OpenAI Breaks Microsoft Exclusivity as ElevenLabs Hits $500M Revenue

Avatar photo

ByLisa Grant

May 9, 2026

OpenAI expands to AWS and launches production-ready voice models while ElevenLabs secures a massive $11 billion valuation through celebrity and institutional backing.

The landscape of the AI industry shifted dramatically this week as OpenAI transitioned from a Microsoft-dependent lab into a multi-cloud power player. Following the expiration of a six-year exclusivity agreement on April 27, OpenAI moved with startling speed to integrate its latest models, including GPT-5.5 and GPT-5.4, into Amazon Bedrock. This expansion is supported by a strategic deal worth up to $35 billion, which requires OpenAI to utilize two gigawatts of Amazon’s proprietary Trainium accelerators.

This ‘jailbreak’ from the Azure ecosystem coincides with a major push into the voice sector. OpenAI has released three new models via its Realtime API: GPT-Realtime-2, GPT-Realtime-Translate, and GPT-Realtime-Whisper. By moving these tools out of beta and into general availability, the company is signaling that voice-driven AI is no longer an experimental novelty but a production-ready utility for enterprise applications. GPT-Realtime-2 notably features a 128K context window, addressing long-standing complaints regarding the ‘memory’ of voice agents.

While OpenAI scales its infrastructure, ElevenLabs has emerged as a dominant force in the specialized voice market. The startup reported a surge in annual recurring revenue from $350 million to over $500 million in the first four months of 2026 alone. This growth trajectory is backed by a Series D extension exceeding $550 million, featuring a diverse cap table that includes BlackRock, NVIDIA, and high-profile figures like Jamie Foxx and Eva Longoria. Now valued at $11 billion, ElevenLabs counts major corporations such as Deutsche Telekom and Meta among its clientele.

However, this rapid commercial expansion has caught the eye of federal regulators. Reports surfaced this week that the White House briefed executives from OpenAI, Anthropic, and Google on a potential working group to vet AI models before public release. The discussions were reportedly triggered by Anthropic’s ‘Mythos’ model, which possesses the capability to map software security vulnerabilities at scale—a tool deemed too sensitive for public distribution. Although officials later characterized these reports as ‘speculation,’ the incident highlights a growing tension between Silicon Valley’s desire for deregulation and Washington’s fear of AI-enabled cyber threats.

Despite the concentration of AI development in the United States, domestic adoption continues to lag. A report from the Microsoft AI Economy Institute ranks the U.S. 21st in global AI adoption at 31.3%, trailing significantly behind leaders like the UAE at 70.1%. This disparity suggests that while American firms are winning the race to build the most advanced models, the domestic workforce has yet to integrate these tools as effectively as international counterparts.

As the Algorithmic State expands, the consolidation of power among a few key players—OpenAI, NVIDIA, and Amazon—raises critical questions about digital sovereignty. While the market celebrates $500 million revenue milestones and multi-billion dollar cloud deals, the average citizen remains caught in a web of increasing surveillance and data capitalism, where the tools of liberty are increasingly owned by the architects of the new digital panopticon.

Leave a Reply

Your email address will not be published. Required fields are marked *