Google’s Gemini to power Apple’s AI features like Siri

January 20, 2026

🔊 Soundcheck

  • Apple–Google Siri AI Deal Moves Voice Forward

  • Deepgram Becomes Voice AI Unicorn

  • VoiceRun raises $5.5M to scale voice agents

  • Modulate's Ensemble Listening Breaks Ground

Read time: 4 minutes

🔥 Hot Mic

Big moves, deep dives, and standout stories.

Apple struck a multi‐year deal with Google to license Gemini AI for the next Siri overhaul, targeting spring 2026 launch.

Apple is shaking up Siri by licensing Google’s Gemini AI in a multi‑year deal worth around US$1 billion annually. The arrangement powers Siri’s more complex reasoning and summarization tasks using Apple’s own Private Cloud Compute infrastructure.

This move accelerates Apple’s AI roadmap while preserving user privacy, since Gemini runs behind the scenes. The upgraded Siri, part of the broader Apple Intelligence platform, is slated to arrive with iOS 26.4 in spring 2026.

Key Points:

  • Multi‑year deal to integrate Google’s Gemini into Siri

  • Approximate cost: US$1 billion per year

  • Gemini runs on Apple’s private cloud, not Google’s servers

  • Upgraded Siri launching spring 2026 with iOS 26.4

Takeaway: By leveraging Google’s Gemini AI while keeping it within Apple’s private infrastructure, Apple can rapidly level up Siri’s intelligence without compromising privacy, laying critical groundwork for its AI future.

Deepgram raised $130M at a $1.3B valuation to boost multilingual, real‑time Voice AI expansion.

Deepgram has secured $130 million in Series C funding at a $1.3 billion valuation, officially earning unicorn status. This round was led by AVP and includes support from both long-standing backers and new strategic investors. The capital will fuel multilingual voice AI infrastructure, global growth, and continued innovation in speech models. Deepgram also acquired OfOne, extending its real‑time voice offering into the restaurant sector.

Key Points:

  • Raised $130M in Series C at a $1.3B valuation.

  • Led by AVP; includes YC, BlackRock, Twilio, SAP, Citi Ventures.

  • Deepgram acquired OfOne to expand restaurant voice AI.

  • Funds support growing language support and real‑time infrastructure.

Takeaway: Deepgram’s latest funding round and OfOne acquisition underscore its leadership in real‑time multilingual Voice AI, positioning the company to scale across new markets and sectors with robust infrastructure and innovation.

VoiceRun raised $5.5M to help developers code, test, deploy, and scale AI voice agents enterprise-wide.

VoiceRun emerged from founders Nick Leonard and Derek Caneja’s frustration with existing voice agent tools being either too simplistic or too custom. They built a platform that favors a code-first approach, providing flexibility and deep configurability developers need for complex use cases. The company just closed a $5.5 million seed round led by Flybridge Capital, with backing from RRE Ventures and Link Ventures. With this funding, VoiceRun aims to accelerate its enterprise tooling, deployment flexibility, and early access developer program.

Key Points:

  • Raised $5.5M seed funding from Flybridge, RRE Ventures, and Link Ventures

  • Code-first platform for AI voice agents offers deeper customization than no-code tools

  • Supports enterprise-grade deployments with testing, observability, and rapid iteration

  • Launching early access program to onboard independent developers and startups

Takeaway: VoiceRun’s code-centric voice agent platform offers a middle path between simplistic no‑code tools and heavy custom systems—empowering developers and enterprises to build and scale reliable, customizable voice interfaces at pace.

Modulate has unveiled Velma 2.0, its latest Ensemble Listening Model (ELM), designed to perceive not just spoken words but the full emotional and acoustic context of voice. Rather than relying on a single, massive language model, Velma 2.0 orchestrates hundreds of specialized sub-models to capture tone, prosody, background noise, deception cues, and more. This architecture allows it to interpret voice conversations with richer insight and greater nuance, overcoming the limitations of traditional LLMs that operate on text alone. Modulate claims Velma 2.0 surpasses leading voice AI in accuracy while cutting costs dramatically, and it's already deployed across hundreds of millions of enterprise-level conversations.

Key Points:

  • ELM uses over 100 specialized sub-models analyzing tone, emotion, and audio features.

  • Velma 2.0 interprets intent, sarcasm, stress, deception, and more from real-time voice.

  • It outperforms top LLMs by about 30% in conversational understanding accuracy.

  • This system is 10–100× more cost-effective than traditional foundation models.

Takeaway: By layering multiple, focused sub-models into an ensemble, Modulate's Velma 2.0 offers a smarter, more nuanced, and cheaper way to understand real-world voice—advancing enterprise conversational AI beyond text-only systems.

🎙️ Mic Drop

What else is making noise in voice AI.

Deepgram is now a unicorn, bolstered by major funding and positive cash flow for its voice AI infrastructure. (eweek.com)

VoiceRun debuts its full-stack voice AI platform, empowering enterprise builders after raising a $5.5M seed round. (prnewswire.com)

VoiceRun secures funding to provide developers granular control over enterprise voice agent workflows and deployment. (siliconangle.com)

ElevenLabs is reportedly in funding talks that could value the text-to-speech leader at $11B—major validation for generative voice AI. (pymnts.com)

Healthcare provider CareXM launches AI voice agents to automate non-clinical calls and ease workload for clinicians. (businesswire.com)

Zadarma launches a multilingual AI voice agent for global VoIP, boosting customer call automation capabilities. (prnewswire.com)

Bolna secures funding to build a self-serve voice AI agent platform targeting enterprise users with minimal deployment friction. (moneycontrol.com)

Ringg AI raises Series A to enhance its multilingual voice AI offering and expand its presence in voice-first markets. (thehindubusinessline.com)

Startup 'Kai Calls' launched after piloting a custom voice AI solution for a local business, showcasing hands-on innovation. (openpr.com)

Secondesk is deploying free AI-powered voice tutors to universities, demonstrating educational impact of conversational voice agents. (natlawreview.com)

Story showcases how personalized AI voice tech supports accessibility and quality of life for individuals with speech impairments. (eweek.com)

NLPearl outlines its strategy to build scalable, developer-centric infrastructure for next-generation human-compatible voice AI. (entrepreneur.com)

Kapture CX moves beyond traditional IVR, deploying AI voice agents that take customer-support actions, not just answer calls. (etedge-insights.com)