Opaque Systems Team Joins Anthropic as Confidential-AI Startup Winds Down, Terms Undisclosed

Opaque Systems is closing its doors and sending its people to Anthropic. The startup announced the move on October 13, 2026, in a post on its own site titled "From Berkeley Research to the Future of AI: OPAQUE's Next Chapter." The financial terms of the deal were not disclosed.

What Opaque said

"Today, we're proud to share that our team is joining Anthropic to carry OPAQUE's work forward," the company wrote. Its website now says Opaque Systems is winding down operations.

CEO Aaron Fulkerson, who became the startup's CEO in October 2023, and CTO Rishabh Poddar are among those heading to Anthropic. Opaque describes the arrangement as mostly an acquihire. Anthropic will also license technology built around privacy, compliance and data residency.

The companies did not share terms. Opaque's own post does not use the word "acquired," but several outlets report the move as an acquisition: the team goes to Anthropic, Anthropic licenses the technology, and the standalone company shuts down.

What Opaque built

Opaque began about a decade ago as a research project at UC Berkeley's RISELab. Its founders include Ion Stoica, co-founder of Databricks and Anyscale, and Raluca Ada Popa, a security researcher now at Google DeepMind.

The company's core product used hardware-based protections, known as confidential computing, to let organizations run AI workloads on sensitive data without exposing it while it is being processed. More recently Opaque pushed into what it calls "verifiable AI," meaning cryptographic evidence of what model ran, where it ran, and under what policies.

Opaque also contributed to open projects aimed at AI agents and models: Agent Governance Toolkit, TRACE, Confidential MCP and Weight Custody Manifest. Opaque says it partnered with Nvidia, Intel, Microsoft and Google, and had won adoption in Europe and the UAE for work on open standards for sovereign and on-premise AI deployments.

The startup raised a $24 million Series B at a $300 million valuation. Its customers included Anthropic, ServiceNow and Accenture. Anthropic was already a client before it hired the team.

Why it fits Anthropic

Softonic notes that this kind of setup makes sense for finance, healthcare and government buyers that need compliance they can audit.

Dealroom's read is that the deal adds security credentials for Anthropic as it chases enterprise customers with strict data-residency and privacy needs, and that model providers are increasingly absorbing the security layer rather than leaving it to partners. That is outside analysis, not an Anthropic statement.

What nobody has answered

Opaque has not said what the wind-down means for its current customers. Softonic lists support, migration paths, access to tooling and the future of the open-source projects as murky. Customers of a startup that is winding down are left to wait for word; one outlet advises Opaque customers to monitor official communications from Anthropic regarding support and product continuity.

The open-source projects are a separate question. Opaque says it worked with a broader community on them through an initiative called AgenTrust. It has not said who will maintain them now that the company is winding down.

Softonic also raises the bigger question of whether Anthropic should own verification tools at all. The deal puts those tools under the roof of a company whose models they are meant to check. Opaque's own pitch was that customers should replace trust assumptions with cryptographic evidence.

Observers are watching how Anthropic integrates Opaque's technology and personnel into its operations and what that means for its AI offerings.