Opaque Systems
Opaque Systems (styled OPAQUE) built a trust layer for running AI on sensitive data. Using confidential computing, it kept data protected by hardware even while it was being processed, and later focused on “verifiable AI”: cryptographic evidence of what code ran, where it ran and under what policies. The company grew out of a research project at UC Berkeley’s RISELab about ten years ago, and its co-founders include Ion Stoica and Raluca Ada Popa.
What happened
In October 2026, Opaque replaced its website with a thank-you notice saying: “Our team is joining Anthropic to carry our work forward, and OPAQUE Systems is winding down its operations.” Deal terms were not disclosed, and reports describe it either as an acquisition or as an acquihire in which Anthropic licenses the technology and hires the team. Opaque says its open-source work continues at AgenTrust and that the Confidential Computing Summit continues at the Linux Foundation. No public date has been given for when the commercial service ends.
Key dates
- October 2026: Opaque announces its team is joining Anthropic and the company is winding down.
- October 19, 2026: the deal is widely reported in the tech press.
- Service end date: not published.
What users should do now
- If you are an Opaque customer, ask the company in writing how long the service will run and how to export your data and configurations.
- If you depend on Confidential MCP, Confidential A2A, TRACE or the manifest projects, move to the AgenTrust repositories and confirm they are maintained.
- Plan a move to another confidential computing platform for production workloads.
Alternatives
- AgenTrust open-source projects: Opaque’s open-source work, including Confidential MCP, Confidential A2A and TRACE, continues here under existing licenses.
- Azure confidential computing: Microsoft’s cloud offering for running workloads in hardware-protected enclaves.
- Google Cloud Confidential Computing: confidential VMs and related services on Google Cloud.
- Fortanix: a commercial confidential computing and data security platform.
The lesson
Acquihires usually end the original product. For security-critical infrastructure, prefer open standards and open-source components you can keep running.