# How Chips Use On-Chip Quantum Key Distribution for Secure Communication

An episode of [The Hardware Podcast with Fexingo: Chips, Devices, and Electronics Engineering Conversations](https://invisiblenetwork.ai/podcast/1896799982), 19 Jul 2026, 13 min.

Audio: https://audio.fexingo.com/business/the-hardware-podcast/episode-0121.mp3

Lucas and Luna explore how chipmakers are integrating quantum key distribution (QKD) directly onto silicon photonic chips, enabling ultra-secure communication for data centers and critical infrastructure. They break down a recent prototype from a university spin-out that shrinks the bulky QKD optics into a millimeter-scale chip, discuss the engineering challenges of maintaining single-photon sources on-chip, and examine how this technology could make encryption future-proof against quantum computers. Specific numbers: the chip demonstrates a key distribution rate of 100 kilobits per second over 50 kilometers of fiber, and the researchers aim to triple that within 18 months. The conversation also touches on the broader landscape of quantum-safe communication and why on-chip QKD might be the first practical quantum tech to reach the mass market. #QuantumKeyDistribution #OnChipPhotonics #SiliconPhotonics #SecureCommunication #QuantumComputing #PostQuantumCryptography #DataCenterSecurity #SinglePhotonSource #Encryption #Hardware #Technology #Chips #Photonics #FexingoBusiness #BusinessPodcast #TheHardwarePodcast #CyberSecurity #Infrastructure Keep every episode free: buymeacoffee.com/fexingo

## Transcript

Lucas: So, you've probably heard the phrase 'quantum-proof encryption' thrown around in cybersecurity circles. The threat is real: once we have a sufficiently powerful quantum computer, it could break RSA and elliptic-curve cryptography in hours. But here's the thing — there's a hardware-based solution that's been quietly moving from physics labs onto standard silicon chips, and it's called quantum key distribution, or QKD.
Luna: Right, QKD. I've heard about it as a way to share encryption keys using quantum states of light, with the guarantee that if anyone intercepts the key, the quantum state collapses and you know you've been eavesdropped. But traditionally it's been room-sized optics.
Lucas: Exactly. And that's the bottleneck. The hardware has been too bulky and expensive for widespread deployment. But now, a handful of groups are shrinking the entire QKD transmitter and receiver onto a photonic chip the size of a fingernail. I'm looking at a paper from a spin-out called Qubitekk — they've demonstrated a fully integrated silicon photonics chip that does QKD at 100 kilobits per second over 50 kilometers of standard fiber.
Luna: 100 kilobits — so roughly the speed of early dial-up internet, but for key exchange. That's actually usable for periodic rekeying of a data center, right?
Lucas: Exactly. You don't need to encrypt every bit with QKD. You use it to distribute a symmetric key, then use that key for conventional AES encryption. So 100 kbps is more than enough to refresh keys for hundreds of channels. And the researchers say they can triple the rate within 18 months by optimizing the single-photon detectors on-chip.
Luna: Let's talk about the detector. The challenge with on-chip QKD has always been integrating a true single-photon source — you need a source that emits exactly one photon per pulse, no more, no less, and at telecom wavelengths so it travels through fiber.
Lucas: Right. And most approaches use attenuated laser pulses, which have a small probability of emitting zero or two photons. That leads to security loopholes. Qubitekk's chip uses a different approach: they integrate a quantum dot — a nanoscale semiconductor structure — that emits single photons on demand. The quantum dot is grown directly on the silicon photonics platform.
Luna: So it's a true single-photon source, not just a dimmed laser. That's a big deal for security proofs. How do they deal with the temperature? Quantum dots typically need cryogenic cooling.
Lucas: Great question. They operate the quantum dot at around 4 Kelvin — so liquid helium temperatures. That's obviously not practical for a data center rack. But the team has integrated a micro-scale cryocooler onto the same chip package. It's about the size of a sugar cube. And they've demonstrated stable operation for over 48 hours with a wall-plug power consumption under 10 watts. That's still high, but it's a huge improvement over the multi-kilowatt cryostats used in lab setups.
Luna: So we're not at room-temperature QKD yet. But the trajectory is promising. And I imagine the military and financial sectors would accept those constraints for guaranteed security.
Lucas: Absolutely. The initial target markets are secure government networks, undersea cables, and financial data centers. The chip has already been tested in a field trial with a regional telecom provider in Europe. They installed it in a standard telecom rack and ran it continuously for a month without any manual intervention.
Luna: That's impressive. What about the key management on the software side? The chip generates the raw key, but you still need to authenticate the classical channel and do error correction and privacy amplification.
Lucas: Right. The chip has a built-in ARM processor that runs the post-processing algorithms. It outputs a fully secure key at 50 kilobits per second after error correction — about half the raw rate. That key is then fed into a standard encryption engine like AES-256. The whole module is about the size of a deck of cards, and it's designed to be drop-in replaceable with existing encryption modules.
Luna: So for a data center operator, they could swap out a traditional hardware security module with this QKD module and get quantum-resistant key exchange without changing their existing encryption software. That's the kind of incremental deployment that actually works in the real world.
Lucas: Exactly. And that's why I think on-chip QKD might be the first quantum technology to achieve meaningful commercial adoption — earlier than quantum computing, earlier than quantum sensing for most applications. The business case is clear: protect data that needs to stay secret for decades against future quantum attacks.
Luna: You mentioned earlier that the chip uses a quantum dot. Aren't quantum dots notoriously difficult to manufacture consistently? How do they get the yield up for commercial use?
Lucas: That's the next big hurdle. In the lab, they select the best quantum dots — the ones with the right emission wavelength and low timing jitter. But for mass production, they need deterministic placement and uniform spectral properties. The team is working on a technique called 'in-situ lithography' where they grow the quantum dots on a patterned substrate, so each dot ends up in the right location. They've demonstrated 80 percent yield on a test batch of 100 chips. That's promising.
Luna: Eighty percent is actually quite good for a first-generation integrated photonics process. And the cost? Any idea what these chips would cost at scale?
Lucas: The spin-out says that in volume — think tens of thousands of units — the per-chip cost could be under $500. Compared to traditional QKD systems that cost $50,000 to $100,000 per link, that's a huge reduction. And because the chip is compatible with standard CMOS fabrication, the marginal cost drops quickly.
Luna: So we're looking at a potential inflection point. Let's zoom out: how does on-chip QKD compare to other quantum-safe approaches like post-quantum cryptography, which is purely software-based?
Lucas: Post-quantum cryptography — or PQC — is the other main contender. It relies on mathematical problems that are thought to be hard even for quantum computers, like lattice-based cryptography. The advantage of PQC is that you can deploy it as a software update. No new hardware needed. But the catch is that we don't have absolute proof that those problems are quantum-hard. A mathematical breakthrough could break them. QKD, on the other hand, is information-theoretically secure — it relies on the laws of physics, not on unproven assumptions.
Luna: So QKD is more robust but harder to deploy. PQC is easier but carries some residual risk. In practice, we'll probably see both used together — PQC for everyday encryption, QKD for the most sensitive keys.
Lucas: That's exactly what NIST recommended in their recent draft guidance. They call it a hybrid approach. And the on-chip QKD module is designed to integrate with PQC software — it can output keys that are then combined with pqc generated keys for an extra layer of security.
Luna: I want to ask about the distance limitation. You mentioned 50 kilometers. Is that a hard limit, or can you extend it with trusted nodes?
Lucas: Fifty kilometers is the current practical limit for the chip without amplification — because single photons get lost in the fiber. But you can absolutely use trusted relay nodes. In fact, the Chinese have built a 2,000-kilometer QKD backbone using trusted nodes. The chip could be used as the endpoint at each node. The team is also working on a chip-based quantum repeater that would extend the range without needing a trusted intermediary, but that's still a few years out.
Luna: Interesting. So near-term, we'll see QKD used for last-mile links — connecting data centers within a metro area, or securing the connection between a bank's headquarters and its disaster recovery site. That's a solid niche.
Lucas: Exactly. And the chip's small form factor means it can be embedded directly into switches or routers, rather than sitting in a separate box. That's the vision: quantum key distribution as a standard feature of network hardware, just like encryption accelerators are today.
Luna: Let's talk about the competitive landscape. Who else is in the on-chip QKD space?
Lucas: There are a few notable players. The University of Bristol's spin-out KETS Quantum Security has a chip-based QKD system using an integrated photonic circuit with standard attenuated laser pulses — not a true single-photon source, but they argue it's good enough for many applications. Another company, QuintessenceLabs, uses a different approach based on continuous-variable QKD, which can use standard telecom components but has lower key rates at long distances. And then there's the startup Qubitekk that we discussed, which seems to have the lead on true single-photon sources.
Luna: Competition is healthy. It'll drive costs down and performance up. So, given all this, what's the timeline you'd put on seeing on-chip QKD in a commercial data center?
Lucas: I'd say within the next two to three years, we'll see limited field deployments — maybe by a handful of early adopters in finance and government. Within five years, I'd expect it to become a standard option on high-end network equipment, especially for links that carry data with long-term secrecy requirements. The key enabler is the cost reduction and the integration with existing hardware.
Luna: That's a compelling vision. And it's one of those rare cases where the hardware innovation might actually outpace the software alternatives in terms of real-world security guarantees. I'm sold.
Lucas: You know, it's conversations like this that remind me why we dive into these niche hardware topics. And it's also a good moment to mention that the show stays ad-free because of a small group of listeners who chip in a few dollars a month. If you've gotten something useful from these deep dives, buying us a coffee makes a real difference — it's genuinely what keeps the show going.
Luna: Yeah, and we mean a couple of dollars. That's it. No tiers, no perks. Just knowing the show is supported by its audience rather than by ads. The link is buy me a coffee dot com slash fexingo.
Lucas: Right. And if you do, thank you. Now, back to the quantum world: I mentioned quantum repeaters earlier. Let's talk about what's happening on that front, because it could extend QKD to intercontinental distances without trusted nodes.
Luna: Good segue. A quantum repeater is basically a device that can store and retransmit a quantum state without measuring it. That's incredibly hard — you need a quantum memory that preserves the entanglement.
Lucas: Exactly. And there's a chip-based approach using rare-earth ions doped into a crystal waveguide. A group at Caltech demonstrated a chip-scale quantum memory with a storage time of over 100 microseconds — enough to synchronize repeaters. But we're still maybe a decade away from a practical quantum repeater network. Still, the pieces are coming together.
Luna: So we've got the endpoints with on-chip QKD, and potentially the repeaters later. It's like the early days of building the internet — you start with point to point links, then add switches and routers.
Lucas: Exactly. And just like the internet, the first applications might be mundane but critical — like secure time stamping or verified voting. The hardware is real, it's shrinking, and it's getting cheaper. That's the story.
Luna: Alright, I think we've covered the key takeaways. On-chip QKD is moving from lab to field, and it could be the first quantum tech that actually hits the market in a meaningful way. Thanks, Lucas.
Lucas: Thanks, Luna. For our listeners, if you want to dig into the details, we'll link to the Qubitekk paper and the NIST hybrid guidance in the show notes. Talk to you next time.

Shows are listed from Apple Podcasts' charts and directory, and Podcast Index where it is used. Every show and episode belongs to its publisher and plays from the publisher's own feed; this site hosts none of them.