Perovskite stability
The single greatest obstacle, and the one we are working on first. Perovskites degrade under humidity, oxygen, heat — and under the very light used to excite them. Nothing downstream matters until this is understood.
Research
Making quantum chips accessible.
Quantum computing is locked behind cryogenics, national laboratories and capital. Photonics at room temperature is the way out — and we are working on the part that gates everything else. The reason we work on it at all starts somewhere else: a question about what artificial cognition would eventually have to run on.
Why this exists
The motivation comes from Shanaya.
Heliosync’s other long research line is a digital human — Shanaya, and the cognitive system being built around her. That work sets the target this page answers to: artificial cognition, growing more capable over time.
Which raises a question that cannot be settled from inside the software. If the objective is increasingly sophisticated artificial cognition, is scaling today’s conventional AI architecture enough — or will fundamentally different computational architectures eventually be necessary?
We do not know, and neither does anybody else. What we can do is take the second possibility seriously enough to study it, rather than assume the first and find out late. That is the entire reason a company shipping business software also keeps a research line in physics.
Two directions are worth studying for it. Neuromorphic computing is the brain-inspired one — architectures modelled on how nervous systems appear to compute rather than on the von Neumann machine. Quantum computing is the other — not a faster classical computer but a fundamentally different computational paradigm. Quantum photonics is a route into that second one: how photonic systems could contribute to future quantum hardware and to quantum technologies that scale.
Only one of the three is bench work. Quantum photonics is where our own effort goes, and it is what the rest of this page describes. Quantum neuromorphic computing is a direction under study, not a programme under way.
The research chain
Photonics at one end, cognition at the other.
Written out, the reasoning runs six links long. Each arrow is a research question rather than a delivery, and several of them are questions the whole field has open.
01
Quantum photonics
How photonic systems could contribute to future quantum hardware and to quantum technologies that scale. The one link on this chain we can put on a bench, and the subject of the rest of this page.
02
Quantum hardware
Sources, circuits, detectors and eventually memory integrated into something a machine could use, rather than an optical table standing in for one.
03
Quantum computing + neuromorphic computing
Two different departures from the machine we compute on today. Quantum computing changes the computational paradigm outright. Neuromorphic computing keeps the physics and changes the model, taking the brain rather than the von Neumann architecture as its reference. Whether the two belong in one system is itself an open question.
04
New computational architectures
Substrates that are something other than a larger version of what we already run. This is the link the chain exists to reach, and nobody has built one.
05
Artificial cognition
The capability those architectures would be for. Whether any of them is required to get there is exactly what nobody can currently answer, us included.
06
Shanaya
The digital human the question came from. If conventional architectures turn out to be sufficient, this chain was unnecessary — and we would rather establish that than assume it either way.
What the chain is not
This is a research hypothesis. It is not a claim that quantum or neuromorphic computing produces cognition.
Nothing above says that a quantum computer thinks, that a neuromorphic chip is a mind, or that either one automatically yields consciousness or cognition. The chain is a line of reasoning we are testing. Every link in it could fail, and the last three are not established by us or by anybody else.
Conventional architectures may well turn out to be enough. If they are, this research line was a detour — which would itself be a result, and a better one than assuming the answer in advance.
The problem
Quantum computing today remains difficult to scale and to access, because building practical, scalable and manufacturable quantum hardware is still an open problem.
What we are building
Our research focuses on quantum photonics and on scalable photonic quantum-chip technologies. The long-term vision is to make quantum processing units increasingly accessible as a computing component, the way CPUs and GPUs are already integrated into modern computers.
The current work is foundational, with a long-term objective of developing the technology toward scalable quantum-chip manufacturing and eventually establishing dedicated manufacturing capability.
The thesis
Room temperature is the whole argument.
A quantum computer that needs dilution refrigeration is an instrument. It belongs to whoever can afford the building around it. Take the cryogenics away and the size, power draw and cost of the machine collapse together — and the thing stops being an instrument and starts being a component.
Photonics is the most credible route there. Photons carry quantum information at room temperature, resist decoherence, and travel down the fibre infrastructure that already exists. The open question is not whether photonic quantum computing works in principle — it is whether the parts can be made cheaply enough, and reliably enough, to matter.
That is the question we are working on. Not a faster quantum computer. A quantum processing unit you would fit to a machine the way a CPU or a GPU already is — accessible as a part, not booked as a facility.
That argument and the chain above are two separate reasons to do the same work, and both hold on their own. Accessible quantum hardware would matter even if artificial cognition never needed it. The chain is why this particular company is the one chasing it.
The architecture
Three stages, two materials.
Generation is deliberately decoupled from processing. Photons are made in perovskite and processed in silicon, which sidesteps the two-photon absorption that limits designs doing both in the same material.
01
Generate
Perovskite quantum dots
Single photons from colloidal perovskite quantum dots, coupled to plasmonic nanoantennas to force emission fast enough to be useful. Solution-processed rather than grown epitaxially, which is a large part of why the approach could ever be cheap.
02
Process
Silicon photonics
Routing and logic on a silicon-on-insulator circuit — the same CMOS processes the semiconductor industry already runs at scale, rather than a bespoke fabrication line.
03
Read out
Perovskite-plasmonic detectors
Self-powered photon counting at zero bias. A processor needs a detector per channel, so a detector that needs no bias supply and no active cooling changes what the whole machine costs to build.
Where we actually are
We have an architecture and a test plan. We have not run the experiments.
No results. No hardware. No published papers — and none implied. The figures in our internal analysis are drawn from the published literature, not measured in our lab, which is why none of them appear on this page.
That is the photonics work, which is the furthest along. The neuromorphic direction is earlier again: reading and argument, with nothing designed, simulated, fabricated or measured behind it.
This is founder-led research, ongoing and non-commercial. There is nothing to buy here and there is not intended to be.
Open problems
What stands in the way.
Listed because they are real. A research page that describes only the upside is a brochure.
Heterogeneous integration
Perovskites are low-temperature and solution-based; silicon photonics is high-temperature and vacuum-based. Putting them on one chip is a fundamental process incompatibility, not a tuning exercise.
Gate fidelity
Demonstrated fidelities for silicon photonic gates in the published literature remain far below the threshold that fault-tolerant computation requires. That gap is the field's problem, not only ours.
No quantum memory
The architecture as drawn has nowhere to store a photon. Without memory a processor is limited to feed-forward operations, which bounds what it can ever run. A parallel track, openly missing today.
Current work
Testing whether perovskites survive.
Stability is the bottleneck, so it is where the experimental work starts. We have specified a full factorial study — 48 device configurations across material composition, interfacial passivation and encapsulation — benchmarked against ISOS and IEC protocols with 1,000-hour accelerated lifetime testing: damp heat, light soaking and thermal cycling.
The outcome we want is unglamorous and specific: one to three architectures that survive all of it. Everything downstream depends on that answer, and we do not have it yet.
Phase 1
Prototype validation
A single-qubit processing unit: a passive silicon photonic circuit from a commercial multi-project wafer run, with the perovskite source and detector integrated onto it by transfer printing.
Phase 2
Multi-qubit integration
A small programmable processor with two-qubit gates and the beginnings of error detection, with control electronics stacked more tightly to manage the I/O density.
Phase 3
Chip-scale processor
Fully monolithic fabrication, on-chip pump lasers and integrated quantum memory. A working component rather than an experiment.
Phase 4
Manufacturing capability
Taking the design toward scalable manufacture, and eventually toward dedicated fabrication of our own. A chip that cannot be made in volume is a paper worth reading, not a component anyone can buy — so this is the objective the rest of the work is aimed at, and the most distant thing on this page.
Deliberately undated, and photonics only. We will publish a timeline when we have one we believe — and the neuromorphic direction has no roadmap here because it has no programme yet.
Programmes
Two, and only two.
They are the two ends of the chain above — one at the hardware end, one at the cognition end. Listing a third would mean inventing it.
Accessible quantum photonics
A room-temperature, chip-scale photonic quantum computer. Room temperature is the whole argument: it is what removes the cryogenics, and the cryogenics are what keep quantum computing inside national laboratories.
Synthetic presence & digital humans
What makes a digital human read as present rather than rendered, and what it takes for one to operate safely on its own. Mostly a problem of timing, consistency and trust rather than of graphics.
Publications
Nothing published yet.
Heliosync has not published a paper, preprint or technical report. When it does, they will be listed here in full — not summarised into a claim.
Work with us
Building something that needs to be right.
Partnerships, research collaboration, investment, or a role on the team — the door is the same one.