Confidential Computing

Confidential Computing is the protection layer that closes the last gap in ordinary data protection. Encryption protects data at rest and in transit, but not while it is being processed — at that moment the data is in memory in clear text, and whoever controls the machine can read it.

This module addresses exactly that window. It uses a hardware trusted execution environment (TEE) so that data is encrypted in memory and isolated from the operator of the hardware, and a remote attestation proves which software is running before any key is released. The result: data stays sealed while it is being used.

What it delivers

• Protection beyond at-rest and in-transit encryption. The processing window — the one encryption normally leaves open — is covered too.
• A defensible answer to “who can read our data”. Even the operator of the host hardware cannot read memory inside the enclave or dump it.
• Proof instead of trust. Remote attestation verifies the running image against a known-good digest before secrets are released.
• Sovereignty that also holds on foreign hardware. An organization can run on hardware it does not own and still keep its data unreadable to the host.
• Practical fit for the AI layer. Models and prompts can be processed in the enclave, so AI workloads benefit from the same protection as other sensitive processing.

Main capabilities

Capability	What it gives you
Encrypted memory	Data is encrypted while in use, not only at rest and in transit.
Hardware isolation	A trusted execution environment separates the workload from the host operator and hypervisor.
Remote attestation	The CPU signs a report proving the running image is the trusted one.
Key release gated by attestation	A key is released only after the attestation has been verified.
Integrity protection	The enclave protects memory integrity as well as confidentiality.
Portable across platforms	The same model applies to on-premises servers and to confidential cloud instances.

The trust chain in plain terms

1. The protected workload starts inside the enclave.
2. The processor produces an attestation report describing what is running.
3. A verifier on the trusted side compares it with the known-good image.
4. Only then is the data key released, over an attested channel.
5. The workload processes the data — encrypted in memory, unreadable to the host.

Typical use

• An organization runs a sensitive workload on hardware it does not own — a colocation or cloud host — and still keeps the data unreadable to the operator.
• An AI workload processes confidential content inside the enclave, so neither the prompts nor the results are exposed to the host.
• A regulated deployment requires evidence that only approved software ever handled the data; the attestation report provides it.
• A disaster-recovery or burst scenario places processing on a partner’s infrastructure without giving that partner access.

How it fits the suite

• VNClagoon AI can process sensitive requests inside the enclave.
• VNCdirectory provides classification, which decides what must be protected.
• VNCfiles and every product benefit wherever their data is processed on untrusted hardware.
• VNCaxf and VNCuxf define where the protected services run.

Roles

Role	What they do
Administrator	Enable and operate the confidential deployment.
Security / compliance	Verify the attestation and the trust chain.
Auditor	Review the evidence that only approved software handled the data.

Related

• VNClagoon AI — the AI workload that benefits most
• How VNClagoon keeps your data sovereign