JASON KOSHY OPTICAL TRANSCEIVERS
Optical transceiver design is changing as higher performance is demanded in every aspect of operation and capabilities, but how do they remain compatible with existing network infrastructure? Optical Connections editor Peter Dykes spoke with Jason Koshy , Vice President, Global Sales at U.S.-based Infinite Electronics to find out the latest developments. ADAPTING TO THE NETWORK OPTICAL TRANSCEIVER DESIGN
last summer, even though they were not yet available in the market. Starting at the end of last year, we started seeing 1.6T solutions with deliveries in the next three to six months. It’s interesting because usually, you spend a lot of time releasing one technology, and then you talk about the next one. Today, advancements in technology and product innovation are driving solutions to the market for adoption, with the next technology projected to be released in six months versus years.
Are design changes to optical transceiver design also impacting on other network equipment? It depends. When you look at the telecom industry, it’s moving much slower than data centres,
We’re also seeing a growing need for greater temperature tolerance. Internet service providers’ equipment is located outdoors to support fiber to the home (FTTH), and can see very high temperatures, especially during the summer months. Data centers require more power to support network equipment growth and more processing, which generate more heat. As a result, transceivers are now being designed to use less power and have submersion capabilities. In addition, transceivers need to carry more data than they used to. That said, the form factors are not changing much. There is a lot of conversation about CPO (co-packaged optics) technology, but I think CPO solutions are still years away. CPOs have limited flexibility when it comes to servicing, single-point-of-failure concerns, and no option for infield recoding of transceivers. With CPO, if there’s something wrong with the transceiver at the board level, you’re going to have to replace the whole thing. As with all advancements, there are always trade-offs, and it’s going to really depend on what trade-offs network engineers decide they can live with. A high priority for data centres is the need for reliability. Given what’s happening in data centres these days, the requirement for uptime is becoming a much greater concern than before. What could bring your network down? What could cause a rack to fail? Those are considerations that engineers look at when considering new technology. Technology shifts can be seen at trade shows, which are a great place to see vendors’ latest solutions and future technology road maps. We started seeing 800G solutions
PD
JK
for example, because the demand for data from AI, autonomous vehicles and streaming services is drastically driving the need for faster speeds. These processing needs are causing a rapid evolution in transceiver speed capabilities. The introduction of a new generation of transceivers, which once took around three years, is now happening in less than a year. We were doing 100G just a few years ago, then we jumped to 400G, and 800G came out last year. Now, we’re about to release 1.6T into the market this summer. That evolution has been very drastic, and so you’re seeing network equipment essentially being replaced to enable the accommodation of faster speeds. A lot of that is driven by AI, and you’re seeing the impact of Nvidia’s accelerated processing solutions on some of the technology that has been developed with its processing chips. A broad range of things are happening in the market that are really causing that technology to be introduced in a faster timeframe.
Are there limitations when buying transceivers from OEMs?
PD
The problem is the reliability. If a transceiver goes down, you’ve got somebody unhappy somewhere.
JK
What we’ve seen with some of our customers is that there’s a higher failure rate because transceivers are not the core of what they do, and they don’t want to keep inventory. OEM transceivers are more expensive than third-party transceivers. Most operators are looking for redundancy, uptime and backup, which means they need a lot of products if something goes down. With our transceivers, we can code them into a different platform very quickly. Network operators are not going to save on the CapEx of buying new equipment, but there are areas in OpEx and long-term ownership of the network that can be managed in a more cost- efficient way. We can code in the field with our smart coder, as well as access transceivers remotely from anywhere and see their full history. These are things that can’t be done with OEM transceivers, and that’s why our customers include some of the largest telecom companies
Will these advances impact on the form factors of transceivers?
PD
Not necessarily. We’re really talking about what transceivers are capable of doing. While data
JK
centres don’t require transmission over great distances, the telecom industry requires faster speeds and longer distances, in some cases 120 kilometres.
6
www.opticalconnectionsnews.com
| ISSUE 41 | Q2 2025
Made with FlippingBook interactive PDF creator