Ranovus is an Ottawa-based company which makes highly specialized computer hardware, specifically designed for high-capacity fiber optic connections inside data centres.
CCI President Benjamin Bergen recently sat down with Ranovus CEO Hamid Arabzadeh to talk about the company, and the current state of the Canadian semiconductor industry.
This interview has been edited for length and clarity.
Benjamin Bergen: Hi Hamid, I’ve been really looking forward to this chat. We’re talking so much lately about the semiconductor sector, and how we can build this industry in Canada. Let’s get into it.
Just to start off, to give folks a bit of background, how did Ranovus get started? And where in the semiconductor space are you working today?
Hamid Arabzadeh: We started by focusing on interconnect — the information highway, the roads that connect different sources and destinations of data.
There are three things that happen to data in every information technology system that you can think of — from PCs to hyperscale data centres. Data gets processed; it gets transported somewhere; and it gets stored. And we work in the interconnect space, which is moving the data around.
To move data around, typically you can move it on different types of roads. Electrical signals over copper wire would be like a gravelly road. You can’t go fast, you may damage your car, and so on. Or you can have a superhighway, which is fiber optics. These are strands as thin as your hair, and in each of these strands you can carry two or three orders of magnitude more traffic compared to copper.
Now, to carry this traffic you need a source on one side, the light that turns on and off that generates the signal. And you need a detector on the other side. So we started Ranovus working with the National Research Council (NRC) to develop a laser, and without getting into the technical details, it’s a special kind of laser source. It can generate a lot of colours of light from a single chip, and these colours are like lanes in a highway, so a single strand of hair can have let’s say 36 or 72 lanes on the highway. It’s a very unique technology based on quantum dots.
We started from that, and then in 2012 we quickly saw that data centres would be the engine of the next generation of the data-driven economy. So we worked with hyperscale data centre players, connecting servers together inside the data centre, linking up the storage and the processors and all of that. And in the past four years, with the growth of artificial intelligence and the demand for training algorithms that run in data centres, we’ve seen tremendous growth in the demand for these high-capacity data highways inside data centres.
BB: And in terms of those interconnects, is that helping reduce energy usage? I can understand the demand for more capacity, but is it increasing efficiency?
HA: Yeah, absolutely. When the fiber optic technology was invented, the first application was for very large distances on submarine cables. And in that environment, the price and the power consumption were not really an issue, because there was no other way to get it across the ocean. These were like millions of dollars per system.
Then the next step was to bring it inside the country and connect different cities together, which was called a long-haul system. And that became an order of magnitude cheaper, and an order of magnitude lower power consumption than the submarine version.
Then the next step was to bring fiber into metropolitan areas, like if you wanted to connect different data centres around Toronto. The distances became much shorter. And then the order of magnitude of price and power consumption went down one more step.
Now, once you go inside the data centre, the distances become like 500 meters. So now you have to have a technology that competes with electrical cables. So you’re three orders of magnitude away, in terms of cost, from outside data centre technology, and about two orders of magnitude in terms of power consumption; you have to be more power efficient and you have to be miniaturized, by the way, because you don’t have that much space in these server racks. You need to package things together, instead of having a server, you can have a server on a chip. And that’s really the direction to reduce the power consumption and reduce costs. And these are significant reductions, not just some 2% or 30%, we’re talking about 70% power consumption reduction versus today’s technology, doing the same type of work.
BB: You kind of alluded to in your earlier comment, but semiconductors are really powering our future — everything will be built on silicon components, from AI to autonomous vehicles to even weapons and national security.
Building a semiconductor company in Canada is kind of unusual right now. It’s been a long time since we’ve been industry leaders in this space. We’ve seen a lot of capacity dissipate. What’s it like to build a company here, doing the work that you’ve been focused on?
HA: Canada had a huge history in this space. Nortel’s optical division was a leader in this space. It was purchased by Ciena, and the location they have here in Canada is still the world class interconnect technology supplier. There’s a lot of expertise still in Ottawa with the same people that were doing it in the Nortel days.
Now, Ciena is specialized in long-haul and metropolitan optical systems that I mentioned earlier, so a window opened up for us to invest in the next generation of this technology, and the key ingredient of it is the system know-how. And that know-how exists right here in in Ottawa. That’s why we started our headquarters here in Ottawa.
But at the same time, we’re sourcing expertise and components from all over the world. We work with NRC to build the laser, and we partnered with GlobalFoundries to build our chip in Malta, New York. We couldn’t find anybody around the world who could manufacture with the necessary precision to put it all together, so we developed that know-how ourselves and created our facility in Kanata, Ontario.
So in terms of input costs, the chip costs maybe $10 and the laser would be like $2. But then the value add is putting it together and testing it. And then you sell it for $400. And the revenue is going to be recognized in Canada, taxed in Canada, and our IP will be in Canada.
That’s the real difference.
BB: Obviously semiconductors are having a bit of a moment in terms of public policy, national security, and a general awareness of their importance.
How has that been for a company like Ranovus? And where do you see kind of the industry moving as the larger geopolitical pieces play out, and the near-shoring trend progresses?
HA: We repatriated all of our manufacturing from Asia to Ottawa, and this was a very big step for us, with all the investment we had there. We realized that if we are sharing some of our IP with partners in Asia, then we are educating them on how to build the next generation of these things — not just at the level of a silicon wafer, but also then the process of how you transfer that wafer into something that is part of an AI system.
So we looked around and partnered vertically here with the U.S., and we listened to our customers, who are the hyperscale data players. They were very, very interested in having a local supply chain. They were one of the key drivers behind our strategy to repatriate our manufacturing, and also try to develop more proprietary technologies for AI interconnect.
The U.S. government has limited chip sales to Asia, specifically China. They’ve said that you cannot sell your high-end chips to them because AI hardware is now part of a sovereignty equation. So this is becoming sort of weaponized in a way that even the finished products are becoming a sovereignty issue.
And you know, from a shareholder perspective, we have a very, very strong Canadian base with BDC, EDC and Omers Ventures as our supporters. So yeah, it’s been a really important aspect of our business, that we’re Canadian and we’re not taking funding from anybody outside of North America and Europe.
BB: On the domestic focus, maybe that’s a thread we can pull on a bit more. When you think about how to build a domestic sector that actually leads to a lot of the positive outcomes with Ranovus — bringing good manufacturing jobs back to North America, and selling a finished product for not $12 but $400.
What do you think the steps are that get us to a stronger place?
HA: I think the key is to create a critical mass around the successes we have. Ranovus may be one of them, but there’ll be other companies as well. But in a market like this, where there are only a couple really successful firms out there, you really need to see how you can make these firms un-aquirable.
It’s a word I use a lot with folks. What I mean is building and not having an endgame for the company. With most startups, the venture capitalists ask right away, what’s the endgame? Who are you going to sell the company to? When are you going to IPO? So there’s an endgame for some of the participants in the company.
But if you have a venture where there’s a market window and you can insert yourself as a critical component of the market, like Nortel did for optical interconnect, and like we feel Ranovus can do for optical interconnect in AI, then there’s no need to have an endgame. You’re just scaling the company, and the formula is to see if you can acquire some additional companies and add to the scale.
I feel that in areas where you don’t have critical mass, you need to support those companies that exist. The federal government could be doing much more to include Canadian semiconductor companies in their infrastructure projects. With the amount of digital infrastructure the federal government is responsible for, this would be a powerful way to support growth. Other countries do this, and there’s no reason Canada can’t do it too. I think in our industry, because of the scarcity of Canadian-owned entities, we need to come up with a way of nurturing consolidation and scale. This is a very different approach than trying to create more startups; our solution is not to create more startups.
Most of the Canadian startups you look at in the semiconductor space are founded by university professors who spun off something from their research. It’s very unlikely that without additional globally successful executive talent and more business expertise — the people who’ve done this type of thing and know how to make it successful — these startups can grow and scale.
I heard from somebody that, “You aspire to what you see.” We need to create more examples to inspire the next generation to create a robust semiconductor ecosystem in Canada.
The Council of Canadian Innovators is a national business council of more than 150 scale-up technology companies headquartered in Canada. Our members are job-creators, philanthropists and leading commercialization experts in the 21st century digital economy.