Search

Development of Silicon Photonic Micro-Transceiver Chip

AIO Core is a Japanese manufacturer of an optical multi-channel micro-transceiver chip based on silicon photonics technology. In this article, Ivan Nikitski, EPIC’s Photonics Technology Program Manager, talks to Hidetaka Fukuda, CEO about how he came to co-found this forward-thinking company.

After graduating with a MSc in Materials Science from the Tokyo University in 1980, Hidetaka joined Japan’s Ministry of Economy, Trade, and Industry (MITI), where he was responsible for technology and industrial policy for strategic industries including semiconductors, electronics, telecommunication equipment, and flat panel displays. In 1998, he went to San Francisco as Deputy Director of the Japan External Trade Organization (JETRO), a Japanese government-related organization promoting mutually beneficial trade and investment relations. In 2001, he returned to Japan and from 2003-2005 was Managing Director of Information and Communications Equipment Division at MITI. Following his retirement from MITI, in 2008, he became a as Special Advisor (Value Creation) at Silver Lake, a US private equity firm and one of the world’s largest investors in high-tech. As a member of Silver Lake’s team in Asia, he advised on a range of business development, professional networking, investment analysis, and value creation activities.

AIO Core

In 2017, Hidetaka was invited to be a co-founder of AIO Core, a start-up launched by PETRA, the Japanese Photonics Electronics Technology Research Association. The aim was to manufacture and sell an optical multi-channel micro-transceiver chip based on silicon photonics technology, the ‘Optical I/O Core’, which had been developed by PETRA and NEDO, Japan’s national research and development agency. He initially joined the company as a board member and CSO (Chief Strategic Officer) and was appointed CEO in June 2021.

Hidetaka was attracted to AIO Core because he saw it as an opportunity to show the world that Japan was capable of producing cutting-edge silicon-based technology that could penetrate worldwide markets. As he explains, over the previous two decades, despite a large amount of government funding, the Japanese electronics industry had been losing ground particularly to China and Korea, and in the field of computer technology it was 5 years behind the US. The main problem was a conservative and old-fashioned culture prevalent in the industry, among both senior management and engineers especially in the big corporations. What was needed was a more gutsy approach and to be with unique technology that was not dependent on Chinese manufacturers.

AIO Core’s technology

The Optical I/O Core is the world’s smallest (5mm x 5mm) 100 Gbps Optical micro-transceiver that integrates optical transmission and receiver functions onto a single chip and is designed for short distance interconnections over multimode fibres. It’s basically an optical building block that can be flexibly combined with optical components to form the basis of pluggable transceiver module as co-packaged/near-packaged optical modules.

The device is focused on computing applications where low power consumption and low latency are important. For this reason, in addition to 25 Gbps products, they are developing 32, 50 Gbps NRZ and 64.4 Gbps PAM4 products. A Quantum Dot Fabry-Perot laser is used as the light source, which exhibits low temperature dependence in its L-I characteristics and can provide sufficient optical output even at very high temperatures. Also, as the reflected return light noise is very small, the device can be used over a wide temperature range from -40 °C to 125 °C as junction temperature with extremely high reliability.

One of the key features of the I/O Core is the optical pin structure, which couples light from the I/O Core transmitters to the multimode fibre (MMF) interface, and from the MMF interface to receivers in the I/O Core. The optical pins on the transmitter side are designed to confine the light emitted from the grating couplers in their cores, even when the emission angle varies. This allows stable optical connections to be maintained even at temperatures to 125 °C as Junction temperature.

Finally, they are developing an automated assembly machine, that can mount optical fibres passively onto the Optical I/O Core. Passive automated alignment is made possible by the increased misalignment tolerance afforded by the optical pins and the MMFs. By using lower cost, lower precision MMF components instead of higher precision single mode fibres, the I/O Core is thereby cheaper compared with competing single mode devices currently on the market.

Markets

As AIO Core only manufactures the core of the transceiver, not the transceiver itself, their customers are those companies who build complete transceiver modules and systems, who are based mainly in Japan and the US, and increasingly in Europe. With a workforce of around 30 engineers, they have a production capacity ready now, currently shipping demonstrators for verification by potential customers. As Hidetaka points out, there is a huge potential for the I/O Core market not only in HPC, AI and machine learning but in field application like 5G, various kinds of machine including medical, semi, robot, aircraft, automobile, space etc.

Future challenges

For Hidetaka understand that an excellent company must have both an excellent supply chain and an excellent demand chain. So far, AIO Core has built up a supply chain firmly. Now it’s ready for manufacturing. Creating an excellent demand chain is more difficult task. Fortunately, several big transceiver module makers are developing ones with Optical I/O Core and the more shows interest in adopting Optical I/O Core. Actually, some blue-chip system companies are starting to evaluate transceivers with Optical I/O Core. As a start-up, AIO Core should accelerate grabbing customers, otherwise, the delay in accessing them makes it difficult to manage the supply chain as well as expectations on the part of both employees and investors.

A related issue is that many of AIO Core’s engineers are researchers from traditional manufacturing companies, in part lacking knowledge and experience of how to market the technology and how to network at an international level.  Mindset changing to understanding the importance of selling and adapting to the customer’s needs has been proceeding and improving steadily by confronting customers’ requests.

A third challenge is investment. Although Hidetaka has been successful in raising investment due to his long career in the semiconductor industry, finding backers for AIO Core has been tough due to the fact that VCs in Japan tend to shy away from deep-tech in favour of quick return ventures like IT service and AI solution only in Japan.

On a more positive note, Hidetaka is confident that with its key features of ultra-compactness, robustness in wide operating temperature, high reliability, low power consumption, the I/O Core will eventually become a sought-after product in worldwide markets. Not only is the device applicable for high performance computing like AI, ML in data centre but for 5G, medical equipment, semi, aircraft, auto, and space industry because of its requirement of robustness and high reliability in harsh environments.

What’s your advice to the next generation of entrepreneurs?

“I often think that in Japan, start-ups like ours are an endangered species. In the last 30 years, no high-tech start-up has succeeded in marketing deep-tech technology to a worldwide market from Japan. As I mentioned before, Japan’s conservative corporate and industrial culture prefer to maintain the status quo, which makes it virtually impossible for small companies to challenge their business and technological skills.

Another problem for start-ups is that many VCs in Japan are reluctant to support deep-tech as they prefer to fund quick return projects like IT service. For this reason, I believe that the Japanese government should set up funding programs to help small companies develop deep-tech as the EU does in Europe.

Finally, and this is advice to the photonics industry in general, although photonics has a great future, from my experience in the semiconductor industry, I think that the market should be expanding further in a broad field. For example, optical transceivers and related components for computing is huge emerging market and needs different way of usage and technologies. One innovates and tackles it persistently and practically has a chance to take it.

up