Executive Summary

In November 2026, researchers from ETH Zurich and Google disclosed two hardware-level attacks—Mior (CVE-2026-0462) and Skyfall (CVE-2026-40671)—targeting AMD SEV-SNP confidential computing on EPYC server processors. Mior exploits a flaw in the Platform Security Processor's initialization of the Reverse Map Table, allowing a rogue hypervisor with BIOS control to read and write encrypted VM memory and forge attestation reports. Skyfall leverages transient execution timing side-channels to extract VM register values and disk-encryption keys without hardware tampering. Both attacks require privileged host access and were demonstrated on Zen 3, Zen 4, and Zen 5 architectures. AMD has released SEV firmware and microcode patches dated August 12, 2026, with full platform firmware updates targeting November–December 2026.

These disclosures arrive as enterprises and cloud providers accelerate adoption of confidential computing to satisfy data-sovereignty mandates, HIPAA, and PCI DSS 4.0 requirements. The attacks expose a systemic gap: tenant workloads in multi-tenant cloud environments cannot independently verify host firmware integrity, making supply-chain trust and cloud-provider patch velocity critical new risk dimensions for security teams.

Why This Matters Now

Confidential computing is now a cornerstone of regulated-industry cloud strategy, with hyperscalers actively marketing AMD SEV-SNP as an isolation guarantee. Mior and Skyfall prove that a compromised or malicious cloud host can silently bypass those guarantees today. With platform firmware fixes not fully available until December 2026 and embedded EPYC patches delayed until early 2027, every organization running confidential VMs on unpatched AMD EPYC infrastructure faces an unmitigated window of exposure—precisely as regulators increase scrutiny of cloud data-protection controls.

Attack Path Analysis

Related CVEs

MITRE ATT&CK® Techniques

Potential Compliance Exposure

Sector Implications

Sources

Frequently Asked Questions

Organizations and cloud tenants running confidential VMs on AMD EPYC 7003, 8004, 8005, 9004, and 9005 series processors are affected. Customers renting confidential VMs from cloud providers depend entirely on those providers to apply AMD's firmware and microcode patches, which are not fully available until late 2026.

Cloud Native Security Fabric Mitigations and ControlsCNSF

Based on the attack progression modeled above, these are the defensive controls that would constrain each stage.

Aviatrix Zero Trust CNSF is relevant here because the attacker, operating from within the hosting boundary, would still need to reach networked workloads and exfiltrate stolen secrets over cloud network paths that CNSF enforces policy on. While firmware and hypervisor compromise sits below the network fabric, CNSF segmentation and egress controls could meaningfully reduce the blast radius by constraining how far compromised workload identities can reach and how stolen data can leave the environment.

Initial Compromise

Control: Cloud Native Security Fabric (CNSF)

Mitigation: Network-layer segmentation enforced by CNSF would likely limit the reachability of infrastructure components involved in hypervisor and firmware management, reducing the number of cloud workloads and control paths accessible from a compromised host operator position.

Privilege Escalation

Control: Zero Trust Segmentation

Mitigation: Zero Trust segmentation would likely reduce the network-level privilege scope available after hardware isolation is defeated, constraining how far an attacker who has compromised one VM's memory can extend access to other workloads across adjacent network segments.

Lateral Movement

Control: East-West Traffic Security

Mitigation: East-west traffic enforcement would likely constrain a forged or impersonated workload's ability to reach other segments and services, as network policy would still require permitted routing paths that an attacker-controlled identity may not satisfy beyond what attestation alone would grant.

Command & Control

Control: Multicloud Visibility & Control

Mitigation: Multicloud visibility and traffic telemetry could likely surface anomalous patterns associated with the host's orchestration behavior if any part of the side-channel extraction sequence generates observable network-layer artifacts, potentially reducing the attacker's ability to operate undetected across the cloud environment.

Exfiltration

Control: Egress Security & Policy Enforcement

Mitigation: Egress policy enforcement would likely constrain the outbound network paths available to a rogue hypervisor or compromised workload attempting to transmit extracted secrets, reducing the number of permitted destinations through which stolen encryption keys and register contents could leave the cloud environment.

Impact (Mitigations)

If prior CNSF controls had been in place, the residual impact would likely be scoped to VMs on the directly compromised physical host rather than spreading across the broader cloud environment, reducing the number of tenants and workloads exposed to stolen credentials and forged identity claims.

Impact at a Glance

Affected Business Functions

  • Confidential Computing and Trusted Execution Environment (TEE) Services
  • Cloud Infrastructure and Hypervisor Operations
  • Cryptographic Key Management and Disk Encryption
  • VM Attestation and Supply Chain Trust Verification
Operational Disruption

Estimated downtime: 45 days

Financial Impact

Estimated loss: N/A

Data Exposure

Confidential VM memory contents including disk-encryption keys, CPU register values containing sensitive application secrets, System Management Mode (SMM) memory data normally inaccessible to the OS kernel, decrypted microcode update contents, and forged SEV-SNP attestation measurements that could allow malicious VM images to masquerade as trusted workloads. Cloud tenants relying on AMD SEV-SNP for confidential computing isolation are exposed to host-level exfiltration of cryptographic material and workload secrets by a rogue or compromised cloud provider. The vulnerability is particularly impactful for multi-tenant cloud environments where tenant data confidentiality is contractually and regulatorily guaranteed (e.g., healthcare, financial services, government classified workloads).

Recommended Actions

  • • Apply AMD SEV firmware (August 12 release), platform firmware, and microcode patches immediately upon availability from your server OEM or cloud provider — tenants renting confidential VMs must pressure cloud providers to confirm patch deployment and provide updated TCB attestation values per AMD-SB-3059 and AMD-SB-3061.
  • • Deploy Multicloud Visibility & Control (CNSF) to monitor for anomalous hypervisor-to-VM interaction patterns, repeated failed VM start sequences, and abnormal performance-counter activity that may indicate Skyfall-style transient execution probing — centralized traffic observability across the control plane is essential for detecting host-layer exploitation attempts.
  • • Enforce Egress Security & Policy Enforcement on all cloud host network paths to detect and block unauthorized outbound data flows that may carry exfiltrated VM secrets (encryption keys, register state) from compromised hypervisor nodes to attacker-controlled destinations, using FQDN filtering and anomaly-based data loss prevention.
  • • Apply Zero Trust Segmentation and East-West Traffic Security controls to ensure workload-to-workload trust is not solely dependent on SEV-SNP attestation — layer identity-based policy and microsegmentation so that a forged attestation report cannot alone grant a malicious VM lateral access to services, secrets stores, or other tenant workloads.
  • • Leverage Encrypted Traffic (HPE) controls such as MACsec or IPsec for all inter-node and inter-region traffic to ensure that even if hypervisor-layer secrets are extracted, data in transit between cloud nodes remains protected from interception, and deploy Inline IPS (Suricata) signatures tuned to detect known SEV-SNP exploit traffic patterns and anomalous PCIe/PSP interaction sequences where network-visible.

Secure the Paths Between Cloud Workloads

A cloud-native security fabric that enforces Zero Trust across workload communication—reducing attack paths, compliance risk, and operational complexity.

Cta pattren Image