iQrypto / Technology
How the quantum entropy cell works.
A cell under 5 µm, validated in silicon on a European 180 nm CMOS process.
The physics, in one paragraph
Our entropy source is a reverse-biased, deliberately high-leakage PN junction in standard CMOS. Defect-assisted transitions create single electron and hole pairs at random moments: a discrete, Poisson process whose next event cannot be predicted, even in principle.
We count the electrons, watch the source continuously with health tests, and condition the output into random numbers. There is no laser, no LED and no optics.
Readout, entropy assessment, continuous health tests and conditioning turn that physical signal into a usable entropy source. Software must use the source correctly when generating cryptographic secrets.
Want to see a public randomness demo? Explore QRNG.io (opens in a new tab). It is an independent demonstration, not an iQrypto product.
From electrons to random numbers.
AI-assisted explainer, 1 min 48 sec. Diagrams and application examples are illustrative.
Read the video transcript
On-screen opening, no narration: Quantum randomness, generated in silicon. No light source needed.
Every secure connection starts with a random number.
If that number can be guessed, the strongest encryption doesn’t matter. The attacker doesn’t break the maths. They try the likely numbers first.
It keeps happening. In 2008. In 2012. And in July 2026, hardware wallets were drained, linked to a bypass of the random number generator.
The algorithm was fine. The randomness was not.
So where does a good random number come from?
Inside an iQrypto chip, a tiny junction is made in standard CMOS silicon.
Electrons cross it one by one, and physics makes each crossing unpredictable. This is electron shot noise, and it doesn’t require any light.
Many tiny events add up to one noisy signal, and that signal becomes our random bits.
The entropy cell is under five microns, smaller than a red blood cell.
Around it: readout electronics and metal wiring. The die is one and a half millimetres on a side, with one square millimetre of active area. The whole chip draws very little power.
The chip makes random bits. The system needs keys.
We mount it on a board, inside a portable aluminium device. That is VERA-Q.
It runs continuous health tests, and if it detects tampering, the output halts and the host is informed.
Plug it in, and your software can draw random numbers from it.
For example, in a payment terminal, every secure transaction starts with a random number. Draw it from physics and you’ll get better security for your savings.
VERA-Q is available on a shortlist basis, in very limited quantities. Request evaluation access at iqrypto.com.
Closing card, no narration: Quantum randomness on a 1.5 mm chip. Request evaluation access.
The historical examples refer to the 2008 Debian OpenSSL advisory, 2012 network-device research and the 2026 COLDCARD incident account.
The numbers
- Cell size
- Under 5 µm
- Cell power
- Under 1 nW on average, cell only
- Today's chip
- 1.5 × 1.5 mm die, under 15 mW in total
- Active silicon area
- 1 mm² within the 1.5 × 1.5 mm die
- Raw entropy today
- Over 150 bits per second per cell. Cells run in parallel.
- Next-generation performance
- About 10 times faster than the current chip, from our internal measurements.
- Entropy
- 0.99 bits per bit, from our internal NIST SP 800-90B assessment
- Process
- Standard CMOS on a European 180 nm process. No light source and no exotic steps.
Smaller than a red blood cell.
Inside the hardware, the quantum entropy cell is under 5 µm. Compare the source, the 100 × 100 µm integration block and the 1.5 × 1.5 mm chip die below.
The axis is logarithmic: 10 µm, 100 µm and 1 mm are equal tenfold steps. The red blood cell is a typical-size reference.
Add cells, add speed.
One cell is tiny, and cells work in parallel. The IoT-class integration block measures 100 × 100 µm; the individual noise cell remains under 5 µm. A full die serves high-throughput systems. You size the block to the product.
Why integration matters.
When the entropy source sits inside the chip that uses it, random numbers never travel over an external bus that can be probed. There is no extra part on the board, no extra package and no extra power rail.
100 × 100 µm, scaled to billions of devices.
Each integration block occupies 0.01 mm². In an illustrative scenario with 8.2 billion blocks, their combined footprint would be about 82 m² of silicon.
Block-area calculation: 0.1 mm × 0.1 mm × 8.2 billion = 82 m². Wafer margins, manufacturing yield losses and packaging add to the resources required.
Evidence and assurance
Review the source and processing chain for the hardware revision you intend to evaluate.
Formal NIST SP 800-90B validation and AIS 31 readiness are on our roadmap. Today's figures come from our own internal assessment.
- Current evidence
- Internal entropy assessment, with results interpreted for the tested source and conditions.
- Before your evaluation
- Confirm the hardware revision, software releases, operating conditions and measurement point.
- Supporting material
- Ask which source-characterisation, health-test and conditioning documentation can be reviewed for that configuration.
- A characterised physical source
- Enforced operating conditions
- A conservative entropy estimate
- Continuous health tests
- Cryptographic conditioning
Intellectual property
Our random-number-generator technology is protected by granted patents. Patent family WO 2021/240143, "Random number generator".