The future
The next generation of computing won't be defined by one kind of processor, device or interface. Quantum processors may work alongside classical machines. Intelligent instruments will analyse information near where it is collected. Robots will cooperate with remote services. AI agents will assemble capabilities their developers never explicitly anticipated.
Why should every new technology require us to reinvent the systems connecting everything together?
The most important example
Quantum computers won't replace PCs. They'll sit alongside classical machines as a specialised resource, and the hard part is everything around them: preparing problems, queueing, cost, access, validating results and keeping a record of how each answer was produced.
Prepare the interface, not the quantum computer.
Operations, data formats, submission, cancellation, status, latency, cost, locality, quotas, authority, results and provenance, described once. Quantum-specific details belong in the adapter.
GPU clusters, FPGAs, photonic processors and remote simulators could fit the same pattern without pretending they're the same thing. A shared fabric coordinates them; it doesn't erase physics or safety requirements.
Technologies to explore
Filter by how close each one is. Nothing in this list is built today.
Plan for algorithm agility: versioned keys and signatures, and migration paths for envelopes, updates, publishers, custody receipts and fabric sessions. Strands uses conventional modern cryptography today; post-quantum algorithms are planned, not deployed.
New equipment publishes its capabilities, and authorised workflows discover it. Machines already discover each other on a network; equipment that describes itself is the next step.
Specialist agents working with narrow permissions and shared capabilities, holding grants like any other participant.
Connect evidence from protected execution environments to envelope authority and custody. Requires integration work and a threat-model review.
Event-driven, low-energy processing of sensor data near its source, which fits a system that is already event-driven.
Strands could coordinate the classical control and telemetry around them. Moving quantum states needs its own physical systems.
Optical accelerators offered as specialised compute services for suitable signal-processing and AI workloads.
A speculative interface to experimental processors and laboratory equipment. Not a near-term product promise.
Local autonomy while out of contact, with synchronisation when communications return.
Architecture priorities
The groundwork for the future, and where each piece stands today.
Machines and services find each other on a network today. Stable, rich capability descriptions are still to come.
Scheduling and reservations run on real hardware. Leases and quotas are built but not yet proven; metering is in design.
A migration strategy that lets keys and signatures move to post-quantum algorithms without breaking what they protect.
One way to submit work and get results back with their provenance, whatever runs it.
Work that must happen nearby, on time, or never over a best-effort link, described explicitly.
The GPU already joins the fabric as a display node. Compute adapters will follow as hardware and partnerships justify them.
The future of computing won't be one computer. It will be everything working together.
Further reading: NIST post-quantum cryptography (FIPS 203, 204 and 205, published August 2024) · Confidential Computing Consortium · Quantum Economic Development Consortium · Open Compute Project · IEEE Standards · Nature Photonics. Links are for background; none of these organisations is affiliated with StrandsOS.
If you build instruments, accelerators or the systems around them, we'd like to compare notes.
