With the increasing demand for massive computing in fields such as AI, communication, and autonomous vehicles, the evolution of integrated circuits (ICs) has reached a physical limit under the premise of Moore’s Law. How can this limit be surpassed? The answer lies in the realm of optics. Currently, many domestic and international companies are actively embracing “Silicon Photonics” technology. When electronics meet photons, it not only addresses the signal transmission loss issue but is also considered a key technology that could usher in a new era, potentially revolutionizing the future world.

Integrated circuits (ICs) cram millions of transistors onto a single chip, performing various complex calculations. Silicon Photonics, on the other hand, represents integrated “light” paths, where light-conductive pathways are consolidated. In simple terms, it is a technology that converts “electronic signals” into “optical signals” on a silicon platform, facilitating the transmission of both electrical and optical signals.

As technology rapidly advances and computer processing speeds increase, communication between chips has become a critical factor in computing performance. For instance, when ChatGPT was first launched, there were issues with lag and interruptions during the question and answer process, which were related to data transmission problems. Therefore, as AI technology continues to evolve, maintaining computational speed is a crucial aspect of embracing the AI era.

Silicon Photonics has the potential to enhance the speed of optoelectronic transmission, addressing the signal loss and heat issues associated with copper wiring in current computer components. Consequently, semiconductor giants such as TSMC and Intel have already invested in related research and development efforts. In this context, we interviewed Dr. Fang Yen Hsiang, director of the Opto-Electronics Micro Device & System Application Division and Electronic and Optoelectronic System Research Laboratories at the Industrial Technology Research Institute (ITRI), to gain insights into this critical technology.

What Is the Relationship Between Silicon Photonics and Optical Transceivers?

An optical transceiver module comprises various components, including optical receivers, amplifiers, modulators, and more. In the past, these components were individually scattered on a PCB (printed circuit board). However, to reduce power consumption, increase data transmission speed, and minimize transmission loss and signal delay, these components have been integrated into a single silicon chip. Fang emphasizes that this integration is the core of Silicon Photonics.

Integrated Circuits’ Next Step: The Three Stages of Silicon Photonics

• Silicon Photonics Stage 1: Upgrading from Traditional Pluggable Modules

Silicon Photonics has been quietly developing for over 20 years. The traditional Silicon Photonics pluggable optical transceiver modules look very much like USB interfaces and connect to two optical fibers—one for incoming and one for outgoing light. However, the electrical transmission path in pluggable modules had a long distance before reaching the switch inside the server. This resulted in significant signal loss at high speeds. To minimize this loss, Silicon Photonics components have been moved closer to the server’s switch, shortening the electrical transmission path. Consequently, the original pluggable modules now only contain optical fibers.

This approach aligns with the actively developing “Co-Packaged Optics” (CPO) technology in the industry. The main idea is to assemble electronic integrated circuits (EIC) and photonic integrated circuits (PIC) onto the same substrate, creating a co-packaged board that integrates chips and modules. This co-packaging, known as CPO light engines (depicted in figure “d” below), replaces optical transceivers and brings optical engines closer to CPU/GPU chips (depicted in figure “d” as chips). This reduces transmission paths, minimizes transmission loss, and reduces signal delay.

According to ITRI, this technology reduces costs, increases data transmission by over 8 times, provides more than 30 times the computing power, and saves 50% in power consumption. However, the integration of chipsets is still a work in progress, and refining CPO technology will be the next important step in the development of Silicon Photonics.

• Solving the CPO Bottleneck and Beyond – Silicon Photonics Stage 2: Addressing CPU/GPU Transmission Issues

Currently, Silicon Photonics primarily addresses the signal delay challenges of plug-in modules. As technology progresses, the next stage will involve solving the electrical signal transmission issues between CPUs and GPUs. Academics point out that chip-to-chip communication is primarily based on electrical signals. Therefore, the next step is to enable internal chip-to-chip communication between GPUs and CPUs using optical waveguides, converting all electrical signals into optical signals to accelerate AI computations and address the current computational bottleneck.

• Silicon Photonics Ultimate Stage 3: The Arrival of the All-Optical Network (AON) Era

As technology advances even further, we will usher in the era of the “All-Optical Network” (AON). This means that all chip-to-chip communication will rely on optical signals, including random storage, transmission, switching, and processing, all of which will be transmitted as optical signals. Japan has already been actively implementing Silicon Photonics in preparation for the full transition to all-optical networks in this context.

Where Does Silicon Photonics Currently Face Technological Challenges?

Currently, Silicon Photonics faces several challenges related to component integration. First and foremost is the issue of communication. Dr. Fang Yen Hsiang provides an example: semiconductor manufacturers understand electronic processes, but because the performance of photonic components is sensitive to factors such as temperature and path length, and because linewidth and spacing have a significant impact on optical signal transmission, a communication platform is needed. This platform would provide design specifications, materials, parameters, and other information to facilitate communication between electronic and photonic manufacturers.

Furthermore, Silicon Photonics is currently being applied in niche markets, and various packaging processes and material standards are still being established. Most of the wafer foundries that provide Silicon Photonics chip fabrication belong to the realm of customized services and may not be suitable for use by other customers. The lack of a unified platform could hinder the development of Silicon Photonics technology.

In addition to the lack of a common platform, high manufacturing costs, integrated light sources, component performance, material compatibility, thermal effects, and reliability are also challenges in Silicon Photonics manufacturing processes. With ongoing technological progress and innovation, it is expected that these bottlenecks will be overcome in the coming years to a decade.

This article is from TechNews, a collaborative media partner of TrendForce.

(Photo credit: Google)

According to a report by Taiwanese media Money DJ, after establishing a stable position as a major supplier for NVIDIA GPU baseboard, Wistron has secured orders for AMD MI300 baseboards. Reliable sources indicate that Wistron has expanded its involvement beyond AMD baseboards and entered the module assembling segment.

In addition to NVIDIA and AMD, Wistron has also entered the Intel AI chip module and baseboard supply chain, encompassing orders from the three major AI chip manufacturers.

The NVIDIA AI server supply chain includes GPU modules, GPU baseboards, motherboards, server systems, complete server cabinets, and more. Wistron holds a significant share in GPU baseboard supply and is also involved in server system assembly.

Currently, NVIDIA commands a 70% market share in AI chips, but various chip manufacturers are eager to compete. Both AMD and Intel have introduced corresponding solutions. While Wistron was previously rumored to have entered AMD baseboard supply, it has also ventured into AMD GPU module assembling, serving as the sole source, according to reliable sources.

Regarding the news of Wistron’s involvement in AMD and Intel chip manufacturing, the company has chosen not to respond to market rumors.

(Photo credit: Google)

According to a report by Taiwan’s Economic Daily, AI is driving a massive demand for data transmission, and silicon photonics and Co-Packaged Optics (CPO) have become new focal points in the industry. TSMC is actively entering this field and is rumored to be collaborating with major customers such as Broadcom and NVIDIA to jointly develop these technologies. The earliest large orders are expected to come in the second half of next year.

TSMC has already assembled a research and development team of over 200 people, aiming to seize the business opportunities in the emerging market of ultra-high-speed computing chips based on silicon photonics, which are expected to arrive gradually starting next year.

Regarding these rumors, TSMC has stated that they do not comment on customer and product situations. However, TSMC has a high regard for silicon photonics technology. TSMC Vice President Douglas Yu recently stated publicly, “If we can provide a good silicon photonics integration system, it can address two key issues: energy efficiency and AI computing capability. This could be a paradigm shift. We may be at the beginning of a new era.”

Silicon photonics was a hot topic at the recent SEMICON Taiwan 2023 with major semiconductor giants like TSMC and ASE giving related keynote speeches. This surge in interest is mainly due to the proliferation of AI applications, which have raised questions about how to make data transmission faster and achieve signal latency reduction. The traditional method of using electricity for signal transmission no longer meets the demands, and silicon photonics, which converts electricity into faster optical transmission, has become the highly anticipated next-generation technology to enhance high-volume data transmission speeds in the industry.

Industry reports suggest that TSMC is currently collaborating with major customers like Broadcom and NVIDIA to develop new products in the field of silicon photonics and Co-Packaged Optics. The manufacturing process technology ranges from 45 nanometers to 7 nanometers, and with mass production slated for 2025. At that time, it is expected to bring new business opportunities to TSMC.

Industry sources reveal that TSMC has already organized a research and development team of approximately 200 people. In the future, silicon photonics is expected to be incorporated into CPU, GPU, and other computing processes. By changing from electronic transmission lines to faster optical transmission internally, computing capabilities are expected to increase several tens of times compared to existing processors. Currently, this technology is still in the research and academic paper stage, but the industry has high hopes that it will become a new driver of explosive growth for TSMC’s operations in the coming years.

(Photo credit: Google)

In the realm of specifications competition, desktop computers continue to possess numerous irreplaceable advantages. These include ease of upgrading, superior heat dissipation capabilities, and robust and durable construction, resulting in extended usage lifespans. As a result, desktop computers maintain a steadfast market demand. Due to the ease of component replacement in desktops, expandability remains a significant advantage for PC gamers. For creators and business professionals, desktop computers satisfy extensive external connectivity needs while offering superior heat dissipation. Furthermore, owing to the size limitations of laptops, desktop computers continue to provide a more comfortable user experience during prolonged usage.

Windows 10 Exit and Hardware Updates Set to Drive 2024 Upgrade Trend

In the latter half of 2022, brands and retailers aggressively cleared their inventories, a trend that continued into 2023, resulting in a sustained challenging period for the PC market. In recent years, the PC market has approached saturation, making it difficult to drive market growth through sheer quantity. Consequently, brand manufacturers have focused on business, gaming, and creator products. However, PCs inherently belong to a cyclical terminal market. With the Windows 10 operating system set to retire in October 2025 and Windows 11’s heightened hardware specifications requirements, products released before 2017 will require replacements. Additionally, it is anticipated that companies like Intel, AMD, and NVIDIA will gradually unveil new products in the latter half of 2023. This, coupled with the demands of the new operating system, is expected to trigger a noticeable upgrade trend among consumers, ultimately providing a glimmer of hope for the PC market. (Image credit: Unsplash_Alienwaregaming)

In its FY2Q24 earnings report for 2023, NVIDIA disclosed that the U.S. government had imposed controls on its AI chips destined for the Middle East. However, on August 31, 2023, the U.S. Department of Commerce stated that they had “not prohibited the sale of chips to the Middle East” and declined to comment on whether new requirements had been imposed on specific U.S. companies. Both NVIDIA and AMD have not responded to this issue.

TrendForce’s analysis:

• Close ties between Middle Eastern countries and China raise U.S. concerns:

In NVIDIA’s FY2Q24 earnings report, it mentioned, “During the second quarter of fiscal year 2024, the USG informed us of an additional licensing requirement for a subset of A100 and H100 products destined to certain customers and other regions, including some countries in the Middle East.” It is speculated that the U.S. is trying to prevent high-speed AI chips from flowing into the Chinese market via the Middle East. This has led to controls on the export of AI chips to the Middle East.

Since August 2022, the U.S. has imposed controls on NVIDIA A100, H100, AMD MI100, MI200, and other AI-related GPUs, restricting the export of AI chips with bidirectional transfer rates exceeding 600GB/s to China. Saudi Arabia had already signed a strategic partnership with China in 2022 for cooperation in the digital economy sector, including AI, advanced computing, and quantum computing technologies. Additionally, the United Arab Emirates has expressed interest in AI cooperation with China. There have been recent reports of Saudi Arabia heavily acquiring NVIDIA’s AI chips, which has raised concerns in the U.S.

• Huawei is expected to release AI chips comparable to NVIDIA A100 in the second half of 2024; competition is yet to be observed:

Affected by U.S. sanctions, Chinese companies are vigorously developing AI chips. iFlytek is planning to launch a new general-purpose LLM (Large Language Model) in October 2023, and the AI chip Ascend 910B, co-developed with Huawei, is expected to hit the market in the second half of 2024, with performance claimed to rival that of NVIDIA A100. In fact, Huawei had already introduced the Ascend 910, which matched the performance of NVIDIA’s V100, in 2019. Considering Huawei’s Kirin 9000s, featured in the flagship smartphone Mate 60 Pro released in August 2023, it is highly likely that Huawei can produce products with performance comparable to A100.

However, it’s important to note that the A100 was already announced by NVIDIA in 2020. This means that even if Huawei successfully launches a new AI chip, it will already be four years behind NVIDIA. Given the expected 7nm process for Huawei’s Ascend 910B and NVIDIA’s plan to release the 3nm process-based Blackwell architecture GPU B100 in the second half of 2024, Huawei will also lag behind by two generations in chip fabrication technology. With the parameters of LLM doubling annually, the competitiveness of Huawei’s new AI chip remains to be observed.

• China remains NVIDIA’s dominion in the short term:

Despite the active development of AI chips by Chinese IC design house, NVIDIA’s AI chips remain the preferred choice for training LLM models among Chinese cloud companies. Looking at the revenue performance of the leading Chinese AI chip company, Cambricon, its revenue for the first half of 2023 was only CNY 114 million, a YoY decrease of 34%. While being added to the U.S. Entity List was a major reason for the revenue decline, NVIDIA’s dominance in the vast Chinese AI market is also a contributing factor. It is estimated that NVIDIA’s market share in the Chinese GPU market for AI training exceeded 95% in the first half of 2023. In fact, in the second quarter of 2023, the China market accounted for 20-25% of NVIDIA’s Data Center segment revenue.

The main reason for this is that the Chinese AI ecosystem is still quite fragmented and challenging to compete with NVIDIA’s CUDA ecosystem. Therefore, Chinese companies are actively engaged in software development. However, building a sufficiently attractive ecosystem to lure Chinese CSPs in the short term remains quite challenging. Consequently, it is expected that NVIDIA will continue to dominate the Chinese market for the next 2-3 years.

(Photo credit: NVIDIA)