Sanas technical moat narrowing

The key shift is that speech quality is becoming a feature inside larger voice stacks, not a standalone moat. Sanas still wins when a contact center needs accent conversion that sounds natural and does not make agents sound like a different person, but OpenAI and Deepgram now offer real time voice, translation, and speech infrastructure through broad APIs, while contact center platforms can bundle adjacent AI into existing seats and workflows.1 2 3 4 5 6

• Sanas sells into a narrow workflow with high willingness to pay. A contact center can roll Accent Translation out agent by agent, then add speech enhancement, language translation, and speech intelligence. That expansion motion has supported about 160% net revenue retention, roughly $1M average contracts, and estimated revenue growth to $77M by August 2026.1
• The overlap is getting more direct. OpenAI now offers GPT-Realtime-2, GPT-Realtime-Translate, and Realtime-Whisper for live voice and translation in one API. Deepgram has moved from speech to text into a fuller stack with multilingual Flux STT, Flux TTS, Aura-2, and a Voice Agent API, aimed at customer service and IVR workloads.2 3 7 8 9
• If model quality converges, distribution matters more than demos. Genesys positions AI as part of an end to end orchestration platform across channels and workflows, and Sacra research on Cresta, Retell AI, and Parloa shows the same procurement pattern, large enterprises prefer fewer vendors when AI can be bundled into the contact center platform they already run.4 5 6 10 11 12

The next phase favors whoever controls the customer relationship and the broader call workflow. Sanas can keep growing by becoming a larger voice layer across translation, enhancement, and analytics, but as base speech models get cheaper and better, enterprise deals will increasingly be won through bundled distribution, integration depth, and pricing power.1 2 3 4 5 6