FTTH stands for Fiber to the Home: an access network in which optical fiber runs from the operator's central office all the way to the subscriber's premises, where an optical network terminal converts light back into the electrical signals that feed the router, the set-top box and the Wi-Fi access point. Nothing in the last mile is copper.
That single architectural decision is why FTTH has become the default target for telecom operators, ISPs, municipalities and network integrators worldwide. The global FTTH market was valued at USD 56.0 billion in 2024 and is projected to reach USD 68.3 billion in 2026, according to Grand View Research. Knowledge Sourcing models growth of roughly 12.37% CAGR to USD 115.90 billion by 2031 from USD 64.70 billion in 2026. Estimates differ between research houses; the direction does not.
"FTTx" describes how far fiber reaches before copper takes over. FTTN stops at a street cabinet and runs VDSL over existing copper; FTTC stops within a few hundred metres of the home; FTTB stops at the building riser and relies on in-building wiring; FTTA backhauls radio sites. FTTH and FTTP terminate fiber inside the premises itself — no copper segment exists. Every step closer to the subscriber removes a bandwidth ceiling, and FTTH removes all of them.
Bandwidth headroom without re-digging. Fiber's capacity is limited by the transceivers at each end, not by the glass. Operators upgrading from GPON to XGS-PON reused the same outside plant, and the move to 50G-PON will be the same. Trenching, ducting and aerial strand are paid for once.
Symmetrical service. Video calls, cloud backup and remote work depend on upstream capacity. Copper access is inherently asymmetric; PON scales both directions more evenly.
Reach without powered field electronics. A passive optical network typically spans about 20 km from the OLT, with extended-reach designs going further. There is no powered equipment in the field to power, cool or repair.
Lower operating cost. Fewer active elements means fewer faults, fewer truck rolls and less electricity, and fiber is immune to electromagnetic interference and the corrosion that degrades copper.
A 25-year asset. Properly installed outside plant commonly serves a quarter of a century — far longer than the electronics it carries — and delivers consistently low latency and jitter under load.
An FTTH network is three domains: the active equipment at each end, and the passive optical distribution network (ODN) in between.
The OLT sits in the central office or a street cabinet. It aggregates subscriber traffic onto the upstream network, manages PON framing, dynamic bandwidth allocation and encryption, and provides the optical ports that feed the ODN. Port counts range from a handful in a small deployment to hundreds of PON ports in a metro head-end.
The ODN is the passive infrastructure, and it determines both the cost and the quality of the deployment. It normally uses three cable segments: feeder cable (high-count, from the OLT to the first splitting point or distribution hub), distribution cable (mid-count, from the hub to the access points serving a street or building), and drop cable (the final span into each home, usually 1- or 2-core in 2.0 × 3.0 mm flat construction with bend-insensitive G.657 fiber).
The ODN also contains every passive component that terminates, splits, protects and interconnects those cables:
At the subscriber end, an optical network terminal (ONT) receives downstream light, transmits upstream, and presents Ethernet, voice and sometimes RF video ports to the customer's equipment. Modern ONTs are increasingly integrated into the residential gateway or Wi-Fi router.
Nearly all FTTH networks today are passive optical networks: one OLT port is shared among many subscribers by time-division multiplexing, with the splitter deciding who is physically reachable.
The practical implication for buyers: the passive layer installed today must survive two or three PON generations. Wide-band splitters, low-loss connectors, clean splices and generous loss margin are not extras — they are what makes the next upgrade possible.
Point-to-point gives every subscriber a dedicated fiber and suits business parks and high-value enterprise connections, but consumes enormous fiber counts and OLT ports. Point-to-multipoint with passive splitters is the economic default for residential FTTH.
Splitting can be centralised (one 1×32 splitter in the distribution hub) or distributed (cascaded 1×2s and 1×4s placed progressively closer to subscribers). Distributed splitting uses less cable and lets operators build as take-up grows, at the cost of more field enclosures and splice points; centralised splitting is simpler to manage and easier to test.
Split ratio trades directly against reach and loss budget. Typical PLC splitter insertion loss, including excess loss, is roughly:
| Split ratio | Typical insertion loss |
|---|---|
| 1:2 | ~3.6 dB |
| 1:4 | ~7.3 dB |
| 1:8 | ~10.5 dB |
| 1:16 | ~13.7 dB |
| 1:32 | ~16.9 dB |
| 1:64 | ~20.5 dB |
Those numbers explain why 1:32 is the common residential choice on GPON, and why 1:64 requires either a larger loss-budget class or a shorter reach.
Every FTTH design starts with a power budget: available budget must exceed total loss, with margin left over. On the budget side, GPON defines power-budget classes — Class B+ at 28 dB and Class C+ at 32 dB are the two most common — while XGS-PON defines its own classes with budgets ranging from roughly 29 dB to 35 dB. Higher classes cost more in optics but permit longer spans and higher split ratios.
On the loss side, the contributors are fiber attenuation (approximately 0.35 dB/km at 1310 nm and 0.25 dB/km at 1490 nm in standard single-mode fiber), splice loss (typically ≤0.1 dB per fusion splice), connector loss (typically ≤0.3 dB per mated pair for quality APC connectors), and splitter insertion loss.
Worked example. A GPON Class B+ link (28 dB) with a 1×32 splitter, 20 km of fiber, six splices and four mated connector pairs: 16.9 dB splitter + 7.0 dB fiber + 0.6 dB splices + 1.2 dB connectors = 25.7 dB, leaving roughly 2.3 dB of margin.
That margin is thin. Ageing, temperature-driven loss changes, imperfect splices and dirty connectors all consume it — which is why component quality is a design parameter, not a purchasing detail.
Most FTTH cost overruns and field faults trace back to passive components rather than electronics. Specify:
Four instruments cover most field work. An OTDR characterises spans, locates breaks and identifies macrobends and bad splices. An optical loss test set proves end-to-end loss against the design budget. A visual fault locator finds tight bends and breaks close to the technician. A fiber inspection scope prevents the most common cause of intermittent faults: a contaminated connector end face.
The recurring failure modes are unglamorous: dirty connectors, macrobends where cable was dressed too tightly, splices outside the accepted loss window, mismatched APC and UPC mating, and split ratios that consumed the entire loss budget.
10G is the new baseline. XGS-PON is no longer an upgrade project; it is the default platform for new builds in many markets.
50G-PON is moving from trial to commercial. With the ITU-T G.9804 series complete, operators are planning coexistence overlays rather than forklift replacements.
Rural buildout continues. Publicly funded programs across Europe, North America and emerging markets are extending fiber into areas previously considered uneconomic.
Home networks are catching up. Wi-Fi 7 and multi-gigabit Ethernet are making a 1 Gbps access line the bottleneck in a typical household.
Density and outside-plant innovation. Higher-density ODFs, MPO-based cross-connects, pushable fiber, micro-duct and pre-connectorised terminals all reduce labour content per subscriber.
FTTH is a 25-year infrastructure decision carrying a five-year technology cycle. The electronics at both ends will be replaced; the fiber, the splices, the enclosures and the splitters will not. Get the passive layer right — realistic loss budget, honest component specifications, disciplined installation and proper test records — and every future PON upgrade becomes a card swap instead of a rebuild.