Short answer: SIP trunk pricing on a CPaaS platform breaks into three line items — a per-channel trunk fee, origination and termination per-minute rates by destination tier, and a region-and-failover multiplier that prices redundancy — and the post works the full TCO on a 50-seat mid-market deployment with two trunks, one failover chain, and three destination tiers. The per-minute voice model contrasts with the per-message model on the messaging side: voice billing is wall-clock duration across a live session, not a per-event meter, so the cost equation trades message-count predictability for minute-count granularity. The pricing-overview hub and the voice-intelligence route already exist on the comparison surface; this post adds the SIP-trunk cost-arithmetic sibling.
The quick answer by deployment shape
Which line items dominate depends on call volume, destination mix, and redundancy posture.
| Deployment shape | Dominant cost line | Why |
|---|---|---|
| Low-volume, single-region (under ~1,000 talk-minutes/month) | Per-channel trunk fee | The trunk fee is a fixed per-channel line; at low minutes it dwarfs the per-minute rate |
| Mid-volume, single-region (~1,000–10,000 talk-minutes/month) | Per-minute origination and termination | The per-minute rate becomes the main lever; destination-tier mix determines the blended effective rate |
| Enterprise, multi-region with failover | Region multiplier on the per-channel fee, plus the destination-tier per-minute blend | Each additional region adds a trunk channel; failover regions price at the same per-channel rate but consume minutes only when active |
The rest of this post is the arithmetic behind those three rows, worked on a concrete deployment.
SIP trunk building blocks — channels, trunks, and carrier posture
A SIP trunk is the IP session between your PBX or SBC and the voice platform. A channel is one concurrent call slot on that trunk — one INVITE carried at a time. A trunk is the logical configuration: credentials, failover chain, routing rules, and the set of channels it provisions. The billing unit is the channel, not the trunk: a trunk with ten channels bills ten per-channel fees, whether those channels are in use or not.
Two carrier postures change the per-minute arithmetic:
- BYO carrier — you keep your incumbent carrier contract and phone numbers, and point your trunk at the platform for the intelligence layer (AI voice agents, programmable voice, call routing). Your carrier bill stays yours; the platform bills only the trunk channels and its own per-minute value-add. Read the full tradeoff in the BYO carrier guide.
- Carrier-of-record — the platform terminates and originates calls on its own wholesale softswitch, and the per-minute rate includes the PSTN leg. This is the bundled posture: one vendor, one bill, and the platform owns carrier health end to end. The carrier-of-record vs aggregator post walks through the route-depth question that determines how many wholesale parties sit between the platform and the destination operator.
The billing difference between the two postures is a line-item split, not a model change. BYO splits the carrier minute onto your own carrier invoice; carrier-of-record bundles it into the platform's per-minute rate. The per-channel trunk fee is the same in both postures, and the TCO equation below works for both — just move the carrier-minute term from one column to the other.
TCO worked example — a 50-seat mid-market deployment
Total cost of ownership for one month of SIP trunk voice on a CPaaS platform:
TCO = (channels on primary trunk × per-channel fee) + (failover channels × per-channel fee) + (talk-minutes × blended per-minute rate)
A 50-seat deployment with a two-region failover posture, three destination tiers, and a representative call mix:
| Line item | Value | Monthly cost (illustrative) |
|---|---|---|
| Primary trunk channels (10 concurrent) | 10 × $2.00/channel | $20.00 |
| Failover trunk channels (5 concurrent) | 5 × $2.00/channel | $10.00 |
| US/CA origination (3,000 min) | 3,000 × $0.014/min | $42.00 |
| US/CA termination (2,500 min) | 2,500 × $0.014/min | $35.00 |
| Tier-2 destinations — UK, DE, FR, AU (1,200 min) | 1,200 × $0.035/min | $42.00 |
| Tier-3 destinations — rest of world (300 min) | 300 × $0.085/min | $25.50 |
| Total monthly TCO | $174.50 |
The per-minute rates above are illustrative shapes of how destination-tier pricing works; substitute the actual per-country rates from the pricing page or your Enterprise rate card. The per-channel trunk fee is the illustrative $2.00/channel/month shape consistent with CPaaS trunk pricing; replace with the actual figure in your quote.
Read per line:
- Trunk channels are the fixed floor. At 10 primary channels, $20/month is the sunk cost before the first minute. At low volume this is the dominant line; at mid-volume and above, the per-minute lines overtake it. A deployment that provisions 20 channels on a hunch and averages three concurrent calls pays for 17 idle channels every month — size channels to observed concurrency, not to seat count.
- The failover region adds channels, not per-minute cost. A failover trunk in a second region is a set of channels at the same per-channel rate. Those channels carry minutes only when the primary region is unreachable, so the monthly line is the channel fee alone under normal operation. The failover multiplier is therefore bounded: it adds a fixed number of channel fees, not a percentage uplift on every minute.
- The destination-tier blend is the cost lever at mid-volume. In the worked example, the blended effective rate across all 7,000 minutes is roughly $0.0206/min — higher than the US/CA stand-alone rate because Tier-2 and Tier-3 destinations pull the blend up. The single largest cost move at this volume is shifting call mix toward lower-tier destinations or negotiating a volume rate card on the higher-tier rows.
- BYO carrier shifts the carrier-minute line, not the equation shape. Under BYO, the origination and termination per-minute lines move to your own carrier invoice, and the platform's bill carries only the per-channel trunk fees plus any platform value-add per-minute uplift. The TCO equation still holds — just reconstitute the minute-rate term from two invoices instead of one.
The decision rule: per-channel trunk fees are the fixed cost anchor; the blended per-minute rate is the volume lever; the failover region adds a bounded fixed multiplier. Run the equation with the actual quote, and the winning move at any given volume names itself.
Multi-region failover math
A second region is priced as a second set of trunk channels, not as a percentage surcharge. The multi-region SIP failover guide covers the architecture — geo DNS, route registration per region, and the cycle-safe failover chain — and the failover explainer walks through the four failure domains (trunk health, route disposition, webhook dependence, and media geography). This section adds the cost arithmetic to that architecture.
A two-region deployment with failover:
| Region | Channels | Monthly channel cost | Minutes carried (normal month) | Per-minute cost (normal month) |
|---|---|---|---|---|
| Primary (US East) | 10 | $20.00 | 7,000 | $144.50 |
| Failover (EU West) | 5 | $10.00 | 0 (idle) | $0.00 |
| Total | 15 | $30.00 | 7,000 | $174.50 |
When the primary region fails and the failover region carries the full 7,000-minute load:
| Region | Channels | Monthly channel cost | Minutes carried (failover month) | Per-minute cost (failover month) |
|---|---|---|---|---|
| Primary (US East) | 10 | $20.00 | 0 (down) | $0.00 |
| Failover (EU West) | 5 | $10.00 | 7,000 | $144.50 |
| Total | 15 | $30.00 | 7,000 | $174.50 |
The total monthly cost is identical in both scenarios — the per-minute rate is the same across regions, and the channel count is the only variable. The failover region's five channels must be sized to carry the primary region's peak concurrency, which is the capacity-planning question the architecture post addresses. An under-sized failover region that can carry only three concurrent calls during a failover event throttles traffic, not cost — but the cost math stays flat.
What changes in a three-region deployment is only the fixed channel-count term: three regions each at their channel count, with the primary sized to peak concurrency and each failover region sized to carry it. The per-minute rate does not multiply per region.
Cross-channel cost governor — the LLM spend governor analog on the voice side
On the messaging and AI-agent side, the LLM spend cost governor bounds the one cost component that scales superlinearly with usage — model inference tokens — through per-tenant caps and automatic model-downshift routing. On the voice side, the equivalent control is the per-channel trunk cap: a tenant sets a maximum channel count per trunk, and provisioning beyond that cap is rejected at configuration time rather than silently added at the next invoice.
The voice-side cost controls a tenant can set:
- Per-trunk channel cap. Write-time enforcement: the trunk API rejects channel counts above the tenant-configured ceiling, so no trunk silently grows to an unbounded channel count.
- Per-destination spend alerts. Usage alert rules on per-country minute consumption, configured on the same usage alert rules surface that governs LLM token thresholds. A rule on Tier-3 destination minutes notifies the operator before a high-rate destination consumes the month's voice budget.
- Auto-top-up caps on the prepaid wallet. The same prepaid wallet that carries SMS, voice, email, and AI agent spend can carry a hard monthly top-up ceiling, so no single channel's overrun — voice or otherwise — draws the wallet past a bound.
The voice-side governor is simpler than the LLM side because voice cost scales with wall-clock duration, not conversation depth: a minute is a minute, regardless of how complex the call is. The risk on voice is not token-count drift; it is channel-count drift — a trunk left at "unlimited channels" during a traffic spike — and the per-trunk channel cap answers that directly.
Frequently asked questions
What are the line items on a SIP trunk pricing quote?
Three: the per-channel trunk fee (a fixed monthly charge per concurrent call slot), the origination and termination per-minute rate by destination country or tier, and any region-and-failover multiplier on the channel count. A quote that bundles these into a single per-minute figure hides the channel-cost floor, which is the dominant line at low volume.
How does BYO carrier change the SIP trunk bill?
Under BYO carrier, the per-minute carrier rate moves to your own carrier invoice, and the platform bills only the per-channel trunk fees plus any platform value-add per-minute uplift. The per-channel trunk fee stays the same; the TCO equation still works — just reconstitute the minute-rate term from two invoices instead of one. The full BYO tradeoff is in the BYO carrier guide.
Does a failover region double the per-minute cost?
No. A failover region adds per-channel trunk fees for the channels it provisions — those channels are billed at the same per-channel rate whether they carry minutes or sit idle. The per-minute rate does not multiply per region; the same destination-tier rate applies regardless of which region carries the call. The multi-region failover post covers the architecture.
How does SIP trunk pricing compare with per-message pricing?
Voice is billed per wall-clock minute across a live session; messaging is billed per discrete event (SMS segment, RCS message, WhatsApp conversation). A voice minute is a continuous cost driver — the session meter runs from answer to hang-up, and every second of silence between turns is billed. A message is a point cost driver — one event, one debit. The consequence: voice cost is predictable per call if average handle time is stable, but unpredictable if AHT drifts; messaging cost is predictable per campaign if volume is forecastable, but unpredictable if per-message segmentation (multi-part SMS, media attachments) is not accounted for. The CPaaS pricing models TCO post walks through the per-message arithmetic.
How many trunk channels should a deployment provision?
Size channels to observed peak concurrency, not to seat count. A 50-seat deployment where at most 8 concurrent calls have ever been observed needs 8 channels plus a concurrency buffer (10 is a safe default), not 50. Every provisioned channel above observed concurrency is a fixed monthly cost with no corresponding minute volume. The voice dashboard's concurrency panel shows peak observed channels per trunk over a rolling window; use that number, not the seat count on the org chart.
Does Devotel Orbit charge a per-trunk setup fee?
No. Devotel Orbit prices usage-based, end to end: the pricing page ships one Pay-as-You-Go track with per-channel trunk fees and per-minute voice rates drawn from the same prepaid wallet as SMS, RCS, email, WhatsApp, video, and AI agents. There is no monthly platform fee, no seat-count charge, and no setup fee for a trunk. The Enterprise track negotiates volume rate cards on the same pay-as-you-go structure for tenants moving past the self-serve rate card.
The takeaway
SIP trunk pricing on a CPaaS platform is three terms — per-channel trunk fees, per-minute destination-tier rates, and a region multiplier on the channel count — and the TCO equation (channels × per-channel fee) + (failover channels × per-channel fee) + (minutes × blended per-minute rate) names the winner at any volume band. Size channels to observed concurrency, blend the per-minute rate against the actual destination mix, and price the failover region as added channels rather than a percentage uplift. The pricing page publishes the self-serve rate card; the equation stays the same at Enterprise volume, with a negotiated rate card in place of the published one.
Published 7 October 2026.