In this report, the analyst analyzes the new 800G-and-above technologies and assesses their market potential. The coverage of this report is of the three main development programs in this area. The most conventional is to create an 800G pluggable transceiver which will be the natural successor 400G. This project has been taken on (somewhat paradoxically) by what used be the 25G Ethernet group, as well as by a largely Chinese group who plan to deliver an 800 MSA.
The other two research programs are much more radical. Specifically, they dispose with pluggability and deliver high data rates on a board. The point here is to move the ASIC and the optics closer to each other to reduce power consumption, the major factor constraining high data rate networking at the present time. One strategy is on-board optics, which migrates the optics from the pluggable transceiver to a board. The other version is co-packaged optics, in which the optics and the chip share a package.
Both these board approaches have existed in simple forms for more than a decade - often in connection with supercomputers or the largest routers. They have proved difficult to implement and the markets that can implement them for this kind networking have proved stubbornly niche-like. Indeed, there would no need for this kind of embedded optics if it wasn’t for the speed at which bandwidth is under demand. In not so many years’ time 51.2Tbps switching will demand new ways of interconnecting optics and electronics.
Although the technologies discussed in this report are novel, the analyst provides a quantitative and qualitative forecast of the revenues they are likely to generate, along with how revenues are likely to change under a variety of scenarios. We also discuss the challenges that networking at 800G face with device complexity and size, power consumption, and thermals. Finally, the report profiles the leading vendors and their strategies active in this area, as well as the standards and MSA activity that is emerging the new high-speed networking sector.
Chapter One Data Beyond 400G: Real Need or Science Project?
1.1 Power consumption: Devil in the detail
1.2 Cloud providers: Growing market power
1.2.1 How much power do cloud providers have to shape the hardware market
1.3 Edge computing and its impact on bandwidth demand
1.4 COVID-19 and the bandwidth explosion
1.5 Need for reduced latency
1.5.1 Technology improvements
1.5.3 Silicon photonics
1.5.4 Single-photon optics
1.6 Manufacturing issues
1.7 Possible scenarios and forecasts for data networking at 800G and above
1.8 Key points in this Chapter
Chapter Two Next-generation Ethernet: 800G and 1.6T
2.1 The Ethernet Technology Consortium’s 800G standard: An assessment
2.1.1 Proposed technology, standards and its proponents
2.2 Proposal for 800G MSA: China’s entry in 800G
2.3 The path to Terabit networking
2.4 Scenarios, forecasts and timetable for pluggable 800G
2.4 Key points in this Chapter
Chapter Three On-Board Optics: Whither COBO?
3.1 The philosophy of on-board optics
3.2 COBO and its future
3.2.1 The COBO Module: Products to date
3.2.2 Related switching technology: Products to date
3.3 Scenarios, forecasts and timetables for on-board products
3.4 Key points in this Chapter
Chapter Four Is Co-Packaging The Next Big Thing?
4.1 The Co-Packaged Optics Collaboration: Goals and roadmap
4.1.1 Proposed technical solution
4.2 Likely vendor participants in the co-packaged future
4.3 Manufacturing and packaging issues
4.4 Scenarios, forecasts and timetables for co-packaged products
4.5 Key points in this Chapter
Chapter Five: Vendor Strategies for 800G and Above
This report states that the market opportunity for 800 Gbps transceivers (800G) and above will reach $245 million in revenues by 2025 with ramp up to $2.5 billion by 2029. Driven initially by video, the analyst expects the traffic in data centers will be further driven by 5G and IoT applications requiring data center infrastructure to be rebuilt using 800G interfaces.
The analyst sees 800G being used for interbuilding connectivity but will become a necessity for data centers that make significant use of 200G servers. The analyst also sees some pioneer efforts to build the first few public network 800G links using technology that is somewhat proprietary to the equipment makers.
According to Lawrence Gasman, author of this new 800 Gbps transceivers report and President, ”800G represents a new era in optical networking speeds and latencies to accommodate the substantial uptick in video conferencing, streaming and digital entertainment as well new applications including virtual reality, augmented reality and artificial intelligence services.”
About the report:
The main objective of this 800 Gbps transceivers report is to assess the market potential of optical networking operating at 800G and above. The report is especially focused on technologies coming out of the efforts to build high-speed interfaces based on pluggable optics, on-board optics and co-packaged optics. The analyst compared the approaches, discuss their viability and construct roadmaps for each technology.
Another key objective of this report is to better understanding of the factors driving the need for networking in this space. It looks at the impact of video/streaming media, which has grown to an even greater extent because of COVID-19. 5G and IoT seem certain future bandwidth hogs, but what about virtual reality (VR), augmented reality (AR) and AI? The analyst notes that the advent of the 51.2T switching chips will be a key enabler for switching gear with 800G and above ports.
The report also analyzes current efforts to deploy 800G in the public network. Here the drivers are almost identical; video, 5G and so on, although 5G is more emphasized. But the players are quite different.
Although the technologies we discuss in this report are novel, it provides a quantitative and qualitative forecast of the shipments and revenues they are likely to generate, under different (high-end and low-end). We also provide an appendix to the report providing profiles of leading firms active in this space. Among the companies discussed in this report are Broadcom, Ciena, Cisco, Facebook, Huawei, IBM, Infinera, Intel, Mellanox, Microsoft, Ranovus and Samtec.
From the Report
Commercial 800G modules will be available in the next couple of years. This means that the 800G “revolution” is an event we can expect to happen soon. Previous generations of modules have been 10x efforts – 10G to 100 G. 800G is
can be implemented quite quickly because it can be achieved in effect be gluing 400G modules together.
The 800G public network market will never be a large one. Here transceivers will be marketed as part of larger systems packages. They will be proprietary so that the equipment companies can squeeze as many features as possible out of their boxes in order to impress the large telcos, which is not an easy thing to do. In the context of public networks, the companies that matter in terms of 800G trials are Ciena, Huawei, and Infinera. But given political realities, Huawei’s market will largely be limited to China and countries in China’s sphere of influence.
A few companies some singling out in the 800G and above. Facebook and Microsoft are important as providing both a testbed for the most advanced transceiver technology in this area as well as leadership for their technology development through organizations such as COBO. Cisco because of its ability to shape the entire datacom market. The analyst also believes that IBM, Intel and a few other firms are important because they have extensive and growing experience of co-packaged/integrated optics which will eventually help Terabit networking to emerge
Most engineers would like to see the pluggable paradigm preserved as long as possible and a major effort has been launched to make this happen. Unfortunately, it doesn’t seem as if pluggability can survive past 1.6T. In the long run, co-packaged optics may dominate the large data center - but almost all the technology for co-packaging is still being developed. In the middle is on-board optics, which may turn out to be an interim technology