The IEEE 802.3 Roadmap, Read Honestly

The IEEE 802.3 roadmap PDF that gets passed around the industry every year is a useful document. It is also an aspirational document, and the gap between what it says and what actually ratifies has been widening since around 2023.

I read every revision of this roadmap. I cross-reference against the actual IEEE working group schedules, the OIF interoperability profiles, and what vendors are sampling. The honest version of the timeline is meaningfully different from the version on the slide.

This post is the honest version, for procurement teams trying to plan 2027-2028 deployments against assumptions that may or may not hold.

What the roadmap shows

The IEEE 802.3 roadmap, as of mid-2026, anticipates ratifications across the rest of the decade for:

• 400GBase variants and reach extensions — mostly already ratified, with refinements continuing.
• 800GBase across multiple reach classes — primary ratifications in 2024-2025, secondary variants continuing through 2026.
• 1.6TBase (also called 1.6T Ethernet) — primary ratification targeted for 2025-2026, slipping to 2026-2027 in the latest revisions.
• 3.2TBase — targeted for late decade, currently in early study group phase.

The dates as published are the dates IEEE thinks the standard might ratify. Vendor implementation typically lags ratification by 18-30 months for volume products.

What's actually slipped

Two ratifications have slipped meaningfully in the last 24 months.

1.6T's full ratification dates. Original projection was 2025. Current projection is late 2026 with some reach classes pushed into 2027. The slip is about a year on the primary spec and 12-18 months on some of the reach variants.

OIF 800ZR+ profile finalisation. Not strictly an IEEE document, but adjacent enough that it shows up in customer planning. The 800ZR+ profile work has slipped twice and is currently expected to be finalised in mid-2027, not late 2026.

The slip isn't bad behaviour on anyone's part. It's the standard pattern of multi-stakeholder standards work hitting more disagreement than the optimistic projection assumed.

Where the slip matters for procurement

Three concrete consequences.

Vendor "production-volume" 1.6T products won't ship in volume before late 2027, possibly 2028. Vendors are sampling 1.6T silicon now. Sample availability does not mean volume. Volume requires the standard to be stable enough that all vendors implement the same interpretation, and that's downstream of ratification by 12+ months.

800ZR+ optics aren't reaching multi-vendor interop ratification before late 2027. OIF profile finalisation drives this. Optics that interop across vendors in production reach class are gated on the profile work. Single-vendor 800ZR+ shipments are happening; multi-vendor interop is a year out.

Plans built around 1.6T deployment in 2027 are at meaningful risk. A network that committed to 1.6T uplinks in 2027 in their 2025 planning cycle is depending on multi-vendor 1.6T volume by Q3 2027. The slip math doesn't support that today.

How to read the next roadmap revision

When IEEE publishes the next revision of the roadmap in late 2026, three signals tell you whether the timeline is firming up or slipping further.

Working group meeting attendance and contribution rate. If the 1.6T working group is producing contributions at a consistent rate and convergence on issues, that's a positive signal. If issues are accumulating without resolution, that's negative.

Number of unresolved technical issues at the latest plenary. The agenda from each plenary lists open issues. A shrinking list at each meeting is convergence. A stable or growing list is divergence.

Vendor public statements about volume timing. When all the relevant vendors start using the same approximate timing in their public communications, the timing has firmed up. When vendors are still using ranges or hedged language, it hasn't.

A reader who tracks these three signals quarterly can have a better forecast than the published roadmap itself.

The structural reasons standards slip

It's worth understanding why this happens so you can apply the right discount to whatever the next aspirational date is.

Working group dynamics. Each new generation of Ethernet has more participants and more disagreement than the last. The 100G ratification was contentious; 400G was more contentious; 800G was more again; 1.6T continues that trend. Reaching rough consensus takes longer with more parties at the table.

Concurrent dependencies. 1.6T depends on PMD specifications that depend on signalling specifications that depend on FEC profile decisions. When one of those slips, the rest cascade. Most of the slip in 1.6T traces back to a single FEC profile decision that took longer than projected.

Vendor strategic interest. Vendors who lead a generation push for fast ratification. Vendors who are behind push for slower ratification to give them catch-up time. The working group politics reflect these strategic interests. The committee chair's tactical patience determines how visibly this plays out.

Standards body resource model. IEEE working groups are largely volunteer. Volunteers have day jobs. Multi-year standards projects compete for volunteer attention with everything else. The bandwidth ceiling is real.

None of these slow down by themselves. The 1.6T process is at the edge of what the current resource model can sustain. The next generation, 3.2T, is going to test this even harder.

What 2027 procurement should actually assume

Three default assumptions for procurement plans landing in 2027 that depend on this roadmap.

Default to "800G is the production ceiling" for any 2027 deployment. 800G is well-understood, multi-vendor, and shipping at volume. Anything 1.6T or above is at meaningful timing risk.

Build optionality into 2028 plans. A 2028 plan should be able to land on either 800G (if 1.6T slips further) or 1.6T (if it lands in 2027). The cost of building optionality is small. The cost of being locked to one number when the other is what materialises is large.

Verify volume claims independently. When a vendor tells you "we'll ship 1.6T in volume by Q4 2027", the question to ask is "what's your unit projection for that quarter and what's the SKU mix". If they don't have an answer, the claim is aspirational. If they do, the number probably reflects their honest view.

The genuine question I'd love to be wrong about

Is there any path that ratifies 1.6T cleanly in 2026 and gets volume shipments by mid-2027? I don't see it. The unresolved issues at the latest plenary still need work, the FEC profile decision is still settling, and vendor sampling lags ratification by twelve to eighteen months even when everything goes well.

I'd be happy to be wrong. The networks running large AI fabrics are running into the limits of 800G aggregation right now. 1.6T showing up faster would be welcome.

The honest read says it won't. Plan accordingly. The IEEE roadmap is a useful directional document and a poor scheduling tool. The procurement teams who treat it as the latter consistently miss the actual ship dates.

Read the roadmap. Apply the slip math. Plan for one generation behind the published target. You'll be wrong some of the time, right most of the time, and never embarrassed by a slip you didn't see coming.