NG-PON2 works exactly as designed and lost anyway, because it solved a capacity problem that access networks did not have in the form it offered. The ITU-T G.989 series defines a TWDM-PON that stacks four 10 Gbps wavelength pairs onto one ODN for 40 Gbps downstream and 10 Gbps upstream, with tunable ONUs that can move between wavelengths. The technology is elegant. The commercial problem is that operators wanted symmetric 10G first, and got it from XGS-PON without tunable optics anywhere in the network.
Rayin (Shenzhen Rayin Technology) is a manufacturer of GPON OLTs and industrial Ethernet switches.
NG-PON2 is ITU-T G.989: four wavelength pairs, 40 Gbps downstream, 10 Gbps upstream, tunable ONUs, coexistence with GPON on the same ODN.
XGS-PON (ITU-T G.9807.1) delivers 10 Gbps symmetric with fixed optics, which covered most of the demand at a fraction of the complexity — and won the market.
50G-PON (ITU-T G.9804) then skipped straight past 40G, so NG-PON2 now sits between a standard nobody needed and a successor that arrived before it scaled.
NG-PON2 is the ITU-T G.989 family, and its defining feature is TWDM: time and wavelength division multiplexing. Instead of one wavelength pair carrying a time-shared 10 Gbps, four wavelength pairs each carry a time-shared 10 Gbps, and the OLT assigns subscribers to a wavelength.
| Feature | What G.989 specifies | Why it matters |
|---|---|---|
| Downstream capacity | 40 Gbps, from 4 × 10 Gbps wavelengths | Capacity scales by adding wavelengths, not by faster optics |
| Upstream capacity | 10 Gbps total, from 4 × 2.5 or 4 × 10 Gbps | Upstream stays shared, so the asymmetry remains |
| ONU optics | Tunable, per G.989.2 | The ONU can be moved between wavelengths without a truck roll |
| Wavelength count | Up to 4 pairs in the base system, extensible | More wavelengths mean more capacity per ODN |
| Coexistence | WDM overlay with GPON on the same fibre | A single ODN can serve GPON and NG-PON2 subscribers |
| Reach and split | Comparable to GPON: tens of kilometres, 1:64 or beyond | No new civil works required |
The tunable ONU is the part that distinguishes NG-PON2 from everything before it, and it is also the part that made it expensive. A fixed-wavelength ONU uses a Fabry-Perot or DFB laser chosen at manufacture. A tunable ONU needs a laser whose wavelength can be set in the field, plus a control channel to tell it where to go, plus the calibration to keep it there across temperature.
Three problems, and only one of them turned out to be urgent.
Capacity per ODN without new fibre. A single GPON port delivers 2.488 Gbps downstream and 1.244 Gbps upstream, shared across a 1:64 split. For an operator with a full ODN and no appetite for trenching, adding wavelengths is more attractive than adding fibre.
Wavelength mobility. Because the ONU is tunable, an operator can move subscribers between wavelengths to balance load, or to move a subscriber to a dedicated wavelength for a business service. That is a real operational capability, and no fixed-optics PON has it.
Converged access. One ODN carrying residential, business and mobile backhaul, separated by wavelength rather than by fibre. The PtP WDM overlay option in G.989 exists precisely for this.
What operators actually needed first, in the 2015–2020 window, was more upstream. GPON's 1.244 Gbps upstream was the binding constraint for anything symmetric — business Ethernet, small cell backhaul, cloud backup. NG-PON2 offered 10 Gbps upstream shared across four wavelengths, which is 2.5 Gbps per wavelength in the common configuration. XGS-PON offered 10 Gbps upstream on a single wavelength, symmetric with downstream, with fixed optics.
Because the comparison was never close on cost per bit delivered for the dominant use case.
| Dimension | GPON (G.984) | XGS-PON (G.9807.1) | NG-PON2 (G.989) | 50G-PON (G.9804) |
|---|---|---|---|---|
| Downstream | 2.488 Gbps | 9.95 Gbps | 40 Gbps (4 × 10) | 50 Gbps |
| Upstream | 1.244 Gbps | 9.95 Gbps | 10 Gbps shared | 25 or 12.5 Gbps |
| Symmetry | No | Yes | No, in practice | Partial |
| ONU optics | Fixed | Fixed | Tunable | Fixed, with options |
| OLT optics | One per port | One per port | Four per port plus multiplexing | One per port |
| Coexistence with GPON | — | WDM overlay | WDM overlay | WDM overlay |
| Commercial status | Ubiquitous | Mainstream | Niche deployments | Deploying |
Three forces decided it.
Cost of the tunable ONU. A tunable laser plus control electronics is the single most expensive component in the ONU bill of materials. At scale, XGS-PON ONUs reached commodity pricing years before tunable ONUs did, and the gap never closed.
The 40G step was in the wrong direction. Access networks moved from 2.5G to 10G because upstream demand tripled. The next step from 10G was never going to be 40G asymmetric; it was 25G or 50G, and the industry standardised 50G-PON under G.9804 rather than building on the NG-PON2 stack.
Ecosystem gravity. A PON standard is only as good as the number of OLT vendors shipping it and the number of ONU vendors producing compatible optics. NG-PON2 had credible vendors, but XGS-PON had more, and operators buy from the larger ecosystem because it prices better and interoperates more reliably.
Note: This is not a judgement on the engineering. NG-PON2 deployments exist and work, particularly where an operator needed wavelength-based service separation on an existing ODN. It is a judgement on where the volume went.
Two things, and they are worth knowing before dismissing it.
Wavelength mobility at scale. If a subscriber needs a dedicated wavelength — a mobile backhaul site, a hospital, a business park with a service-level agreement — NG-PON2 can assign one without new fibre. With fixed-optics PONs, the equivalent means a separate port and a separate fibre path.
Incremental capacity on a constrained ODN. Where a 1:64 ODN is fully subscribed and the fibre count cannot be increased, adding wavelengths is the only way to grow without new civil works. That is a narrow case, but it is a real one in dense urban builds and in some wholesale networks.
Both capabilities are now also being addressed by 50G-PON plus WDM overlays, which is the practical reason NG-PON2 did not recover: its differentiators were absorbed into the next standard.
A decision table, on the assumption that the network is being built now. For most of these rows the NG-PON2 vs XGS-PON question is settled before the table is even read: the fixed-optics option wins unless the ODN is out of fibre.
| Situation | Reasonable choice | Reason |
|---|---|---|
| New FTTH, residential, 1 Gbps services | GPON | Cheapest ONU, mature ecosystem, sufficient for the service |
| New FTTH, 1–10 Gbps services, symmetric demand | XGS-PON | Symmetric 10G, commodity optics, coexistence with GPON |
| Business parks, mobile backhaul, symmetric 10G | XGS-PON, or NG-PON2 where wavelength separation is required | XGS-PON unless the ODN is capacity-constrained |
| Existing ODN at capacity, no new fibre | NG-PON2 or a 50G-PON overlay | Wavelength stacking is the only growth path |
| Greenfield with a 10-year horizon | 50G-PON readiness | Avoid building a standard that is already superseded |
| Industrial or campus network, not carrier | GPON, not NG-PON2 | The requirement is reach and simplicity, not 40G |
The last row deserves emphasis. In a factory or a campus, the reason to use PON is passive infrastructure over distance, not headline capacity. A GPON OLT with 1:64 splitting covers a site that would otherwise need a dozen switches and a lot of copper, and the capacity question never arises.
Rayin builds GPON OLTs for exactly that role: the mini GPON OLT line covers 1-port and 2-port models with 1:128 splitting per PON port, sized for edge and campus deployments rather than for a 40 Gbps access network. For an operator weighing NG-PON2 against XGS-PON, the relevant comparison is OLT port economics and ONU availability, and that conversation belongs with vendors who ship both.
Five questions that separate a real proposal from a datasheet copy.
Which G.989 sub-recommendations are implemented? G.989.1 is the general architecture, G.989.2 the PMD layer, G.989.3 the transmission convergence layer. "Compliant with G.989" without sub-references is not a specification.
Is the ONU tunable in the field, and over what range? Tunability is the whole point; a fixed-optics ONU sold as NG-PON2 is a different product.
What is the per-wavelength upstream capacity in the shipped configuration? 2.5G per wavelength and 10G per wavelength are very different products.
How does the OLT handle wavelength assignment — static or dynamic? Wavelength mobility has operational value only if reassignment does not require a maintenance window.
What is the ONU price delta against XGS-PON, and the lead time? If the answer is more than a small multiple, the business case has to rest on something other than bandwidth.
Is NG-PON2 the same as TWDM-PON?
TWDM-PON is the multiplexing technique that NG-PON2 uses: multiple wavelength pairs, each time-shared. NG-PON2 is the ITU-T G.989 standard that specifies it, including the tunable ONU, the coexistence rules and the management model. You can describe TWDM-PON without specifying NG-PON2, but not the reverse.
Why did operators choose XGS-PON over NG-PON2?
Cost and symmetry. XGS-PON delivers 10 Gbps in both directions with fixed optics, which covered the upstream demand that actually existed. NG-PON2 delivered 40 Gbps downstream and 10 Gbps upstream shared across wavelengths, and needed tunable optics in every ONU to do it.
Does NG-PON2 coexist with GPON on the same fibre?
Yes. G.989 defines a WDM overlay so that GPON and NG-PON2 subscribers can share one ODN, in the same way XGS-PON coexists with GPON. Coexistence is one of the stronger arguments for any of these upgrades, because it avoids re-splicing the outside plant.
Is 40G PON still relevant with 50G-PON available?
Largely no for new builds. 50G-PON under ITU-T G.9804 offers higher capacity per wavelength with fixed optics, and the vendor ecosystem is concentrating there. A 40G PON deployment now is defensible only where an existing NG-PON2 estate is being extended.
What is wavelength mobility in NG-PON2?
The ability to move a subscriber from one wavelength to another without changing hardware, because the ONU laser is tunable. It allows load balancing across wavelengths and lets an operator assign a dedicated wavelength to a business or backhaul service on the same ODN.
About the author: Sara — Sara is a Customer Manager at Rayin with over 10 years of experience in the communications field. In her free time, she enjoys badminton and swimming.
About Rayin: Shenzhen Rayin Technology Co., Ltd. — Company Profile
