7

The future & the open problems

goal: form defensible opinions on where this heads
objective An expert sees the trajectory and the weak spots. Be able to argue where attestation is going — and where it's structurally fragile.
Yesterdaytrust in a physical chipsoldered to your board shift Today & tomorrowtrust in the CPU maker's signing keys(Intel, AMD, Arm) — a few global roots the open tension: convenient & powerful, but centralizes trust in a handful of vendors
The biggest shift: confidential VMs run a virtual chip whose trust chains to the CPU maker rather than a discrete part. Great for cloud scale — but the whole edifice now rests on Intel/AMD/Arm's signing infrastructure. Have an opinion on whether that's better or worse.

The live trends

per-program enclave whole-VM confidential computing one program the rest of it enclave one virtual machine operating system your app a protected region inside one program normal software runs unchanged Intel SGX Intel TDX · AMD SEV-SNP · Arm CCA
What changed is the shape of the protected thing. The older style carves a protected region out of a single program; the VM style puts the whole machine inside the boundary and runs normal software unchanged. SGX lives on in servers after Intel retired it on consumer chips; TDX, SEV-SNP and CCA are the VM-style line.
  • Confidential VMs are winning over the older per-program enclave style. Intel retired SGX on consumer chips (~2021–22); it lives on in servers. VM-style (Intel TDX, AMD SEV-SNP, Arm CCA) runs normal software unchanged.
  • Post-quantum: every signature in the chain could someday be broken by a quantum computer. Migration eyes new algorithms (e.g. NIST's ML-DSA), but no chip mandates it yet. Active research, not shipping.
  • The hard unsolved problem: distributing trustworthy "known-good fingerprints" to every verifier at scale, and knowing whom to trust to run the verifier. Transparency logs (public, append-only records) are a leading idea. Emerging, not standardized.
  • New chips: Arm CCA (Confidential Compute Architecture) and open RISC-V designs (Keystone) extend the same loop to more hardware.
lab 7 · position paper

Write ~1,500 words: "Where measured attestation goes 2026–2030, and its unsolved problems." Cite primary sources; take real positions. It's both a learning forcing-function and a portfolio piece that signals expertise.

mastery gate 7

Defend a specific 3–5 year prediction against pushback, grounded in the RATS model and the trust-root shift. Correctly separate what ships today from what is still research (post-quantum attestation and transparency-log distribution are not deployed standards).

sources
Ephemeral virtual-chip research — arXiv · Intel SGX status — Intel · Arm CCA — Arm · RISC-V Keystone — link