive a more focused R&D plan that’s fully aligned with its top business priorities.

This shuffle seems to be reasonable, as earlier this year, Intel has achieved a milestone in integrated photonics technology for high-speed data transmission, and the two arenas seem to be inseparable.

A few months ago, Intel demonstrated the industry’s most advanced and first-ever fully integrated optical compute interconnect (OCI) chiplet co-packaged with an Intel CPU and running live data. According to Intel, the OCI chiplet represents a leap forward in high-bandwidth interconnect by enabling co-packaged optical input/output (I/O) in emerging AI infrastructure for data centers and high performance computing (HPC) applications.

A report by Photonics Spectra notes that Intel’s IPS division focuses on technologies such as light generation, amplification, detection, modulation, CMOS interface circuits, and package integration.

Here’s why this technology matters: As chipmakers advance Moore’s Law, increasing transistor density, signal loss during transmission becomes a significant issue because chips use electricity to transmit signals. Silicon photonics technology addresses this problem by using optical signals instead of electrical ones, allowing for high-speed data transmission, greater bandwidth, and faster data processing.

Intel has been developing silicon photonics technology for over 30 years. Since the launch of its silicon photonics platform in 2016, Intel has shipped over 8 million photonic integrated circuits (PICs) and more than 3.2 million integrated on-chip lasers, according to its press release. These products have been adopted by numerous large-scale cloud service providers.

In addition to Intel, rivals such as AMD and TSMC are also accelerating the development of next-generation silicon photonic solution.

Read more

• [News] Venturing into Silicon Photonics, AMD Reportedly Seeks Partnerships in Taiwan
• [News] TSMC’s Latest Advancements in CFET, 3D Stacking, and Silicon Photonics

(Photo credit: Intel)

On September 9, 2024, China’s Ministry of Industry and Information Technology (MIIT) released the “Guiding Catalog for the Promotion and Application of Major Technical Equipment (2024 Edition),” listing 286 domestically manufactured equipment items. Among these, the KrF and ArF lithography machine, categorized under electronic specialty equipment, have attracted significant market attention.

According to TrendForce’s analysis, the catalog aims to promote the adoption of domestic equipment in key manufacturing sectors. This is achieved through subsidies for domestic equipment to lower the threshold for companies adopting these tools. The catalog will be updated every 2-3 years to align with the latest policy and industry developments.

With the Netherlands expanding its restrictions on ASML’s exposure equipment exports, the disclosure of two domestic lithography machines by China’s Ministry of Industry and Information Technology carries symbolic significance. The KrF Lithography Machine (model 2.1.5) is designed for 12-inch wafers, with a working wavelength of 248nm, resolution ≤ 110nm, and overlay accuracy ≤ 25nm. The ArF Lithography Machine (model 2.1.6) is also for 12-inch wafers, with a working wavelength of 193nm, resolution ≤ 65nm, and overlay accuracy ≤ 8nm. Both machines fall under the category of deep ultraviolet (DUV) lithography equipment.

Despite initial market speculation that the ArF lithography machine might be suitable for 8nm processes, TrendForce’s latest market analysis suggests otherwise. The machine’s overlay accuracy of ≤ 8nm is inadequate for advanced processes, which require ≤ 3nm for 10nm processes and ≤ 2nm for 7nm processes. Additionally, for advanced processes, the resolution must be ≤ 38nm, while this machine’s resolution is ≤ 65nm, making it challenging to handle even 40nm processes.

Although the ArF lithography machine could potentially be used for 55nm or more advanced processes with multiple exposures, its insufficient overlay accuracy would lead to unmanageable yield issues, significantly impacting production efficiency.

Read more

• [News] Breakthrough in China’s Domestic DUV Equipment to 8nm? Experts Call It a Misunderstanding
• [News] China’s DUV Breakthrough: Capable of Producing 8-Nanometer Chips

(Photo credit: SMIC)

In early September, rumors have it that TSMC’s first US fab in Arizona began producing engineering wafers using the 4nm process in April, with yields reportedly comparable to those manufactured in its Southern Taiwan Science Park facility. Now here’s the latest update: the fab has started trial production for Apple’s A16 chip, according to a report by Tim Culpan at substack, a technology columnist.

Tim Culpan notes that the mobile processors are manufactured with TSMC’s 5nm, or the so-called N4P node, which is the same as the node used in Taiwan to manufacture A16. The N4P node is actually a member of the 5nm family, as it is regarded as an enhanced version of 5nm, the report explains.

It is worth noting that Apple’s A16 SoC, though launched two years ago with iPhone 14 Pro, is considered as one of the most advanced mobile chips for the company, as the chip is also be seen in iPhone 15 and iPhone 15 Plus models. Culpan indicates that the move marks a milestone that instead of beginning with some less critical chips, Apple and TSMC intend to aim high from the start.

According to Culpan, Apple’s A16 is currently being trial-produced at TSMC Arizona’s “Fab 21” Phase 1 facility, with a small production volume. However, once the second stage of the Phase 1 fab is completed, the output will significantly increase.

TSMC plans to build three plants in Arizona, each with cleanroom spaces twice the size of typical logic fabs in the industry. The first fab is expected to begin mass production in the first half of 2025.

TSMC’s second fab in Arizona will use 2nm process technology to meet strong AI-related demand, with production expected to begin in 2028. The third fab will employ 2nm or even more advanced process.

However, the situation for Samsung’s investment in the U.S. would be a different story. A previous report from Korean media outlet Business Korea noted that persistent issues with its 2nm yield rate have led Samsung to decide to withdraw personnel from its Taylor, Texas plant, signaling another setback for its advanced wafer foundry business.

As for Intel, which proactively pursues the support of the U.S. government, it is holding steadfast on its investments in the country despite recent announcements to spin-off its foundry business and delaying the projects in Germany and Poland for two years.

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.

Read more

• [News] Showing Positive Progress, TSMC’s Arizona Fab Reportedly Reaches Production Yields on Par with Taiwan
• [News] TSMC Receives USD 6.6 Billion for Chipmaking from the US Government, Commits to Constructing Third Fab

(Photo credit: Apple)

ByteDance, the parent company of TikTok, is said to be collaborating with TSMC, eyeing for the mass production of two self-developed AI chips by 2026, according to reports by Economic Daily News and The Information.

ByteDance’s AI chips are expected to be made with TSMC’s 5nm node, which would be one generation behind the foundry giant’s most advanced process, the reports suggest, making the move comply with the U.S. export regulations to China. The chips are similar to NVIDIA’s next-generation flagship AI chip, Blackwell, which are manufactured with TSMC’s 4NP node.

Citing sources familiar with the matter, the reports note that the tech giant in China aims to reduce its reliance on NVIDIA for AI model development. Though the chips are still in the design phase and the plan is subject to change, ByteDance’s self-designed chips could save billions of dollars compared to purchasing NVIDIA’s products, according to the reports.

The Information estimates that ByteDance’s spending on developing generative AI models has been increasing, and it is rumored that the company has ordered over 200,000 NVIDIA H20 chips this year, costing it over USD 2 billion, with some orders still pending delivery.

In response to US export bans, NVIDIA launched AI chip H20, L20 and L2, specially designed for the Chinese market earlier this year. According to a previous report by Wccftech, H20 GPU has 41% fewer Cores and 28% lower performance versus H100. Still, the product is reportedly seeing strong demand for AI servers among Chinese Cloud Service Providers (CSPs) and enterprises, including Huawei and Tencent.

However, due to its lower computing power, Chinese companies need to purchase more H20 chips to build clusters with equivalent computing capacity, which raises costs, Economic Daily News notes.

According to TSMC’s financial report in the second quarter, North American clients contributed 65% of its total revenue. While China, the second-largest market, contributed 16% of its quarterly revenue, with a significant jump from 9% in the first quarter and 12% during the same period last year.

Read more

• [News] China Plans to Set at Least 50 AI Standards by 2026, from Language Models to Semiconductors
• [News] Broadcom Reportedly Developing 5nm AI Chip with ByteDance, Manufactured by TSMC

(Photo credit: ByteDance)

According to a recent Financial Times report, the US and Japan are nearing an agreement to restrict tech exports to China’s chip industry, despite Tokyo’s concerns over potential retaliation from Beijing. The White House aims to introduce new export controls ahead of the November presidential election, including rules requiring non-US companies to obtain licenses before selling products that could bolster China’s tech sector.

Citing the Financial Times, US officials have been engaged in extensive discussions with Japan and the Netherlands to align export control policies, aiming to prevent Japanese and Dutch companies from being affected by the US “foreign direct product rule.” While sources in Washington and Tokyo indicate that a breakthrough is near, a Japanese official expressed caution, noting that the situation remains delicate due to concerns over potential Chinese retaliation.

To mitigate the impact of any retaliatory measures, Japan and the US are discussing strategies, as Washington and its allies attempt to curb China’s technological advancements. The Biden administration’s export controls are designed to close loopholes and address China’s rapid progress, particularly from companies like Huawei, in chip manufacturing over the past two years.

The new restrictions are intended to make it more difficult for China to access critical chipmaking equipment, which would heavily affect companies like ASML in the Netherlands and Tokyo Electron in Japan. Washington is also pushing for restrictions on servicing these tools, including software updates and maintenance, which would further harm China’s chip ambitions. These measures would mirror those already imposed on US firms and citizens.

Should the new restrictions come into effect, Japanese equipment manufacturers could face significant repercussions. Tokyo Electron, for instance, derived nearly 50% of its revenue in Q1 of its 2025 fiscal year from China, up from 39.3% in the same period a year earlier.

Other major equipment suppliers have also seen a surge in revenue from China. In Q2 2024, ASML reported that China was its largest market, accounting for 49% of lithography unit sales, surpassing South Korea (28%) and Taiwan (11%). Applied Materials recorded 43% of its sales from China in Q2, up 22 percentage points year-over-year. Meanwhile, China contributed 39% of Lam Research’s Q2 revenue in 2024, compared to 26% a year earlier, and 44% of KLA’s revenue in the same quarter.

Read more

• [News] ASML Claims the World Counts on China for Legacy Chips; So Do Chip Equipment Giants

(Photo credit: ASML)