cd /news/ai-infrastructure/the-race-to-1-6t-ethernet-and-cohere… · home topics ai-infrastructure article
[ARTICLE · art-126892] src=networkworld.com ↗ pub= topic=ai-infrastructure verified=true sentiment=↑ positive

The race to 1.6T: Ethernet and coherent optics tackle AI’s bandwidth crunch

The Ethernet Alliance will demonstrate 1.6 terabit Ethernet switches, modules, and cables from more than a dozen vendors at ECOC in Malaga, Spain, later this month, while the Optical Internetworking Forum publishes its 1600ZR Implementation Agreement for a coherent interface carrying a full 1.6T Ethernet connection over a single wavelength. IEEE P802.3dj, the specification defining 1.6T Ethernet, is in balloting and formal review and could be finalized later this year, and direct-detect 1.6T products are already shipping to a stable draft covering distances up to about 2 kilometers over single-mode fiber. "There's a lot of people out there that are doing 1.6 terabit-per-second capacity, and that's not the same thing as 1.6 terabit Ethernet," said John D'Ambrosia, chair of the IEEE P802.3dj Task Force.

read5 min views2 publishedSep 11, 2026

Networks have always been driven by the need for more speed and bandwidth, and in the AI era, that drive is accelerating the path to 1.6 terabit network connectivity.

Work on 1.6T began in 2020 and became a formal IEEE study group in 2021, before AI was the industry’s dominant discussion point. That work became the IEEE P802.3dj Task Force, which has been building out the specifications that are now in the balloting and formal review process before finalization, which could happen later this year.

While the standard is being ratified, the demand is already present, and that’s manifesting as real hardware work, not just abstract roadmaps. Later this month, the Ethernet Alliance is putting 1.6T switches, modules, and cables from more than a dozen vendors on the same show floor at ECOC (European Conference on Optical Communications) in Malaga, Spain, to see if any of it actually talks to each other. At the same event, the Optical Internetworking Forum (OIF) will be highlighting its publication of the Implementation Agreement for 1600ZR, a coherent interface built to carry a full 1.6 terabit Ethernet connection over a single wavelength, across the kind of distance that separates buildings on a campus or spans a metro area. The goal of the demonstration is to show that the technology isn’t hype, it’s real. “That’s always the hope of the presentation and the display, that it’s not just technical mumbo jumbo on a sleep-inducing IEEE specification,” David Rodgers, president and chair of the Ethernet Alliance, told Network World. “It’s actually live. It’s happening.”

Not everything marketed as 1.6T means the same thing. A switch port built on eight lanes can be split into different combinations of port speeds rather than a single 1.6T Ethernet connection.

“There’s a lot of people out there that are doing 1.6 terabit-per-second capacity, and that’s not the same thing as 1.6 terabit Ethernet,” John D’Ambrosia, chair of the IEEE P802.3dj Task Force and a founder of the Ethernet Alliance, told Network World. “You can imagine doing a quad 400 gig through a 1.6T port. That’s not 1.6 terabit Ethernet. That’s four ports of 400 gig.”

IEEE P802.3dj, the specification that defines 1.6T Ethernet, is now in its final stretch. Vendors typically wait for that balloting stage before committing to volume production, since committing too early risks building to a standard that could still change.

“I just think that the biggest challenge that we’re going to have is to not underestimate what the demand is going to look like,” D’Ambrosia said.

D’Ambrosia said that a shift toward volume production has already started, with vendors now bringing a wider range of 1.6T products to market.

Direct-detect 1.6T is already shipping. Vendors are building to a draft of IEEE P802.3dj that is stable enough during balloting to commit to, covering distances up to about 2 kilometers over single-mode fiber, D’Ambrosia said.

The coherent line itself is not something IEEE’s specification defines. Coherent optics are more complex to build, but they let a signal travel much farther than simpler optics can. Karl Gass, OIF Physical and Link Layer Working Group Optical Vice Chair, explained to Network World that none of the existing direct-detect options reach the 80- to 120-kilometer range that coherent links need. OIF’s newly published 1600ZR Implementation Agreement closes that gap, doubling the capacity of its earlier 800ZR specification and carrying a full 1.6T Ethernet client over one coherent wavelength.

Products built to the new spec are only now reaching interoperability testing at ECOC, and deployment will need new pluggable modules that draw more power than direct-detect optics, though they plug into the same switches and routers.

IEEE and OIF are coordinating rather than competing. Initial 1600ZR implementations will use the 1.6TAUI-8 chip to module interface that IEEE defines, Gass said, with OIF’s work focused on the data center interconnect piece.

Understanding what 1.6T actually is turns out to be the easier part. Getting a room full of competing vendors to agree on how it interoperates is another problem entirely, and that is where the Ethernet Alliance comes in. Rodgers said the current cycle is being shaped by two groups moving at different speeds.

“There’s the pioneering track, which the AI and the hyperscalers have really taken, and they’re kind of doing what they can with what they have today,” Rodgers said. “And then there’s the Ethernet track, which is we want interoperability across the board, whether it be from an interconnect company or switch or a server or a storage array.”

Testing itself has gotten harder at this speed. The specification now requires chip-to-module training alongside module-to-module training, with separate inner and outer forward error correction that adds more combinations vendors need to validate against each other. Rodgers said the margin for error shrinks as speed increases.

“A pebble in the roadway at 10 miles per hour doesn’t do anything, a pebble in the roadway at 100 miles an hour sends the car off course,” he said.

OIF is already working on what comes after 1600ZR. Gass said 1600ZR+ will extend reach to 1,000 kilometers, alongside a shorter-reach specification called 1600CL, or coherent lite, aimed at roughly 10-kilometer WDM links.

On the IEEE side, D’Ambrosia said planning for 3.2 terabit Ethernet is expected to begin in the first quarter of 2027.

Asked whether the industry will look back on 2027 as the year 1.6T arrived, Rodgers offered a historical parallel. “In my opinion, we will see 1.6T by this time next year,” Rodgers said. “We should see a vast improvement in the standard and the products available that are compliant with the standard. And if history is any indicator, we had the same kind of toe dipping in the water with 800 gig, and now 800 gig is ubiquitous.”

D’Ambrosia has a different forecast: “I’m looking at it a little differently. 1.6T is already here, the stuff is actually already being deployed and we can see this,” he said. “I’ve looked at the number of forecasts. ’27 is a ramp up year, and ’27 and ’28 are just going to be unbelievable.”

── more in #ai-infrastructure 4 stories · sorted by recency
── more on @ethernet alliance 3 stories trending now
sponsored brought to you by zahid.host 4,200+ EU-deployed projects
reading about agents? ship yours in a single git push.

Run your AI side-project on zahid.host

EU-based hosting, git-push deploys, automatic HTTPS, no cold starts. Free tier with a custom domain — perfect for shipping the agent you just read about.

$git push zahid main
Live at https://your-agent.zahid.host
Get free account → Pricing
from €0/mo · no card required
LIVE [news/the-race-to-1-6t-eth…] indexed:0 read:5min 2026-09-11 ·