This extract from CPaaSAA’s report AI Voice: Who Will Run The Conversation? examines one of the central strategic questions in AI Voice: who controls the runtime, and what key players need to do. It builds on the previous articles in this series covering the ecosystem, telco network AI Voice, customer buying behaviour and competitive scenarios. You can download the full report here:
What is runtime in AI Voice?
In the context of AI Voice, runtime refers to the environment where conversations are executed.
This includes:
- How data flows through a conversation
- How AI models are selected and applied
- How decisions are made and actions triggered
- How policies, compliance and governance are enforced
Runtime is not just infrastructure. It is the layer that determines how interactions are processed and what outcomes are delivered.
Runtime is the orchestration layer governing execution across AI Voice systems.
It highlights how runtime sits between:
- Communication inputs (calls, messages)
- AI components (models, analytics)
- Enterprise workflows and outcomes
This matters because it defines control. The player that governs runtime determines how conversations are executed.
Headline insight: Control of runtime determines who captures most value in AI Voice.
Why runtime matters in AI Voice
As AI Voice evolves, value shifts away from transport and towards execution.
The report makes this explicit:
- The communications infrastructure remains essential
- But value increasingly sits in how conversations are understood, governed and acted on
Runtime is where this happens.
It governs:
- Routing and escalation
- Model selection and switching
- Policy enforcement
- Integration with enterprise systems
This makes runtime the point where technical architecture becomes commercial strategy.
Four runtime scenarios
The report outlines four scenarios for where runtime may sit:
- Intelligent Engagement-led (CPaaS / CCaaS platforms)
- Hyperscaler-led
- Telco-integrated
- Enterprise-controlled
Each scenario reflects a different allocation of control.
The diagram above shows how these scenarios differ in terms of execution control, data flow and integration. It illustrates:
- Who governs orchestration
- How flexible the system is
- Where switching costs emerge
This matters because it defines competitive positioning.
Implications of each model
Each runtime model has distinct strengths and trade-offs.
Hyperscaler-led
- Strong AI capabilities and scale
- Expanding from components into full platforms
- Risk of centralised control and dependency
Intelligent Engagement-led
- Strong integration with communications workflows
- Opportunity to position as orchestration layer
- Requires clear demonstration of business outcomes
Telco-integrated
- Strengths in identity, trust and quality
- Potential to embed AI directly into networks
- Requires alignment with enterprise systems
Enterprise-controlled
- Maximum flexibility and control
- Higher complexity and integration burden
Hyperscaler strategy: from component to platform
The report highlights a clear shift in hyperscaler positioning.
Historically focused on components such as AI models and infrastructure, hyperscalers are increasingly moving into orchestration and platform roles.
This involves:
- Embedding AI Voice into broader AI and cloud stacks
- Integrating with enterprise workflows and data
- Positioning as the default execution environment
The opportunity is to standardise runtime within cloud architectures. The risk is regulatory scrutiny and resistance in markets where sovereignty and governance are critical.
Competitive risk and structural tension
The report identifies a growing tension across the ecosystem.
- Hyperscalers expanding upward into orchestration
- CRM and enterprise software capturing outcomes
- Telcos exploring network-native execution
- CPaaS players defending and extending their role
If Intelligent Engagement players remain solely focused on transport and usage, value will migrate elsewhere.
This is not a marginal shift. It is a redistribution of value across the stack.
Stakeholder “so what”?
For CPaaS and engagement platforms:
- Runtime control is the path to maintaining relevance and margin
- Integration with workflows is critical to defend position
For telecom operators:
- Embedding runtime into the network creates a differentiated role
- Without it, AI remains an external layer
For hyperscalers:
- Platform expansion increases control
- Must address governance, sovereignty and trust concerns
For enterprises:
- Runtime choice determines control of data, flexibility and long-term cost
- This is a strategic decision, not a technical one
From architecture to strategy
Runtime brings together multiple dimensions:
- technology (models, infrastructure)
- operations (workflows, processes)
- governance (compliance, policy)
- economics (pricing, margin, switching costs)
It is the point where these intersect.
This is why the report positions runtime as the central control point in AI Voice.
Conclusion
The evolution of AI Voice is not defined by features or models alone. It is defined by where execution sits and who controls it.
Runtime is where conversations are turned into decisions and actions. It is where value is created and captured.
The battle for runtime is therefore the defining strategic contest in AI Voice.
You can download the full report here