Knowledge Center
Technology Insights · Industry Trends · Product Knowledge · Application Notes · News & Updates
Every optics project starts with the same sentence: the module price is too high. In practice the unit price is a small part of what a network pays for optics. The reach code that is one step above what the route needs, the coded variant bought only because the part number looked familiar, the spare of a 400G module kept at eight sites, a coherent port specified where a fixed wavelength part would do: each of these adds cost that never appears on the module invoice.
This guide sets out the seven levers that actually move optical module cost, what each one changes, and what it does not change, so a procurement decision can be defended in a design review as well as in a budget review.
Seven Levers That Move Optical Module Cost
| Lever | What it changes | What it does not change |
|---|---|---|
| Match the reach code to the measured route | Eliminates amplified and long reach modules on short spans, plus the attenuators and higher power they bring | Link budget margin, which still has to be calculated and kept |
| Standardise on fewer part numbers | Concentrates volume on one SKU per route type, which improves terms and simplifies spares | The routes themselves, which still need the correct reach class |
| Choose the form factor the port needs | Avoids paying for 800G where 100G satisfies the traffic and the chassis | Future upgrade path, provided the chassis has spare higher-rate ports |
| Use fixed wavelength where tunable is not needed | Removes laser tuning and control cost on rings with a fixed plan | Wavelength flexibility, which matters if the ring plan is still changing |
| Buy coded optics from a supplier that manages coding | Avoids platform specific purchases and lets one module type cover several host platforms | Platform behaviour, since the module still has to be coded for each host |
| Plan the wavelength layer | CWDM and DWDM overlays on existing fiber, which delays new cable and new line cards | Fiber count, which is fixed by the duct already in place |
| Plan spares and lead time together | One shared spare per site class instead of a spare per part number at every site | Failure rate, which is a property of the module class and the installation |
None of these levers depends on buying a lower grade of product, only on specifying the product that matches the link.
Where Cost Grows Without Anyone Noticing
Four patterns account for most of the avoidable spend:
- One step too much reach. A 40km or 80km part on a 10km route links only with attenuators and draws more power on every port. The purchase order shows a slightly higher unit price, while the real cost sits in patch panels, spare parts and rack thermal planning.
- Coherent where fixed wavelength is enough. Coherent modules carry real value above 120km or where services share one fiber pair. On a 40km unamplified span with spare channels, a coded amplified pluggable does the job with lower power.
- Part number sprawl. Six similar 10G and 25G part numbers that differ only by a vendor code fragment split the volume and multiply the spares list. Standardising on one coded part per route type usually buys more than any single negotiation.
- Spares planned per site instead of per failure. A spare pool sized for the installed base, not for each cabinet, removes working capital without touching availability.
What Must Not Be Cut
Three items look like savings and are not:
- Link budget margin. Removing the margin to justify a cheaper reach class moves the risk into commissioning, where a marginal span becomes a truck roll.
- Diagnostics. DDM and DOM readings are what turn a site visit into a remote diagnosis. The cost of one avoided call-out is larger than the difference between module grades.
- Temperature class. Outdoor and unheated sites need the industrial temperature range. A commercial module in that position is a warranty claim waiting to happen.
How to Run the Exercise
Cost reduction works best as a specification review, not as a price negotiation. Take the BOM, and for every line ask:
- What is the measured route length and fiber type for this span?
- Which reach code does that route actually require?
- What rate will the service carry in three years, and does the host chassis have a higher-rate port if that changes?
- Is there a shared wavelength plan that would let two services use one fiber pair?
- How many distinct part numbers are on the sheet, and how many of them differ only by coding or a reach step?
- How many spares are held, where are they held, and what failure are they covering?
The output is a specification change, not a discount request, and it survives the next design review because every line still matches the link it serves.
What APEX Supports
APEX quotes across the full rate ladder, so a specification review can be answered with modules rather than opinions. That includes the 1.25G and 2.5G SFP range for access and legacy links, the 25G SFP28 range for 5G transport and 25G server links, the 100G QSFP28 range for metro and DCI, and the 400G and 800G QSFP-DD and OSFP families for AI fabric and core links. Single-fiber BiDi, CWDM and DWDM versions are available in the same families, and modules are coded for the platform in use so one part type can cover more than one host.
For teams replacing or consolidating an installed base, the same review applies to existing stock: which part numbers are over-specified for their route, which are duplicated, and which could be replaced by one coded module across sites. The coding check is usually the starting point there.
Frequently Asked Questions
Does a compatible module cost less than an original one?
Modules coded for the host platform are typically supplied from third party manufacturers at lower cost than the platform vendor's own part, with the same MSA compliance. Third party brand optics are available through APEX, with the actual manufacturer stated on every quotation.
Should I reduce the spare pool to cut cost?
Size the pool by failure rate and site class, not per cabinet. A single spare covering several similar sites usually gives the same availability with less tied up capital, provided the part numbers have been standardised first.
Where does the biggest saving usually come from?
From reach code discipline. Long reach and amplified modules carry more bill of materials, more power and more spares than the route requires. Correcting the reach class on a mixed BOM normally changes more lines than any single price negotiation.
Is it worth using CWDM or DWDM instead of more fiber?
It usually is where fiber is scarce or new cable is expensive. Fixed wavelength modules and passive mux and demux units let one fiber pair carry several services, at a cost that is stable and predictable compared with new duct work.
Next Step
Send the BOM with the measured distance and fiber type for each line, the host platforms in use and the current spare plan. We will mark which lines are over-specified or duplicated, propose the reach code and coding that fits each span, and quote the mix as one order so the comparison is like for like.
