From transceivers to speed-of-light AI
Since 1985, silicon photonics has progressed from the initial development of high confinement waveguides to strategically incorporate CMOS techniques, establishing its dominance in the transceiver space. In the coming years, it has the potential to expand to a wide range of innovative applications.
BY Martin Vallo, senior analyst in photonics at Yole
Since early 2023, there has been a lot of hype around – and massive investments into – silicon photonics, especially optical computing, optical I/O, and diverse sensing applications. It seems logical that primary technologies in various applications will be replaced relatively quickly by optical-based designs and architectures. The giants forecast that optics will be necessary and become ubiquitous relatively soon, while startups are developing new applications through R&D. So, can we expect this prediction to be realised any time soon?
While there are many arguments about the necessity of photonics coupled with electronics, the largest silicon photonics market – datacom pluggables – generates only around 12 percent of datacom transceiver revenue (projected to reach 30 percent by 2028). The semiconductor market is suffering from an extended period of decline, leading to more pragmatic buying behaviour on the part of customers. DC operators prefer long-established and low-cost technology solutions. Yole Intelligence’s market research shows that silicon photonics is not yet a primary technology, even for intra-data-centre interconnects with up to 500 m reach.
Figure 1:Growth forecast for various silicon photonicsapplications.Credit: Yole Intelligence.
In this context, silicon photonics remains a technology in active development, with a wide array of potential applications, hinting at promising opportunities on the horizon. In the coming decade, frontrunners will emerge, leading to industry consolidation. Nevertheless, the broad spectrum of applications will ensure abundant opportunities for the technology to expand and proliferate.
Yole Group, in its new Silicon Photonics 2023 report, estimated the silicon photonics PIC market was worth US$68 million in 2022 and is forecast to generate more than US$600 million in 2028 at a 44 percent Compound Annual Growth Rate for 2022-2028 (CAGR2022-2028). This growth will mainly be driven by 800G high-data-rate pluggable modules for increased fibre-optic network capacity. Additionally, projections of rapidly growing training dataset sizes show that data will need to use light to scale ML models using optical I/O in ML servers.
Substantial data centre requirements, particularly in the domains of artificial intelligence (AI) and machine learning (ML), are expected to fuel the ongoing demand for swifter applications and computations over the next decade. With the conventional processor-centric computing architecture and copper interconnects, the state-of-the-art chips based on 3nm technology are approaching their physical limitations, while the necessity for faster data transmission has surged. Silicon photonics, with its ability to facilitate high-speed communication, has therefore become a prime focus.
Figure 2: A vast array of potential applications hinting at promising opportunities on the horizon. Credit: Yole Intelligence.
Architectures that include optical I/Os can take out streamline access between compute nodes and memory pools, harnessing the fan-out capabilities of optics to minimise the number of switching hops required to access resources. Broadcom’s strategic plan outlines a trajectory for switching chips, projecting an increase from 51.2 Tb/s (5 nm process node) this year to 102.4 Tb/s (3 nm process node) in 2025, and an impressive 204.8 Tb/s (2 nm process node) by 2027. This exponential growth could serve as a significant catalyst for the advancement of silicon photonics in networking applications, paving the way for significantly enhanced data capacity in the future. Silicon photonics provides a versatile platform for applications with high-volume scalability demands.
The primary and most immediate domain for its application is data centres, where Intel holds a dominant position. A second major high-volume application is telecommunications, as exemplified by Acacia, benefiting from the consistent and superior performance of silicon processing. A third broad application area encompassing optical LiDAR systems has significant potential but faces cost and 2D beam-scanning challenges. 3D integration, housing both chips on the same silicon substrate, is vital for seamless control. Optical gyroscopes need sizeable chips for sensitive rotation sensors, benefiting from silicon substrates and SiN waveguides. Quantum computing is pivotal in the evolving AI and machine learning landscape. Optical computing, ideal for efficiency-focused tasks, garners industry attention and promises substantial impact.
Advanced photonic components and their integration for medical use can transform healthcare, enabling faster, more precise diagnostics, treatment, and patient monitoring. Overcoming regulatory and standardisation challenges may be necessary for clinical adoption. The outlook for silicon photonics-based medical applications is promising and holds significant potential for various healthcare and medical fields. Extending silicon photonics into the visible spectrum shows potential for future developments, offering a vast range of innovative applications.
Figure 3: Intel, Cisco, Marvell... The silicon photonics industry is confident of its future value. Credit: Yole Intelligence.
The silicon photonics industrial landscape is forming around diverse players, including: major vertically integrated players (Intel, Cisco, Marvell, Broadcom, Nvidia, IBM, etc.); actively engaged in the silicon photonics industry; startups and design houses (AyarLabs, OpenLight, Lightmatter, Lightelligence); research institutions (UCSB, Columbia University, Stanford Engineering, MIT, etc.); foundries (GlobalFoundries, Tower Semiconductor, imec, TSMC, etc.); and equipment suppliers (Applied Materials, ASML, Aixtron, etc.). All these players contribute to significant growth and diversification.
Intel is a leader in this field, investing heavily in research and development. There are numerous startups focused on silicon photonics technology, aiming to bring innovation to the market. These startups often focus on specific applications or novel technologies, such as high-speed transceivers, optical interconnects, and LiDAR systems. Universities and research institutions play a crucial role in advancing silicon photonics, often collaborating with industry partners to develop cutting-edge technologies and share knowledge.
Foundries provide silicon photonics services, enabling other companies to manufacture their photonic chips. These foundries often use advanced manufacturing processes, such as CMOS (complementary metal-oxide-semiconductor) technology, to produce these chips. Equipment suppliers provide the tools necessary for manufacturing silicon photonic devices. The quality and precision of these tools are critical for producing high-performance photonic components.
The silicon photonics industry is marked by ongoing research and development, strategic partnerships, and collaboration between players to advance the technology. It is also becoming more accessible to a broader range of companies, thanks to silicon photonics foundries and growing expertise in the field. The technology’s ability to improve data transfer speeds, reduce energy consumption, and enable various applications make it a promising area for industrial growth.
Intel remains the market leader in datacom, with a 61 percent market share in both shipments and revenue, followed by Cisco, Broadcom, and other smaller companies.
With the recent strengthening of product portfolios and the commercialization of PICs by other players, Yole Intelligence expects that Intel will lose its dominant market share. In telecom, Cisco (Acacia) has almost 50 percent market share, followed by Lumentum (Neophotonics) and Marvell (Inphi). The telecom silicon photonics market is driven by coherent pluggable ZR/ZR+ modules.
Intel has recently muddied the water, failing to complete the acquisition of Tower as well as divesting its manufacturing line of silicon photonics-based pluggable modules to Jabil. Intel, which is struggling to regain the lead in chip production technology, hoped the merger with Tower would help accelerate a shift to become a major manufacturer for other chip designers. The failure to complete this acquisition will force it to focus its Intel Foundry Services (IFS) division business strategy solely on leading-edge process technologies. It could also send a further chill through American companies with deep ties in China, as technology is the prime battlefield in the tense economic relations between China and the United States. Intel’s recent strategic decision to offload its manufacturing line to Jabil allows it to optimise its operational efficiency, reduce costs, and leverage Jabil’s expertise to serve its customers better, remain competitive in the market, and boost profitability. Intel is shifting its focus towards the development and production of higher-value components, such as processors and compute platforms, which are integral to forthcoming optical interconnects designed for disaggregated data centres. The company is setting its priorities to concentrate on silicon photonics components that are crucial for emerging sensing applications, for example in the automotive industry or in medical uses.
Silicon photonics: what is happening in China?
Silicon photonics is an advanced technology requiring access to high-level manufacturing skills, which China still lacks. Chinese companies are at the prototyping or sampling level and rely on external partnerships to supply silicon photonics transceivers or optical engines in volume. Skorpios-Luxshare-Broadex and Sicoya-Broadex are good examples of collaborations in datacom. The Chinese telecom players Huawei and ZTE usually purchase PICs from Cisco or Nokia.
In 2014, Huawei and imec added silicon photonics to their joint research on optical data link technology. That followed Huawei’s acquisition of Caliopa, a developer of silicon photonics optical transceivers spun out of imec and Ghent University. Ultimately, the Huawei-imec collaboration was terminated, and shipments of ASML’s EUV lithography systems to China were banned in 2019.
After being put on the US Commerce Department’s Entity List, Huawei has continued with its research, which is essential both for its telecom equipment business and its efforts to escape sanctions imposed by the US government. China has strong motivations to make substantial investments in silicon photonics. Huawei is trying to procure American equipment essential for chip production and other restricted materials and thus uses a covert strategy to circumvent international sanctions.
Figure 4: Revenue market shares for datacom and telecom modules in 2022. Credit: Yole Intelligence.
What is the technology pathway for silicon photonics?
Despite silicon’s shortcomings as a light emitter, recent breakthroughs have introduced innovative approaches to creating active optical components on silicon and have achieved mass production in just a few years. It’s worth noting that silicon’s internal quantum efficiency is low, whereas direct bandgap III–V materials boast an efficiency close to 100 percent. After the success of bonded LEDs (GaAs on GaP) back in the 1990s in high-brightness LED applications, it was anticipated that bonding III–V materials to silicon would also prove highly
effective.
A pivotal collaboration between the UCSB and Intel played a crucial role in resolving manufacturing issues and achieving high-volume production.
The pathway for silicon photonics appears to be monolithic integration through quantum dots (QD). Conventional InP PICs require five or six regrowth steps, which are complex and expensive, and have limited yield. Heterogeneous integration offers the advantage of combining multiple materials, and bonding and processing simultaneously. With this approach, modulators, lasers, and detectors can be bonded side by side and processed together, offering inherent benefits. However, the cost of the substrate is not insignificant, as III–V substrates are considerably smaller than 300 mm, prompting a growing interest in monolithic integration. Therefore, the monolithic integration techniques of on-chip lasers offer a promising approach towards high-density and large-scale silicon photonic integration.
The choice between quantum wells (QW) and quantum dots in monolithic GaAs-on-silicon devices has been a critical issue. After four decades of research, QD lasers have demonstrated intrinsic parameters surpassing QW devices, providing a much longer lifetime. For example, the QD gain medium exhibits a large tolerance to material defects, allowing for the epitaxial integration of QD lasers on silicon, while its fast gain response makes it suitable for amplifying high-speed signals.
Additionally, the QD gain medium’s stability at high temperatures enables uncooled operation, while narrow linewidth lasers, low threshold current density, internal loss, and confinement factor contribute to low noise figure operation. Significant improvements in III-V/silicon epitaxy have pushed QD technology to the frontiers of Si photonics and a wide range of applications. Still, much effort is needed to make this technology ubiquitous and affordable for high-volume, high-performance PICs.
Figure 5: Penetration rate of laser technologies in datacom pluggable modules 2021-2028. Credit: Yole Intelligence.
The realm of silicon photonics is not confined to a single substrate or material. Various material platforms, such as thin film LiNbO3 (TFLN), SiN, BTO, and GaAs, among others, have demonstrated their potential for photonic integration. Among these, TFLN on silicon has made rapid progress. With its tight mode confinement, TFLN has proven invaluable for creating high-speed modulators, comb generators, and a diverse array of devices. Notably, HyperLight has played a pivotal role in advancing this technology with remarkable success.
Waveguides extend beyond silicon and encompass a wide range of materials, including LiNbO3 compound semiconductors, compound semiconductor on insulators (CSOI), SiN, and more. SiN waveguides, for instance, support 980 nm tuneable lasers operating at remarkable temperatures, presenting exceptional possibilities.
Figure 6: Roadmap of silicon photonics integration 1992-2030. Credit: Yole Intelligence.
There is a significant disparity in scale when comparing silicon photonics, which operates at 45 nm, with silicon integrated circuits, which have scaled down to just a few nanometres. Remarkably, silicon photonics doesn’t require 3 nm lithography, as the 45 nm technology is perfectly adequate for producing high-performance, high-quality silicon photonics devices. This is advantageous because employing older foundries with lower lithography levels is very cost-effective.
By utilising 3D bonding to connect the PIC to electronics, which may operate at lithography of 3 nm or beyond, we can harness the strengths of both worlds. Consequently, it doesn’t appear rational to integrate photonics and electronics on the same wafer in the same process flow, as it would increase costs and production timelines. Instead, the more sensible approach is 3D integration, aligning the most advanced electronics with the most advanced photonics.
Silicon photonics has the potential to revolutionise the way data is transmitted and processed, offering benefits in terms of speed, power efficiency, and cost-effectiveness. The technology pathway involves a combination of material science, device engineering, and application development to realise this potential.
