800G coherent ships, 1.6T enters the deployment window

For most of the last five years, the marketing layer of optical networking has been ahead of the deployment layer. Vendors announced 800G line-side coherent in 2021. They announced 1.6T pluggables in 2023. By the time those announcements made it into a carrier's actual capacity plan, the numbers in the slide deck had aged out.

That gap is finally closing. LightCounting — the most cited optical market analyst — published a May 2026 update forecasting that 800G transceiver shipments will more than double in 2026, while 1.6T port shipments will rise from a small 2025 base to tens of millions of units. 1.6T chipset sales are expected to surpass $2 billion this year and grow rapidly through 2029.

For a regional DWDM operator, that single forecast carries four practical consequences.

1. 100G ZR+ is finally the floor, not the ceiling

When OpenZR+ consolidated around the 400G coherent pluggable form factor in 2022, the deployment economics of long-haul DWDM started to invert. A pluggable that fits into a router slot and runs over an alien wavelength has a fundamentally different cost stack from a dedicated transponder shelf.

In 2026 the question is no longer whether to use 400G ZR/ZR+ — that ship has sailed. The question is whether to design a new build for 400G as the channel unit with 800G as the upgrade path, or to over-build to 800G ZR from day one. With 800G pluggables now shipping in volume, the second option becomes the cheaper one on a five-year amortisation for any link above roughly 600 km.

2. The C+L band debate has a different answer

C+L band amplification roughly doubles a fibre pair's capacity. Until 2026 the decision was straightforward — C+L is "obviously more efficient" but the optical filters, amplifiers and ROADM line cards cost a meaningful premium.

With 1.6T per wavelength entering the deployment window, C-band-only systems can carry 115.2 Tbps per fibre pair (72 channels × 1.6T) if the photonics line system supports it. That is enough headroom on most regional routes for the next five-to-seven years.

The honest answer for SEE routes in the 600 to 1,500 km class — Sofia–Frankfurt, Sofia–Istanbul, Sofia–Bucharest, Sofia–Belgrade — is that C-band only with a 1.6T-ready optical line system delivers more upgrade headroom than most carriers will exhaust before the next coherent generation. C+L is for hyperscale backbones, not for SEE carrier backbones, in 2026.

3. IP transit pricing has its physical floor coming

The cost per bit at the optical layer is now falling faster than the cost per bit at the routing layer. A 1.6T port in a coherent pluggable, amortised over a five-year DWDM build, will end up under €0.20 per Mbps per month on long-haul European routes by 2027. The IP transit retail price has been falling toward that physical floor for years. With this generation of optics, it gets there.

Wholesale buyers — content networks, ISPs, public-cloud hyperscalers — will keep getting better unit pricing. Carriers who are still selling IP transit on a marketing-driven discount curve will be repriced by physics, not by a competitor. The carriers who win are the ones who already monetise wavelengths, dark fibre and managed services alongside IP transit, because those products have different cost curves.

4. Peering policy gets renegotiated

When per-port cost falls and per-bit volume rises, the cost of carrying settlement-free peering at an IXP also falls. That makes the 3-to-5-year cycle of peering ratio disputes between large eyeball networks and content distributors economically less interesting to fight. We expect a cooling-off period on peering policy litigation through 2027 — and a renewed acceleration of paid-peering offers from Tier-1 carriers who need to monetise their declining transit margins.

For SEE operators, this is the cycle to lock in long-term peering arrangements at DE-CIX Frankfurt, AMS-IX Amsterdam and the regional IX (BIX, NIX.CZ, SOX) — before the next round of policy resets.

The window

The optical hardware cycle is rarely synchronised with a single year. 2026 is unusual because three trends are converging at once: 800G shipping in volume, 1.6T entering deployment, and a sharply rising appetite for diversified routes in the wake of the 2024 Red Sea events.

A network designed in 2026 will be in service through 2031 to 2033. The capacity, the bands and the pluggables chosen this year are the contour lines for the next architecture conversation. We will be conservative on the high-touch decisions — line systems, fibre pair counts, regen sites — and aggressive on the low-touch ones — coherent pluggables, alien wavelength flexibility, automated lambda provisioning.

---

Sources: LightCounting transceiver forecast cited by C-Light; 800G optical transceiver market update, FICER Technology / CommunicAsia 2026; Cignal AI optical research press notes.