OVH Groupe announced on June 11 that it is entering exclusive negotiations to acquire Gladia, a French AI startup specialising in real-time speech transcription and audio intelligence. If completed, it would be OVH’s second AI acquisition in 2026, following its purchase of Dragon LLM in March.

The deal signals something important happening in the voice AI market: cloud infrastructure providers are no longer comfortable routing voice workloads through third-party APIs. They want the capability in-house.

What Gladia Does

Founded in Paris in 2022, Gladia built an API-first platform for transcription and audio processing. A single integration gives developers real-time transcription, batch processing, and structured audio intelligence across more than 100 languages. The platform serves over 300,000 developers and more than 2,000 enterprise customers, including HeyGen, Livestorm, Circleback, Leexi, and Recall.ai.

On June 10 — the day before the acquisition announcement — Gladia launched Solaria-3, described as its most accurate speech-to-text model for business audio in core European languages. The timing suggests OVH was already in late-stage negotiations when the model shipped.

Why OVH Is Buying

OVH Groupe is Europe’s largest cloud provider and has spent 2026 building an AI stack it can call its own. The Gladia acquisition adds a native speech layer to a portfolio that already includes large language models (via the Dragon LLM deal), GPU compute, and cloud infrastructure.

The stated purpose is to internalize the speech-to-text building blocks that OVH’s customers currently rely on, enabling OVHcloud and OVHai to offer new voice AI services directly. OVH has been explicit about its ambition: become the sovereign European alternative to AWS, Google Cloud, and Azure for AI workloads.

That word — sovereign — matters in Europe right now. Enterprise customers in financial services, healthcare, and government are increasingly required to keep voice data on European infrastructure, under European data governance rules. Owning a competitive STT layer removes a dependency on US-headquartered vendors and makes that compliance story cleaner.

What This Tells Us About the Voice AI Market

Two patterns are converging.

First, voice AI is no longer a specialist niche. Three years ago, real-time transcription at enterprise scale was an edge case. Today it’s table stakes for call centres, meeting intelligence, customer service platforms, and internal knowledge systems. The market has grown large enough to attract major acquirers.

Second, the infrastructure layer is consolidating fast. When large cloud providers start acquiring voice AI startups, the independent mid-tier gets squeezed. Developers and businesses building on standalone voice AI APIs need to think carefully about their vendor’s acquireability and long-term independence.

The Gladia acquisition follows a series of moves in the voice AI space in 2026 — including Microsoft’s MAI-Voice-2 launch at Build 2026, 3CLogic’s enterprise voice agent rollout, and Aircall’s acquisition of Vogent. Every major infrastructure player is trying to own a piece of the audio stack.

What This Means for Business

For enterprise buyers: The market is consolidating, which typically means better integration but fewer choices. Businesses evaluating voice AI vendors should ask directly: what is your acquisition roadmap, and what happens to my contract if you are acquired? Getting this in writing now avoids painful migrations later.

For European businesses specifically: OVH’s sovereign cloud positioning could simplify GDPR compliance for voice workloads. If voice transcription and processing stays entirely on European infrastructure, audit and compliance reporting gets significantly easier.

For businesses already using voice AI: The real story here is not which company owns Gladia. It is that the enterprise voice AI market has matured to the point where global cloud providers see it as strategic infrastructure worth owning. That is a signal that voice AI has crossed from experimental to essential.

The companies moving the fastest are the ones already running voice agents in production — handling calls, processing audio from meetings, surfacing insights from customer conversations. They are not waiting to see how the infrastructure landscape settles.

---

Omni Voice by Enterprise DNA is built for businesses that want to deploy voice AI employees without building the underlying infrastructure themselves. If you are exploring what a voice AI deployment looks like for your business, book a discovery call.