Anthropic licenses Opaque Systems’ confidential AI technology and hires key engineers

Anthropic has licensed confidential AI technology from Opaque Systems and hired some of the startup’s engineers. The arrangement was announced on October 13, 2026.

This is not a full acquisition. Anthropic took the technology and a slice of the team, and Opaque continues to exist as its own company.

What Anthropic actually got

The deal has two parts: a technology license and a partial team integration. The industry calls the second part an acquihire, which means hiring a company’s people rather than buying the company itself.

Opaque’s core product lets AI workloads run on sensitive data inside hardware-isolated environments. Data goes in, computation happens, and nobody outside the vault, including the cloud operator, can see inside.

Opaque’s system produces cryptographic evidence that a workload ran as intended and followed the required policies. The company calls these attestations. A bank can do more than say its data stayed private. It can hand regulators mathematical proof of that.

Anthropic has not said publicly which engineers made the move or how many joined. Financial terms of the license were also not disclosed.

The research findings name CEO Aaron Fulkerson and co-founders Raluca Ada Popa and Ion Stoica as key figures connected to the deal. Their roles going forward have not been detailed.

From Berkeley lab to $300 million startup

Opaque traces its roots to RISELab, a research initiative at UC Berkeley. The company was founded around 2021 to commercialize that work.

Opaque raised a $24 million Series B round that valued the company at approximately $300 million. Total funding came to about $31.6 million.

Opaque developed its verifiable AI tools in collaboration with Nvidia, Intel, Microsoft, and Google. Its customer and partner list also included ServiceNow and Accenture. Notably, Anthropic itself was already on that list before this deal.

Opaque’s platform includes components designed to help organizations meet regulatory and audit requirements. Finance and healthcare are the obvious targets, since both sectors handle data that comes with heavy legal obligations.

Why regulated industries are the prize

Anthropic wants to sell AI to enterprises, and the most lucrative enterprises are often the most cautious ones. Banks, hospitals, and insurers sit on enormous amounts of valuable data and face strict rules about where that data goes and who touches it.

The deal also fits Anthropic’s broader push into sovereign AI and on-premise deployments. Sovereign AI refers to systems that keep data and computation within a specific country’s or organization’s control. On-premise means running the software on a customer’s own hardware rather than in a shared cloud.

By not buying Opaque outright, Anthropic avoids absorbing a full company, its contracts, and its obligations. It gets the capability and the expertise with less baggage.

For Opaque, the company keeps its independence on paper, but some of its engineering talent now works elsewhere. How it continues serving partners like ServiceNow and Accenture will be worth watching.

Anthropic has not said when Opaque’s technology will appear in its products or which customers will get access first.