Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
Anthropic has agreed to license technology from confidential-AI company OPAQUE and hire part of its team, including CEO Aaron Fulkerson and CTO Rishabh Poddar. Constellation Research reported the agreement on October 13, 2026, but the parties have not disclosed a price or confirmed that Anthropic bought OPAQUE outright.
The deal gives Anthropic a path toward protecting sensitive data across the AI lifecycle—training, fine-tuning, inference and agent execution—as it pursues enterprise, sovereign-AI and on-premises deployments.
What Anthropic agreed to
The reported transaction combines a technology license with selective hiring. OPAQUE is licensing its confidential-computing and confidential-AI technology to Anthropic, while members of the OPAQUE team are joining Anthropic. The named executives are Fulkerson and Poddar.
That structure is best described as an acquihire-style licensing and hiring agreement. It should not be reported as a confirmed full-company acquisition: Constellation Research said financial terms were undisclosed and did not report a purchase price or an outright transfer of OPAQUE.
#1 Best Overall
“We’re proud to share a new agreement with Anthropic that carries OPAQUE’s work forward. We’re excited to see its impact continue in shaping how AI is built and secured.”
| Aspect | What has been reported |
|---|---|
| Announcement | Agreement reported October 13, 2026, by Constellation Research |
| Structure | OPAQUE technology license plus hiring of some OPAQUE personnel |
| Named executives | Aaron Fulkerson and Rishabh Poddar |
| Purchase price | Not disclosed |
| Confirmed full acquisition | No; the available announcement does not establish one |
Why confidential AI matters to Anthropic’s enterprise strategy
Organizations often cannot send their most sensitive information to a general-purpose cloud service, even when an AI model could deliver useful results. Healthcare records, financial data, government information and defense workloads can be constrained by privacy law, contractual controls, national-security rules or internal policy.
Confidential AI is intended to reduce that barrier by protecting data while it is being processed, not only while it is stored or moving across a network. For Anthropic, that supports an enterprise push in which customers may require on-premises operation, a particular national jurisdiction or a sovereign cloud architecture. Constellation Research connected the OPAQUE agreement with those deployment priorities.
The Tool Desk
Outbyte PC Repair FREERepair Windows errors before they cause bigger problemsFix Now →Outbyte Driver Updater FREEScan for outdated or missing drivers - takes under a minuteDriver Scan →- Data-leak risk: controls are designed to limit who—and what software—can inspect prompts, training data, model inputs and outputs.
- Compliance: hardware-backed evidence and cryptographic policy enforcement can help an organization demonstrate that a workload ran in an approved environment.
- Data sovereignty: customers can seek clearer control over where sensitive information is processed and which operators can access it.
- Regulated production use: stronger isolation can make it more practical to move beyond pilots in sectors with strict handling requirements.
How confidential computing protects an AI workload
OPAQUE describes its platform as covering the full lifecycle: model training, fine-tuning, inference and execution of AI agents. The point is not merely to encrypt a database; it is to establish controls around the environment in which the model and data operate.
1. Establish an attested execution environment
Hardware attestation produces a cryptographically verifiable statement about the computing environment and the software configuration running in it. A customer can use that evidence to check that a workload is executing in an approved confidential environment before releasing data or keys.
2. Apply cryptographic protections to collaboration and computation
OPAQUE says its technology uses several complementary techniques:
- Multi-party computation (MPC) allows parties to jointly compute a result while keeping their individual inputs concealed from one another.
- Fully homomorphic encryption (FHE) permits selected computations on encrypted data, so the underlying values do not have to be exposed to the processor.
- Post-quantum cryptography uses algorithms designed to remain resistant to attacks from sufficiently capable quantum computers.
These approaches have different performance and deployment trade-offs. The announcement does not specify which technique is used for each Anthropic workload, what hardware will be supported or how the components will be packaged in a customer product.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Scan for outdated or missing drivers - takes under a minute3Repair Windows errors before they cause bigger problems3. Keep protections in place across the lifecycle
Training and fine-tuning can involve an organization’s proprietary corpus; inference exposes live prompts and responses; agent execution can give a model access to tools, files or business systems. Covering all four stages matters because securing only inference would leave earlier data preparation or later agent actions exposed to a different class of risk.
Best Value
4. Verify policy compliance rather than trusting a promise
Attestation and cryptographic controls are intended to provide evidence that the approved code and policies are running. They do not eliminate the need for identity management, least-privilege access, logging, incident response, vendor assessments or careful agent permissions. Confidential computing protects a defined execution boundary; it is not a substitute for an organization’s entire security program.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.OPAQUE’s funding and performance context
OPAQUE said it raised a $24 million Series B at a $300 million valuation in 2026. The company also claimed that its platform can move enterprises from isolated AI experiments to production deployment four to five times faster. That acceleration figure is a company statement, not an independently verified benchmark, and the announcement does not identify the workloads, baseline process or measurement method behind it.
“The future of AI depends on unlocking the data organizations have never been able to touch.”
What’s actually slowing this PC down?
Pick the symptom - the matching free tool is one click away.
Special offer. See more information about Outbyte and uninstall instructions. Please review EULA and Privacy policy.
What customers should look for next
The agreement’s practical value will depend on what Anthropic actually ships and supports. Enterprise buyers evaluating confidential-AI claims should ask for concrete answers in five areas:
- Deployment: Can the service run on premises, in a sovereign cloud or only in a provider-operated region?
- Attestation: What hardware and software measurements can a customer verify, and can access be blocked when the environment changes?
- Lifecycle coverage: Are training, fine-tuning, inference and agent tool calls all covered, or only selected stages?
- Cryptography: Which workloads use MPC, FHE or post-quantum algorithms, and what latency, cost and model-size limits apply?
- Evidence: Which production customers, audits and independent evaluations demonstrate that the controls work under real operating conditions?
Until those details are published, the clearest conclusion is strategic rather than product-specific: Anthropic is adding confidential-computing expertise to address the privacy, compliance and sovereignty objections that keep many regulated organizations from putting frontier AI into production.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
