The Government of Canada has announced its intention to merge BDC and EDC. Today, BDC remains BDC, and we remain fully committed to Canadian entrepreneurs. Learn more

The Government of Canada has announced its intention to merge BDC and EDC. Today, BDC remains BDC, and we remain fully committed to Canadian entrepreneurs. Learn more

Why we invested in Qvyon

AI is leaving the screen: Qyvon is building the reality layer for physical artificial intelligence (AI)
4-minute read

For the past few years, AI has lived in our browsers and inboxes.

Now it is moving into the physical world, powering the machines that drive our trucks, inspect our factories, move our goods and fly our aircrafts. The market is moving with it, with the estimated global physical AI market growing to US$960 billion by 2033, according to Grand View Research.

The question that can slow everything down

Building a machine that works in a demo is one thing, proving it is safe enough to deploy is another. The gap between what works in simulation and what works on physical hardware is where most projects stall.

The hardest situations a physical AI system will face are the ones it rarely sees: the unexpected obstacle, the sensor glare at dusk, the failure that happens once in a million miles. It is these cases that are costly to capture and often dangerous to reproduce.

While more data is necessary, it is not sufficient: Most real-world data is routine, so collecting more of it brings diminishing returns. Synthetic data helps, but it introduces a problem of its own: Simulated scenes can miss the physics of real sensors, the noise in real environments and the messiness of real-world edge cases, which creates a gap between what works in simulation and what works in the field. Teams can generate data at scale, yet struggle to show a safety reviewer, a regulator or a customer that it can be trusted for a go or no-go decision.

The result is a deployment confidence gap. Organizations need to know whether they have tested the conditions that matter, whether they can trust the evidence, and whether they are ready for the next release. Too often, they cannot answer with confidence.

Qyvon’s MIMIC platform enables companies to combine their multimodal data to train AI for real-world conditions.

The MIMIC platform: Reality in, evidence out

This is the problem Qvyon, based in Montreal, Quebec, set out to solve. Its MIMIC platform is a reality infrastructure for physical AI. It takes the real-world multi-modal sensor data a company already has, from camera, LiDAR, radar, or any other real-world data and turns it into measurable, traceable evidence for validation.

MIMIC ingests real sensor logs, reconstructs them as reality-grounded 3D and 4D environments, and generates controlled variations of scenarios that are rare, risky or expensive to collect in the field. It then measures how faithfully every output represents the original data, using quantified fidelity metrics, and attaches the provenance and traceability that make results defensible. Finally, it exports validation-ready scenarios into the tools engineering teams already use.

The impact is that each output is tied to reality and carries a measured level of fidelity and provenance. For customers, that means faster validation cycles, less reliance on repeated field-data campaigns, broader scenario coverage and lower risk before costly real-world testing. It also keeps customers in control, with cloud, enterprise and air-gapped deployment options for organizations that guard their data closely. The urgency of the need is already visible, with leading OEMs, Tier-1 suppliers and manufacturers engaging with Qvyon.

Decades at the frontier of sensing and AI

Frantz Saintellemy, President and Chief Executive Officer, Qyvon

At BDC Capital, we back founders who have lived the problem they are solving. Frantz Saintellemy (President and CEO), Antonio Polo (COO) and Pierre Olivier (CTO) are world-class sensing experts who have spent decades in this space, with 25-30-plus years each across AI, sensing, semiconductors, robotics and telecom. They have built and deployed mission-critical systems before, and that experience underpins a company built on more than 10 years of physical AI R&D and a patent-protected platform.

Every industry putting AI into motion will face the same question: Can you prove it’s safe before you deploy it? Qvyon is building the answer. They turn scarce real-world data into evidence that engineers, regulators and customers can trust. With a team that has spent decades at the frontier of sensing and AI, and a platform built for the market, Qvyon has the opportunity to become a foundational layer of physical AI. We are proud to back them, and to see it built in Canada.

As physical AI moves into critical infrastructure and industry, the systems that scale will be the ones that can be trusted. That trust depends on foundational technology that can turn real-world data into reliable evidence. Qvyon is building technology to unlock physical AI across industries and bring it safely into the real world. That is why we invested in Qvyon.

Learn more about MIMIC and Qvyon.

Investing in community growth

The Black Entrepreneurs Fund (BEF) is a $100 million equity fund dedicated to supporting Black-led businesses with strong growth potential and promising business models. Find out if it could be right for you.