Silicon photonics, a technology that combines silicon-based integrated circuits( ICs ) and optical components to process and transmit enormous amounts of data more efficiently, is seeing an increase in investment and an acceleration of competition due to the large language models that enable generative artificial intelligence ( AI ).
NVIDIA, TSMC, Intel, IBM, Cisco Systems, Huawei, NTT, and imec, the Interuniversity Microelectronics Centre with its headquarters in Belgium, are just a few of the best designers and manufacturers of ICs, AI systems and communications technology that have entered the race.
These and other companies have been developing silicon photonics for a long time, some of them( including Intel and NTT ) for almost 20 years.
Douglas Yu, the vice president of TSMC in charge of pathfinding for technique integration, told the Nikkei paper on September 5 that” if we can give a good silicon optics integration structure, we will be able to handle both important issues of energy efficiency and computing power for AI.” This will mark the beginning of a new paradigm change. We might be in the midst of a brand-new time.
Due to its high speed, high-speed data transfer, long transmission range, low power consumption, and applicability to advanced networking, computing architectures, cloud computing, data centers, autonomous vehicles, smart transportation systems, semi( the microelectronics industry association ) noted in its invitation to the Summit that” Silicon Photonics has become a popular buzzword in the semiconductor industry.”
In other words, golden photonics will gain the high-tech sector directly across the cutting edge in a way that will enhance system and system performance while consuming less energy.
The global silicon photonics market, with a compound annual growth rate ( CAGR ) of 25.7 % from its valuation of$ 1.26 billion in 2022, is expected to reach’ a value of US$ 7.86 billion by 2030 ,’ according to semi. ……………………………………
A joint R & amp, D project called COUPE, which stands for Compact Universal Photonic Engine, was introduced by NVIDIA and TSMC in September 2022, according to DigiTimes. The project’s objective is to use silicon photonic ( SiPh ) technology from NVIDIA to combine various AI processors ( GPUs ).
” Source revealed ,” according to DigiTimes,” that SiPh chip and CMOS processes go through co-packaged optics ( CPO ) technology integration, which can connect multiple advanced GPUs with chip-on-wafer – on-substrate ( CoWoS ) 2.5D IC packaging.”
It is possible to produce” an ultra – large GPU set” thanks to the combination of low-latency optical data transfer and cutting-edge package technology, according to DigiTimes.
The fact that ASE Technology, the top Outsourced Semiconductor Assembly and Test ( OSAT ) company in Taiwan, and Advantest, a Japanese company, were both prominent at the Silicon Photonics Global Summit explains why this technology might not be ready” until the SiPh ecosystem matures.”
Following the golden photonics occasion, TSMC announced to the media at the Semicon Taiwan industry show that a lack of CoWoS packaging capacity is likely to limit its ability to provide NVIDIA with AI processors until the end of the following year. The capacity may therefore double by then.
The golden integrated circuits and the silicon laser are two of the most significant 20th-century inventions, according to Intel’s definition of silicon photonics. By utilizing the efficiency of Intel’s high-volume golden production, it enables faster information transfer over longer distances than conventional devices.
The visual interfaces for Ethernet switches, routers, and carry marketing tools, providing connectivity for large-scale cloud and business data centers, are visual transceivers, according to Intel, which combines” the power of optics and the scalability of silicon.”
” Intel manufactures silicon photonics at the largest scale in the world ,” according to SemiAnalysis’ Dylan Patel. They dominate the market for producing visual communication transceivers thanks to Intel’s size and the integrated nature of their solution, which has caused them to have a failure rate that is two orders of magnitude higher than that of the competition.
By 2025, Intel intends to triple its capacity for superior Circuit package, including a fresh 3D presentation facility in Malaysia. ” Malaysia will eventually become Intel’s largest production base for 3D chip packaging ,” according to corporate vice president for manufacturing supply chain and operations at Intel, Robin Martin, told Tech Wire Asia in late August.
Intel is introducing the technology to others in addition to increasing its own production of visual transceivers and another silicone photonics products. China’s FAST Photonics Technologies announced plans to produce visual transmitters based on Intel systems in March.
Shenzhen is home to the shop and offices of Fast Photonics. Additionally, it has a sales and technical assistance facility in San Jose, California, close to Intel’s Santa Clara offices.
Japan’s national telecoms provider and provider of telecommunications technologies, NTT, announced plans to produce a number of increasingly sophisticated photonics-electronics convergence devices over the coming years in May 2023.
These innovations should pave the way for a significant decrease in the power consumption of data centers and telecommunications networks because they do away with the need to change signals from visual to electrical and back again.
By 2030, NTT hopes to achieve an increase of 100 times in energy efficiency, a 125-times increase in transmission capacity, and an end-to-end delay( latency ) of 200 times.
The Innovative Optical and Wireless Network Global Forum has been established by NTT, along with Intel and Sony, with the goal of” accelerating the implementation of a new communication system that will bring together an all-photonics network infrastructure including golden photonic, edge technology, and wireless distributed computing.”
Silicon photonics were added by Huawei and IEC to their joint study on visual information website technology in 2014. That came after Huawei purchased Caliopa, a golden photonics visual transceiver developer created by imec and Ghent University, the past year.
However, after Huawei was placed on the Entity List by the US Commerce Department and ASML’s EUV printing system shipments to China were outlawed in 2019, the partnership between Huawei and IEC was terminated. Imec and ASML have a nearby connection.
Huawei has continued to conduct independent study, which is crucial for both its telecoms equipment business and its efforts to avoid US government sanctions. ” Huawei intends to produce a photonic chip in order to circumvent American chip restrictions ,” the statement reads in the video that was posted on YouTube in October 2022.
It continues by stating that nanostructured chips outperform silicon-based ICs, which are almost impossible to produce at process nodes below 2nm. Additionally, the manufacturing process for photon chips is entirely different; it will use brand-new manufacturing technology and won’t even need a lithography machine.
The film concludes by saying that the research in this area is still in its infancy and that monopolies have not yet been established in European and American nations.
This evaluation seems to predict at least ten years in the future. Huawei’s intentions shouldn’t be disregarded, despite the fact that it is positive but realistic and that the 5G Mate 60 Pro smartphone recently debuted with a 7nm cpu in spite of US sanctions.
Huawei is Creating a Secret Network for Cards was the subject of an article published by Bloomberg on August 23. The US Semiconductor Industry Association( SIA) has issued a warning that Huawei has purchased at least two factories and is constructing three more, with funding from the Chinese government totaling$ 30 billion, according to the article.
It would be foolish to assume that Huawei is doing this merely to transfer common semiconductor manufacturing equipment under various names if the report is accurate. China and Huawei have every reason to make significant investments in golden optoelectronics.
@ ScottFo83517667 is the author’s Twitter account.