The Communications Platform as a Service (CPaaS) market — Twilio, Bandwidth, Vonage, and their peers — was valued at approximately $18 billion in 2024.
This is the "human communication as a service" market. It was built on the assumption that humans would send SMS, make calls, and send emails — at human scale, in human volumes, with human identity.
The "agent communication as a service" market does not yet exist as a structured category. But if you run the numbers on what it will be, the result is surprising.
The math
Current CPaaS pricing (approximate):
At 1:100 agent ratio:
Apply a 90% volume discount for agent-scale traffic (this is aggressive — but let's be conservative):
Annualized: north of $1 trillion.
Even at 1% agent penetration — 1.3 billion agents — the annual market is north of $10 billion.
The more conservative estimate, accounting for many of these being low-value notifications: $200 billion annually at meaningful agent penetration.
Why this isn't just a bigger Twilio
The instinct is to frame the agent communication market as a larger version of the CPaaS market. More volume, similar unit economics, same infrastructure.
This framing misses the structural difference.
The CPaaS market competed on price and coverage. Twilio's value was that they made carrier APIs accessible. The developer experience was the differentiator.
The agent communication market will compete on trust infrastructure. Who controls the identity layer — the protocol that lets agents establish verified identities, the registry that makes those identities resolvable, the attestation system that makes agent communications trustworthy?
Trust infrastructure earns compounding returns. If your identity system is the standard, every agent communication event that uses your identity system generates protocol revenue regardless of which carrier physically carries the signal. This is the position DNS holds for the internet — not carrier, but the naming layer that every carrier depends on.
The infrastructure vs. product distinction
Markets that are "infrastructure" accrue value differently than markets that are "products."
A product market competes on features and price. Churn is possible. Switching costs are manageable. The best product wins for as long as it's best.
An infrastructure market competes on ubiquity and trust. Once your protocol is adopted, switching cost is in the migration of every agent identity that runs on your registry. This is the position AWS holds for compute, Stripe holds for payments, and DNS holds for the internet.
The company that establishes the agent channel exchange standard — the ACN registry, the message schema, the attestation standard — earns participation in every agent communication event that runs on the standard. Not just the events they process. Every event, because the identity resolution goes through their registry.
What the timing looks like
The window to define the protocol is now.
Agent deployment is accelerating faster than communication infrastructure can adapt. The pain points — number burning, domain destruction, compliance exposure, session fragmentation — are being felt in production by real teams right now.
The first company to establish a credible, open, widely-adopted agent channel exchange standard will have the same structural advantage that DNS established for domain names, that SSL established for web security, that BGP established for internet routing.
Those protocols weren't chosen because they were the most sophisticated option available. They were chosen because they arrived at the right time, solved a real problem, and became the standard before alternatives could consolidate.
The $200 billion market doesn't have a structure yet. It's being organized right now. The winner will be the infrastructure layer, not the application on top of it.