Oblivious
Dublin-based confidential computing startup letting organizations run machine learning on sensitive data without ever exposing it in the clear.
Visit Website ↗ + Add to Compare Claim This CompanyOverview
Oblivious, founded in Dublin in 2020 by Oxford PhDs Robert Pisarczyk and Jack Fitzsimons, builds confidential-computing software that lets data scientists and machine learning models operate on sensitive data while enforcing confidentiality constraints throughout processing — an approach the founders describe as ‘eyes-off data science.’ The platform uses hardware-based trusted execution environments combined with attestation and auditing to broker trust between organizations that need to collaborate on data neither party is willing to expose in the clear.
The company raised a €5.35 million seed round in April 2023, led by Berlin-based Cavalry Ventures, to expand its confidential-computing platform for cross-organization data collaboration. In 2022 it partnered with the United Nations to launch a virtual lab on privacy-enhancing technologies for sharing confidential data across borders, giving it a credible, independently-verifiable use case beyond commercial deployments.
Oblivious operates in a technically demanding niche — confidential computing combined with formal privacy guarantees — that remains small relative to broader data-security tooling, and independent, named-customer validation of production deployments is limited in public reporting.
Innovation Matrix Assessment
A working confidential-computing platform, a UN privacy-tech partnership, and a completed seed round within about three years of founding reflects steady, if not explosive, progress for a hard technical niche.
Letting data teams run ML workloads on sensitive data without exposing it in the clear addresses a genuine blocker to cross-organization collaboration on regulated data.
A single €5.35 million seed round closed in 2023 with no publicly reported follow-on funding or disclosed customer roster suggests modest commercial momentum relative to the broader confidential-computing market.
Combining trusted execution environments with formal attestation and auditing to broker multi-party data trust is a technically distinct approach from standard data-masking or tokenization products.
The UN partnership lends independent credibility to the underlying technology, but there is limited public evidence of large-scale commercial production deployments or measured outcomes.
Growing regulatory pressure on cross-border and cross-organization data sharing (AI training data, healthcare, financial data) makes confidential-computing approaches increasingly relevant over the medium term.
Why CISOs Should Care
Enables data science and AI teams to work with sensitive or regulated data across organizational boundaries without exposing the underlying data to the other party or to the platform itself.
What Makes It Different
Applies confidential computing (hardware-enforced trusted execution plus attestation) specifically to multi-party data collaboration, rather than treating confidential computing as a general-purpose cloud security feature.
The Matrix Verdict
57/100 — INCREMENTAL INNOVATOR
An Incremental Innovator: Oblivious has real technical depth and a credible independent partnership (the UN) validating its approach, but commercial traction and independent efficacy evidence remain limited for a company now several years past its seed round.
Editorial Note: Claims vs. Verified Findings
The seed round and UN partnership are independently reported; the company's broader claims about enabling secure multi-party collaboration are grounded in its published technical approach but lack independent third-party efficacy testing in public sources.
Sources
Alternatives to Oblivious
Cyera
AI-powered, agentless data security platform combining DSPM, DLP, and AI-activity monitoring, and one of the fastest-funded startups in…
Cyberhaven
Data detection and response platform that tracks data lineage to detect and prevent data exfiltration and insider risk.
Fireblocks
The category-defining institutional digital-asset custody platform, combining MPC and hardware chip isolation to move roughly $2 trillion in…
Halcyon
Ransomware-focused cyber resilience platform combining endpoint prevention, automated recovery, and data exfiltration protection.
SandboxAQ
Alphabet spinout building AQtive Guard, a cryptographic management platform helping enterprises inventory, assess and migrate to quantum-safe encryption.
Varonis
Data-centric security platform that monitors file, email, and cloud activity to detect insider threats and ransomware before data…