Skip to main content

​
5. Business Model Analysis

​
Revenue Models Across Layers

​
Layer 1: Telephony/SIP (Commodity Pricing)

Typical Pricing:
  • Inbound calls: 0.0085−0.025/minute(US);0.0085-0.025/minute (US); 0.003-0.008/minute (India)
  • Outbound calls: 0.01−0.03/minute(US);0.01-0.03/minute (US); 0.004-0.012/minute (India)
  • DID rental: $1-3/month per number (US); ₹50-150/month (India)
  • Setup fees: $50-500 (often waived for volume)
Margin Structure:
  • Wholesale termination: 5-15% gross margin
  • Retail SIP trunks: 25-40% gross margin
  • Value-added services (fraud, analytics): 60-75% gross margin
Key Drivers of Profitability:
  • Scale: Per-minute margin is thin ($0.002-0.008); profit requires massive volume
  • Network ownership: On-net calls (Bandwidth) avoid wholesale costs
  • BYOC attach: Software margins (APIs, dashboards) higher than minutes
  • International routing: Cross-border calls carry 2-3× margins
Example Unit Economics (US SIP Trunk): India Difference: Lower per-minute pricing ($0.004-0.008) requires 3-5× volume for comparable revenue. Indian providers compensate with:
  • Bundled connectivity (Airtel, Jio) capturing network margin
  • Lower CAC (inside sales vs. US field sales)
  • Compliance services (DLT) as margin enhancer

​
Layer 2: Voice-AI Infrastructure (Usage + SaaS Hybrid)

Typical Pricing:
  • CPaaS voice APIs:
    • $0.0125-0.04/minute (Twilio, Daily)
    • $0.008-0.02/minute (Indian players)
  • WebRTC infrastructure:
    • $0.004-0.008/participant-minute
    • $99-499/month platform fee
  • Recording/transcription: $0.005-0.015/minute add-on
Margin Structure:
  • API minutes: 50-65% gross margin
  • Platform/SaaS: 75-85% gross margin
  • Storage/transcription: 40-50% gross margin (cloud costs)
Unit Economics (WebRTC SaaS): Why Margins Are Higher:
  • Developer-facing: Lower-touch sales vs. telco relationships
  • Cloud-native: No physical infrastructure (vs. Layer 1 POPs/switches)
  • Stickiness: API integration = high switching cost
LiveKit/Daily Disruptor Model:
  • OSS core = land (free community edition)
  • Managed cloud = expand ($99-999/month)
  • Enterprise = harvest ($10k+/month + professional services)

​
Layer 3: Voice-AI Agents (Value-Based Pricing)

Typical Pricing:
  • Per-minute: 0.15−0.40(US);0.15-0.40 (US); 0.05-0.15 (India)
  • Per-interaction: $0.50-2.00 (resolution-based)
  • Subscription: $5k-50k/month (enterprise seat-based)
  • Outcome-based: % of cost savings vs. human agents
Margin Structure:
  • Gross margins: 55-75% (depends on LLM costs)
  • LLM inference cost: $0.03-0.08/minute (falling 30% annually)
  • TTS/STT cost: $0.01-0.03/minute
  • Contribution margin: 40-60% after cloud/model costs
Unit Economics (Autonomous Agent Provider): Why Agent Economics Beat Infrastructure:
  • Value capture: Customers pay for outcomes (deflected calls, faster resolution) not just infrastructure
  • Displacement pricing: AI at 0.15/mincompeteswithhumanagentsat0.15/min competes with human agents at 5-8/min-customers see 95%+ cost reduction
  • Vertical moats: BFSI/healthcare models with compliance = higher pricing power
India Agent Economics: Critical Difference: Indian customers have lower willingness to pay but higher volume tolerance. Strategy must be land-with-volume, expand-with-features.

​
Margin Cascade: Why Layer 3 Captures Most Value

The Integrated Play: Owning Layer 1 → 3 end-to-end allows:
  • Capturing full 0.40/minute(vs.0.40/minute (vs. 0.015 for SIP alone)
  • Eliminating inter-layer revenue sharing
  • Controlling latency end-to-end (competitive advantage)
  • Cross-subsidizing Layer 1 commoditization with Layer 3 margins