Thank you, Madam Chair, for the opportunity to appear before this committee.
I'm Ewan Reid, founder and CEO of Mission Control.
Mission Control is a 100% Canadian-owned, 11-year-old start-up headquartered here in Ottawa that develops intelligent space systems. We are among the world leaders in deploying AI on spacecraft.
In 2023, we became the first organization in the world to send a deep-learning AI to the moon. Later that year, we deployed AI in a European Space Agency earth observation satellite. Last year, we launched Mission Persistence, Canada's giant leap for AI in space. Persistence is a year-long demonstration of using deep learning at the edge on the spacecraft itself to downlink actionable insights, instead of spending millions to send raw data.
Currently, we're under contract with the Canadian Space Agency to develop the technology to use AI on the spacecraft to detect the location and direction of an active wildfire from orbit. We're also a prime contractor on what could be Canada's next flagship space program for the development of a lunar utility rover, leading a consortia of companies from across Canada. This rover will have to survive and operate on its own on the moon, requiring a vast amount of onboard autonomy.
While Mission Control is a leader in AI, it is important for me to stress that the type of physical AI we develop is fundamentally different from what most people think about when they think of AI. Unlike large language models, which consume enormous amounts of power and train with datasets as vast as the Internet, our deep-learning models are specifically designed to operate on the type of very limited computing platforms that can be sent to operate in space.
The deep technical problem that my company solves is enabling the use of AI when computing power is limited. I'm not talking about server farms in space. I'm talking about relatively small spacecraft, incredibly far away or moving incredibly fast. Physics imposes limits on data transmission: constrained bandwidth and high latency. With spacecraft sensors increasingly designed to generate more and more data, onboard intelligence is key to unlocking the most from space missions.
Training data for algorithms and space applications is also very hard to come by. In the case of operating on the moon, we simply don't have tens of thousands of images of the lunar surface to train a model. That's why we built a 4,000 square-foot moon yard right here in the nation's capital and captured thousands of images to create our own proprietary dataset before we sent AI to the moon.
Like Canada's water and minerals, Canadian data is a key national resource that must be protected and stewarded to realize benefits for Canadians and to prevent abuse by bad actors. The world's largest AI chatbot companies are foreign entities, but so are the companies that provide the services and infrastructure that this AI relies on, companies such as Amazon Web Services and Nvidia. It creates risks for Canadian companies and Canadians when they are forced to rely on foreign entities to provide key services.
For example, our company's AI workflow was recently disrupted when OpenAI bought a company that was key to providing AI training services for our models. Even though our models are trained exclusively in Canada and we maintain ownership of all the training data, we had no say and no recourse when OpenAI shuttered the service we were using.
In upcoming regulation and procurement, Canada should preference fully Canadian-owned AI solutions and provide tools and support to companies, universities and individuals to rigorously defend their IP and copyrights against foreign entities, particularly those that use them to train AI models. Here, the Remote Sensing Space Systems Act, used to regulate data generated by space systems that perform earth observation, may serve as a guide. Incidentally, this act is due to be reformed, and the use of AI on board spacecraft to parse data should be taken into account.
Any AI company, but in particular a foreign AI company, should be regulated and required by its regulator to produce training data, training logs and proof that sensitive Canadian data is maintained on servers in Canada. Canada does world-leading work when we pick specific areas where we want to excel. Examples in space include space robotics and synthetic aperture radar.
Rather than trying to replicate successes elsewhere, focusing on physical AI, AI that interacts directly with sensors and robotics, is key to unlocking the benefits of this technology for globally competitive manufacturing, serving remote Arctic communities and responding to the effects of climate change. Canada's next billion-dollar space exploration mission with a lunar utility vehicle, is an excellent north star under which to advance these technologies, inspire Canadians and reap the benefits in terrestrial sectors.