Confidential Computing Summit 2026

Mark Russinovich, CTO, Deputy CISO and Technical Fellow of Microsoft Azure, delivered a headline keynote titled Attested, Transparent, Sovereign: The Evolution of Confidential Computing. The session addressed how trust in AI systems must be hardware-verified rather than policy-assumed, and how Microsoft Azure is building sovereign infrastructure for the next generation of confidential workloads.

Nelly Porter, Director of Product Management for GCP Confidential Computing and Encryption at Google, and Anand Pashupathy, VP and GM of Intel Product Assurance and Security at Intel, delivered a joint keynote on future-proofing confidential AI. The session covered how cloud infrastructure and hardware co-design are converging to protect AI workloads at scale.

Hugo Romero, Corporate Vice President of Product Security at AMD, delivered a keynote titled Protecting Data in the Agentic AI Era. The session explored how hardware security foundations must evolve as AI agents operate autonomously across sensitive systems and data, and how AMD's silicon-to-software security approach addresses these demands.

Mike Bursell, Executive Director of the Confidential Computing Consortium, delivered a keynote titled Agentic AI Is Running Your Infrastructure: Who's Accountable When It Goes Wrong? The session addressed the governance and accountability challenges that emerge when AI agents are granted access to production infrastructure.

Dr. Najwa Aaraj, CEO of the Technology Innovation Institute, delivered a keynote on Trustworthy AI with Confidential Computing, covering how government and sovereign AI programs are deploying confidential computing to build verifiable trust in AI systems across national security and critical infrastructure environments.

A closing keynote panel titled The Sovereign Stack: Nations, Enterprises and the Demand for Verifiable AI brought together Hugo Romero (AMD), Dr. Najwa Aaraj (TII), Mark Russinovich (Microsoft), Nelly Porter (Google) and Mike Bursell (Confidential Computing Consortium) to address how nations and enterprises are demanding cryptographic proof rather than contractual assurances for AI governance.

Diamond sponsors for Confidential Computing Summit 2026 included AMD, Confidential Computing Consortium, Google, Microsoft and Technology Innovation Institute. Additional sponsors supported the event across gold, silver and other tiers.

Session recordings from Confidential Computing Summit 2026 are being posted to the Linux Foundation's YouTube channel. Session slides from speakers who provided them are available via the event schedule.

The central theme of the 2026 summit was how confidential computing is becoming the foundational trust layer for the agentic AI era. Sessions explored how AI agents operating autonomously across sensitive data and systems require hardware-verified guarantees rather than policy-based compliance, covering trust infrastructure, attestation standards and runtime governance frameworks.

Technical sessions addressed the challenge of operationalising confidential computing at enterprise scale within containerised environments. Sessions from Google on GKE Hypercluster, NVIDIA on confidential computing attestation for next-generation AI hardware, and a debate on whether Kubernetes is the right platform for confidential compute were among the confirmed programme highlights.

A dedicated thread on sovereign AI and national infrastructure addressed how governments and sovereign cloud providers are deploying confidential computing to meet data residency, privacy and security requirements. The keynote panel featuring AMD, TII, Microsoft, Google and the Confidential Computing Consortium addressed how nations are demanding verifiable AI stacks rather than trusting contractual assurances.

Sessions covered how confidential computing enables sensitive data collaboration across organisational and jurisdictional boundaries, including a large-scale data clean room case study from Japan presented by Acompany, WhatsApp Private Processing from Meta, and cloud-to-edge infrastructure patterns for protecting AI weights and model intellectual property.

Sessions addressed the shift from identity-based trust toward hardware-verified attestation, covering NVIDIA confidential computing attestation for next-generation AI hardware, AWS Nitro System architecture and trust-building through secure cloud infrastructure, and how attestation results can adopt common standards for seamless consumption across relying parties.

The Courtyard hosted all plenary keynotes across both days, opening with welcome addresses from Jim Zemlin (Linux Foundation), Aaron Fulkerson (OPAQUE) and Nelly Porter (Google), followed by the full keynote programme covering agentic AI accountability, sovereign stack infrastructure and future-proofing confidential AI.

Breakout sessions ran across the Gold Ballroom, Mint Ballroom and Column Ballroom throughout both days, covering the full technical breadth of confidential computing from containerised TEE environments to privacy-preserving data collaboration and hardware attestation standards.

The Column Ballroom hosted dedicated poster presentation sessions on innovation, research and use cases, giving practitioners and researchers a platform to share applied work alongside the main session programme.

The full keynote programme was livestreamed for global participants. Session recordings are being posted to the Linux Foundation's YouTube channel following the event, making the full content accessible to the wider confidential computing community on demand.

The summit's focus on agentic AI accountability, hardware attestation and sovereign AI stacks gave security leaders a direct view of the infrastructure decisions that determine whether AI agents can be safely deployed at scale in regulated and mission-critical environments.

Sessions on confidential VMs, Kubernetes TEE environments, GKE Hypercluster and AWS Nitro gave cloud architects the technical depth to evaluate which confidential computing patterns are production-ready and how to operationalise them across hybrid and multi-cloud environments.

Sessions on privacy-preserving data collaboration, WhatsApp Private Processing, data clean rooms and confidential AI platforms gave data and AI teams a direct view of how sensitive data can be brought into AI workflows without violating regulatory obligations or exposing proprietary model weights.

The sovereign AI track and the keynote panel featuring TII, AMD, Microsoft, Google and the Confidential Computing Consortium gave government and policy teams a concrete view of how nations are demanding verifiable AI stacks and how confidential computing enables data sovereignty without sacrificing cloud efficiency.

With 400+ in-person attendees concentrated across the complete confidential computing value chain, including hardware vendors, cloud providers, software vendors, service providers, regulators and enterprise buyers, the summit served as the highest-density business development and partnership opportunity in the confidential computing calendar.