Trusted compute for constrained infrastructure
Nelix designs compute systems for sites where power, connectivity and physical access are unreliable. The work today is architecture, RTL and FPGA validation. Production silicon is a longer-term direction, not a current product.
- Trust boundary
- Supporting logic
- Signed data in flight
- Derived key
- Array cell
- Light = derivation, one direction only
Hardware first. Cloud assumptions last.
Edge and infrastructure compute must keep working when the link drops and the supply sags. That is a hardware problem first.
Trust belongs in the package: identity, measured boot and attributable results, not only in software wrappers after the fact.
Nelix has not fabricated production silicon. Platform pages describe architecture and engineering status: Research, Prototype or In Development.
Conditions the architecture must absorb
-
Intermittent power
Safe power-down and resume from non-volatile state.
-
Sparse connectivity
Local execute, measure and verify. No cloud dependency.
-
Thermal extremes
Passive operation planned into the floorplan, not bolted on.
-
Physical tamper
Die-level monitoring with cryptographic response.
-
Long service life
Trust and maintenance sized for decades in the field.
Where the compute work sits
-
Edge and infrastructure compute
Architectures sized for metering, remote sites, industrial monitoring and other unattended equipment.
-
Hardware trust
Per-device identity, measured boot, attestation and result binding specified into the system, not added later.
-
FPGA and pre-silicon path
Prototypes and RTL validation before any tape-out decision. Figures on this site are design targets, not measured silicon.
Technical topics
Deeper notes on trust, execution and implementation. Design intent, not measured silicon.
Trust
Execution
Discuss compute requirements
Design services and research collaboration are open now. Platform work remains pre-silicon.