Bring Your Own Carrier (BYOC): How It Works

BYOC — bring your own carrier — means connecting your own carrier’s SIP trunks to a communications platform instead of buying minutes from the platform itself. The platform keeps doing what platforms do — APIs, call flows, agent desktops — while a carrier you chose, at rates you negotiated, carries the calls to and from the phone network.

SIPNEX is licensed by the FCC as a carrier, so this is the view from the trunk side of the setup — platform in front, carrier trunks behind — not from a platform’s marketing page.

What does bring your own carrier (BYOC) mean?

Every platform call has two layers. The software layer is why you picked the platform: programmable call control, IVR builders, agent routing, the developer API. The carriage layer is the unglamorous part underneath. It originates and terminates calls on the public switched telephone network, owns the phone numbers, and signs the calls.

By default the platform resells carriage bundled into its per-minute price. It buys wholesale capacity from underlying telecom carriers, marks it up, and hands you one invoice. Bringing your own carrier breaks the bundle: you keep the platform for its software and point its voice traffic at SIP trunks from a carrier you selected directly.

The major platforms all productized this, under different names:

• Twilio offers BYOC Trunking for Programmable Voice — your carrier connects to a SIP domain you create on Twilio, and Twilio’s voice API runs on top of your carrier’s minutes.
• Zoom Phone documents two deployment models: Cloud Peering (BYOC-C), where the carrier peers directly with Zoom’s data centers and no on-premises SBC is needed, and Premises Peering (BYOC-P), with your own session border controller in the path.
• Microsoft Teams calls the same architecture Direct Routing — your carrier’s trunks reach Teams through a Microsoft-certified SBC.
• Genesys Cloud offers BYOC Cloud and BYOC Premises: Cloud runs SIP trunks straight to the Genesys media tier, while Premises connects through a Genesys Cloud Edge at your site.

Different labels, identical architecture: the platform stops being your carrier and becomes purely your software vendor.

Why operators bring their own carrier

Rate control. A platform’s bundled per-minute price includes the wholesale carrier cost plus the platform’s margin — on every single minute, forever. At low volume the margin is a rounding error. At dialer or contact-center volume it becomes one of the largest line items you can actually negotiate. Bringing your own carrier replaces the opaque bundled rate with a carrier rate you can see, negotiate, and re-shop. That is where the cost efficiency comes from.

Attestation ownership. Under STIR/SHAKEN, the carrier that puts your call onto the PSTN signs it. The attestation level it can honestly assign depends on its relationship with you and your numbers. When the platform is your carrier, you inherit whatever signing arrangement its upstream stack produces. When you bring your own carrier, you choose the entity that signs, and A-level versus B-level attestation is part of that choice: analytics engines treat partial (B) attestation as a lower-trust signal, though no carrier publishes a specific answer-rate gap.

Number ownership and portability. With your own carrier, your DIDs — the phone numbers — live at the carrier, not inside the platform. Switch platforms next year and your numbers do not move at all; you repoint trunks. Without that separation, leaving a platform means porting every number out of it, with the project timeline and reputation-continuity risk that implies.

One carrier, many workloads. A carrier relationship is reusable. The same trunk group and DID inventory can feed a CPaaS integration, a PBX, and a dialer cluster simultaneously. A platform’s bundled minutes feed exactly one thing: that platform.

How it works technically

Strip the product names away and this is ordinary SIP trunking with a platform as one of the endpoints. If the underlying mechanics are new to you, start with how SIP trunking works; here is the platform-specific part.

Outbound (platform to PSTN). When your application or agent places a call, the platform builds the SIP INVITE as usual — but forwards it to your carrier’s ingress instead of the platform’s own upstream carriers. Your carrier authenticates the trunk, validates the caller ID against your authorized numbers, signs the call with its STIR/SHAKEN certificate, and terminates it to the called party’s network. The platform executed the call logic; your carrier originated the call.

Inbound (PSTN to platform). A call arrives at your carrier for one of your DIDs. The carrier forwards the INVITE to the platform’s ingress — typically a fully qualified domain name the platform assigns you (Twilio’s BYOC docs use the pattern yourname.sip.twilio.com, with regional forms such as yourname.sip.us1.twilio.com). The platform answers, runs your call flow, and bridges media. You tell the carrier where to route calls for each number; the platform decides what happens once they arrive.

Authentication: IP-based, not registration. SIP trunks authenticate one of two ways. Registration-based trunks send a REGISTER with digest credentials so the carrier learns where the endpoint currently lives — built for PBXs behind NAT on changing addresses. IP-authenticated trunks skip registration: both sides whitelist each other’s static addresses.

Platform ingress usually runs on the second model. Platform gateways sit on fixed, published address ranges, so carrier and platform exchange IP ACLs, often with TLS signaling and SRTP media on top, and there is no registration handshake to break at 2 a.m. Some platforms also accept digest credentials: Twilio’s BYOC requires an IP access control list, credentials, or both on the domain your carrier sends to.

The SBC variants. Teams Direct Routing and Zoom Premises Peering put a session border controller you operate between carrier and platform. It normalizes SIP dialects, handles encryption, and gives you a demarcation point for traces. Zoom Cloud Peering and the Genesys cloud option skip the customer SBC and peer carrier to cloud directly.

Media. Voice travels as RTP between carrier media gateways and platform media servers, codec negotiated in SDP. Keep G.711u end to end — every transcode hop adds latency and degrades audio, and native pass-through avoids the problem entirely.

Bring your own carrier: who supports it

“Bring your own carrier” support follows a recognizable pattern across platform categories. The CPaaS layer productizes it most explicitly — Twilio’s BYOC Trunking, described above, is the best-known example. The cloud-based UCaaS platforms run parallel programs under their own names, Zoom’s Cloud Peering and Teams’ Direct Routing among them.

Hosted dialers are the mixed category. Managed VICIdial shops generally allow an outside carrier — VICIdial exposes carrier settings, so pointing it at your own trunks is a routine, documented step. SaaS dialer bundles often permit no carrier substitution at all. Outside-carrier support is a useful proxy for how much lock-in a hosted contract carries.

Watch the direction of the trunk, too. RingCentral’s bring-your-own-carrier program runs the inverse way: your existing carrier feeds RingCentral’s cloud PBX, built for multinational deployments outside its native coverage. That is not the same product as a platform accepting your trunks beneath its dialer or API.

When bringing your own carrier is not worth it

It is an operational commitment, and pretending otherwise sells trunks to people who will regret buying them.

Skip it at low volume. The platform’s markup on a few thousand minutes a month is less than the cost of the hours you will spend standing up and maintaining the integration. The bundled price buys real convenience; at small scale, take it.

Skip it with no telecom ops capacity. Bringing your own carrier means someone on your team owns trunk configuration, reads SIP traces when calls fail, manages DID inventory and CNAM, and coordinates two vendors when a problem could live on either side of the ingress. If nobody owns that, bundled minutes are the correct product — a single vendor to call is worth the markup.

Check feature dependencies first. Some platform features assume platform-purchased numbers or platform-side carriage. Before committing, confirm the capabilities you rely on are supported on outside trunks — the gaps are occasionally surprising.

Swapping the carrier versus leaving the platform entirely

Bringing your own carrier and full migration answer different questions. The first says: the software is worth keeping, the carriage is not. You keep the APIs and agent tooling and swap only the minutes underneath. Full migration says: neither layer earns its cost — replace the platform with your own stack (a PBX, an open-source dialer, direct trunks) and drop the software fees too.

In practice, the carrier swap is often stage one of a migration whether or not you planned it that way. Once your numbers and traffic live at your own carrier, replatforming later becomes purely a software decision — no ports, no attestation reset, no carriage disruption. If you are weighing the two paths for an outbound operation, our Twilio-to-SIPNEX comparison compares rates and billing increments and lays out the migration timeline.

The carrier side of the setup

What to demand from the carrier half of the configuration, and how SIPNEX answers each point:

Signing that survives the hop. SIPNEX holds its own STIR/SHAKEN SP-KI certificate and signs at A-level for verified DIDs under the FCC’s call authentication framework. Because the carrier is the signer, that attestation applies regardless of which platform sits in front of the trunk — your CPaaS integration inherits carrier-grade attestation instead of the other way around.

Published rates instead of platform markup. Entry pricing runs $0.025–$0.030 per minute under 100k minutes per month and steps down to as low as $0.005 at 10M+, on 12/6 billing (a 12-second minimum, then 6-second increments). No per-channel fees, no platform fees, no setup or porting fees — the rate card is public.

Capacity. Unlimited concurrent channels at every volume tier, with G.711u native pass-through. Before you commit traffic, run a test trunk and measure setup time and post-dial delay from your own platform logs.

Frequently asked questions

Why use BYOC with a CPaaS?

Three structural reasons: rate control, attestation ownership, and number portability. Bundled platform minutes carry a markup on every call, which compounds at volume. The carrier that originates your calls determines your STIR/SHAKEN attestation, so choosing the carrier lets you pick one that signs at A-level. And because your DIDs live at your carrier rather than inside the platform, you can change platforms later without porting a single number.

Does BYOC keep STIR/SHAKEN attestation?

Yes — attestation follows the carrier, not the platform. The carrier that puts the call onto the PSTN attaches the signed Identity header, and its attestation level depends on its verified relationship with you and your calling numbers. If your carrier holds its own SP certificate and has verified your DIDs, your calls carry A-level attestation even with a CPaaS executing the call logic in front of the trunk.

Is BYOC cheaper?

At volume, usually — but the mechanism matters more than a blanket claim. Bringing your own carrier replaces the platform’s bundled per-minute price, which includes carriage plus margin, with your carrier’s direct rate plus whatever reduced per-call or per-minute fee the platform charges on outside-carrier traffic. Run the comparison at your actual minute count and check the platform’s fee schedule for that traffic. At low volume, the savings rarely cover the operational overhead; at dialer volume, they typically dominate the decision.

---

SIPNEX is the carrier side of the BYOC equation: A-level attestation under our own certificate, published rates and unlimited channels. Talk to us about a bring-your-own-carrier trunk at (833) 665-2220.