400G Wavelength vs Four 100G Circuits: When Does the Big Pipe Actually Cost Less?
Four 100G circuits sound equivalent to one 400G wavelength. On a capacity chart, they are. On an invoice, router, failure diagram and traffic graph, they can be very different. With 400G availability expanding rapidly in 2026, the real buying question is no longer whether 400G exists. It is when consolidating onto the bigger pipe actually saves money.
I have bought enough telecom over the years to know the dangerous number is often the easiest one to compare. In this case, it is the monthly circuit charge. If a carrier quotes $X for 100G and 3.2 times $X for 400G, the answer seems obvious: take 400G. But that ignores the question that actually determines the economics — how much capacity you need today, how fast traffic is growing, what sits at each end of the circuit, and what happens when one link fails. A 400G wavelength can be a bargain at 330 Gbps and a very expensive piece of unused capacity at 140 Gbps.
400G Has Moved Out of the Hyperscaler Corner
The enterprise market changed quickly. Capacity that was once largely associated with hyperscalers and major carriers is becoming a practical procurement option for large enterprises, cloud-heavy businesses, AI infrastructure and data-center operators.
40+ U.S. Metros
AT&T announced expanded 400G wavelength availability in June 2026, with handoff capability across roughly 440,000 properties serving more than 2.3 million business tenants and more than 130 key locations.
Network-Wide 400G
Zayo says its North American wavelength network is now fully 400G-enabled and that customer adoption of 400G capacity has risen sharply as AI, cloud and data-center traffic expands.
550+ Sites
Lumen advertises wavelength connectivity up to 400G across more than 550 sites, including predefined high-capacity routes intended for faster deployment.
The Number That Matters: The 400G Price Multiple
Divide the monthly 400G quote by the monthly 100G quote. That one calculation tells you a surprising amount about where the economic crossover sits.
TeleGeography’s Q4 2025 analysis found substantial route-to-route variation. On terrestrial routes, individual carrier multiples ranged from approximately 2.2× to 3.5×. Competitive Frankfurt–London pricing was around 2.6×. Subsea economics remained higher, with major route multiples approaching the upper 3× range.
When the Traffic Requirement Changes the Winner
A low 400G cost per bit does not help much if most of those bits sit unused.
| Peak Requirement | Minimum 100G Circuits | 400G Needed? | What Usually Matters Most |
|---|---|---|---|
| 80 Gbps | 1 | No | 100G usually has overwhelming cost advantage |
| 150 Gbps | 2 | No | Two 100Gs normally beat a 400G wavelength |
| 220 Gbps | 3 | Possible | 400G starts getting interesting if its quote is below ~3× |
| 280 Gbps | 3 | Increasingly attractive | Quote multiple, headroom and traffic profile decide it |
| 320 Gbps | 4 | Yes, strong candidate | 400G often wins when quote is materially below 4×100G |
| 380 Gbps | 4 | Likely | Port density and operating simplicity add to savings |
| 425 Gbps+ | 5+ | One 400G is insufficient | Now compare 5×100G, 2×400G and available 800G options |
A Simple Example Shows the Trap
Use intentionally round numbers. Assume a 100G wavelength costs $5,000 per month and a 400G wavelength costs $16,000. That is a 3.2× price multiple.
At 180 Gbps of requirement, the 400G circuit is expensive overkill: two 100Gs cost $10,000. At 270 Gbps, three 100Gs still cost slightly less at $15,000.
The economics flip the moment the network requires the fourth 100G circuit. Four 100G wavelengths cost $20,000 per month while the 400G wavelength remains $16,000.
That is $4,000 per month — $48,000 per year — before counting ports, optics, cross-connects or power.
Figures above are an illustrative 3.2× pricing example, not a carrier quotation. Actual wavelength pricing is highly route-, term-, provider- and building-specific.
Then Come the Costs Most Quote Comparisons Miss
| Item | Four 100G Circuits | One 400G Circuit | Likely Advantage |
|---|---|---|---|
| Carrier wavelengths | 4 recurring services | 1 recurring service | Depends on price multiple |
| Customer router ports | 4 at each endpoint | 1 at each endpoint | 400G |
| Customer optics | Multiple 100G optics | 400G optics | Frequently 400G per delivered bit |
| Cross-connects | Potentially multiple at each endpoint | Potentially one at each endpoint | 400G |
| Power / cooling | More interfaces and optics | Higher-rate interface but fewer ports | Often 400G per bit |
| Configuration | LAG / ECMP / multiple interfaces | Single interface | 400G simplicity |
| Incremental scaling | Add 100G at a time | Large capacity step | 100G |
| Failure granularity | Potential loss of only one member | Entire 400G link is one failure domain | 100G if truly independent |
The Hidden Engineering Difference: 4 × 100G Is Not Always “400G”
This is where the financial spreadsheet can get the engineering wrong.
Link aggregation creates aggregate capacity, not necessarily 400G for one flow.
Ethernet link aggregation normally uses hashing to assign traffic flows to individual member links. Keeping a flow on one member helps avoid packet reordering, but it also means a single flow may remain limited by the capacity of one 100G member.
Up to 400G aggregate with well-distributed flows, but individual flows typically remain assigned to a member link.
One native 400G interface removes the 100G member-link bottleneck at the transport handoff. End systems and applications can still impose their own limits.
But Four 100Gs Have One Major Card to Play: Failure Isolation
Four 100G Circuits
- One failed member can theoretically leave 300G operating.
- Circuits can potentially be split between providers.
- Routes can potentially enter the building differently.
- Capacity can be expanded in smaller 100G increments.
- Maintenance can potentially be performed one circuit at a time.
One 400G Circuit
- One circuit failure can remove the entire 400G path.
- Fewer interfaces reduce configuration and hardware complexity.
- Lower port count can reduce customer-side cost.
- One wavelength may produce a lower total cost per delivered bit.
- A second diverse 400G circuit changes the economics completely.
The 400G Sweet Spot
There is no universal traffic number, but the pricing relationship creates clear buying zones.
Growth Can Make Today’s Cheap Design Tomorrow’s Expensive One
Suppose peak traffic is only 180 Gbps today. Buying a 400G wavelength might look premature. But at 25% annual traffic growth, that demand becomes roughly 225 Gbps after one year, 281 Gbps after two and 352 Gbps after three.
A network architect expecting that growth is not really choosing between 200G and 400G. The architect is choosing between installing two 100G circuits now and then repeatedly adding capacity, or installing the larger handoff early enough to avoid another procurement and implementation cycle.
A Falling 400G Market Makes Contract Language Part of TCO
TeleGeography forecasts average 400G wavelength pricing on key global routes declining approximately 16% per year compounded between 2025 and 2032, slightly faster than the expected decline in 100G pricing.
That forecast does not mean every enterprise route will fall 16% each year. It does mean buyers should think carefully before assuming today’s quote should simply be multiplied across a long contract.
| Contract Item | Why Ask |
|---|---|
| Re-rate rights | Can pricing be revisited if market rates fall significantly? |
| Upgrade rights | Can spending on 100G be credited toward a later 400G upgrade? |
| Ramp pricing | Can the buyer contract for future 400G without paying the full rate immediately? |
| Term length | Does a lower 60-month rate outweigh the risk of being locked above market later? |
| Port / handoff changes | Who pays if the network must move from multiple 100G interfaces to 400G? |
| Protection level | Are both quotes protected, unprotected or physically diverse? |
Seven Questions to Put in the RFQ
| Question | What You Are Trying to Find Out |
|---|---|
| Quote 100G and 400G on exactly the same endpoints and term. | Creates a valid price multiple. |
| Are both prices protected or unprotected? | Prevents paying for different resilience levels without realizing it. |
| What are the exact physical routes? | Reveals whether multiple circuits actually provide diversity. |
| What cross-connects are required at both ends? | Four circuits can multiply recurring colo costs. |
| What handoff and optics are required? | Determines whether existing routers support native 400G. |
| Can 100G spend migrate into 400G later? | May make incremental scaling much less expensive. |
| What happens to pricing at renewal? | Important in a market where high-capacity unit pricing is still falling. |
100G Aggregation vs 400G Growth Calculator
Enter your actual traffic, carrier quotes and expected growth. The model calculates how many 100G or 400G circuits are required each year and includes utilization headroom, cross-connects, installation and customer-side port/optic costs.
Default $5,000 / $16,000 pricing intentionally demonstrates a 3.2× price multiple and is not presented as a market quote. Replace it with actual carrier proposals. “N+1” adds one circuit beyond calculated capacity; it does not prove physical route diversity.
| Year | Projected Peak | 100G Circuits | 400G Circuits | 100G Annual Run Rate | 400G Annual Run Rate | 100G Cumulative | 400G Cumulative | Cheaper To Date |
|---|
This tool is a planning model, not an engineering design or carrier quotation. It does not automatically account for taxes, colo cabinet charges, power billing, managed router fees, route protection premiums, early termination liability, internal labor, financing costs, packet overhead, burst behavior, application bottlenecks or provider-specific SLA structures.
