An $80 million upgrade to Canada’s largest supercomputer is set to help businesses that partner with academics to create products or improve technology.
Component parts for this supercomputer are arriving at Simon Fraser University (SFU), where workers are plugging in cables, installing software and creating an environment where more than six thousand researchers across Canada can remotely conduct their own academic work, or run jobs in conjunction with partner companies.
Burnaby-based Ballard Power Systems Inc. is one of many companies that have benefitted from using the supercomputer.
Alan Young, Ballard’s electrodes and technical collaborations director, told BIV that he has worked closely with Erik Kjeang, an associate professor at SFU in its School of Mechatronic Systems Engineering.
The two used the computer to develop and improve fuel-cell technology.
SFU’s supercomputer runs non-stop. It completes more than 100,000 compute jobs daily and is more productive than any other computer in the country, said James Peltier, who oversees about 25 staff involved with the supercomputer.
What has been known as the Cedar supercomputer since its 2017 launch is set to be fully upgraded by April, after which it will carry a new name: Fir.
The evolution to be known as Fir, and future advancement to eventually include quantum computing, is important for research, business and sovereignty reasons.
In an era when U.S. President Donald Trump regularly refers to Canada as a 51st state and threatens to use “economic force” to achieve that goal, there is a dawning realization for many Canadians that unexpected and harmful protectionist measures from Canada’s longstanding trade partner are likely.
Were Trump to suddenly levy large fees on any foreign individual or company using services from U.S. technology companies—as unlikely as that might seem—Canadian companies using compute resources from Amazon.com Inc. (Nasdaq:AMZN), Alphabet Inc. (Nasdaq:GOOGL) and Microsoft Corp. (Nasdaq:MSFT) would be left at a disadvantage.
“I am concerned if any sovereign nation is beholden to another sovereign nation in terms of their innovation and their ability to move forward at their own pace,” said Peltier, who is SFU’s director of research with the computing group within IT services.
“Canada has long been a leader in artificial intelligence (AI) and technology, and we need to invest in the local infrastructure to allow that kind of innovation.”
Having Canadian companies consistently use U.S. offerings, such as Amazon’s cloud-based computing platform Amazon Web Services, which includes scalable compute services, contributes to Canada not having an effective alternative, Peltier told BIV.
Granted, he said, building equivalent infrastructure in Canada would take a “significant investment from all levels of government” as well as the private sector, given that Amazon has invested billions of dollars in its operations.
Other countries are similarly taking notice.
French President Emmanuel Macron this month announced that foreign and domestic companies plan to invest a total of 109 billion euros ($113 billion) in AI projects in France.
Much of that will likely be spent to enhance France’s compute capacity.
“There’s a requirement for us to have some sort of sovereign capabilities here in Canada,” Peltier said.
Government money helping to make supercomputer speedier
SFU’s supercomputer data centre spans more than 7,800 square feet, although the computer itself occupies less than half of that space.
Administration work is done in the rest of the facility.
The computer has thousands of computer chips, including about 4,000 central processing units (CPUs). There are also hundreds of graphics processing units (GPUs), which are highly prized because they can process work related to AI.
New chips are substantially faster than old ones.
“We’re looking at getting about 10 times the performance for each new GPU,” Peltier said.
“We’re getting about a 40-per-cent increase [in compute power] on each CPU, and we’re getting about a 50-per-cent increase in the number of CPUs. So, it’s a fairly substantial upgrade.”
Helping finance the supercomputer’s upgrade is more than $80 million in funding, announced last year. Nearly $41 million comes from the federal-government funded Digital Research Alliance of Canada (DRAC), SFU’s vice-president of research and innovation, Dugan O’Neil told BIV.
DRAC is separately providing nearly $250,000 to fund three research-support positions.
Other money to finance the supercomputer upgrade include $15 million from in-kind contributions from partners, and $24.5 million from the B.C. government, through the B.C. Knowledge Development Fund.
O’Neil said he plans to apply for some of the $2.4 billion in federal money earmarked for AI that Prime Minister Justin Trudeau announced last year.
Exactly how much he is seeking depends on what rules surround getting the money and whether matched funding must come from provincial governments, he said.
Ottawa has said that of the $2.4 billion, about $1 billion is earmarked to build Canadian public supercomputing infrastructure, while $700 million is designated to help mobilize private-sector investments to increase AI-compute capacity.
“None of that money has been rolled out yet,” O’Neil said.
“Somewhere between $500 million and $1 billion dollars worth of equipment would fit in our data centre.”
The future of computing is quantum
Quantum computing is set to be the next step in computing, and it has been a white-hot but volatile investment niche.
SFU’s supercomputer, and standard desktop computers, are based on binary operations. Computer chips have billions of transistors, with each having a “bit” of data. Those bits have two states: On or off.
CPUs handle the main functions of computers while GPUs can perform many smaller tasks at once.
Quantum computers’ quantum bits, or qubits, can have multiple states. While a CPU or GPU chip’s bits may determine something as being either on or off, a qubit could find that it is both on and off at the same time—something called being in a superposition.
As a result, qubits process data much faster and significantly improve users’ ability to complete complex cryptography, among other tasks, and to do so much faster.
“Right now, the state of quantum computing is as an emerging technology that, within the next five years, is going to become something that is for certain types of problems significantly better than classical computing,” said O’Neil.
He expects SFU’s supercomputer to eventually be updated to include qubits.
“We’re also an institution that has researchers building quantum computers,” he added.
SFU associate professor Stephanie Simmons is one of them. She founded and is chief quantum officer at the Coquitlam-based quantum-computing company Photonic, which raised US$100 million from investors in 2023.
“We currently use Cedar for quantum algorithms development,” Simmons told BIV. “That research is a Photonic-SFU collaboration.”
Photonic this month announced that it had made a scientific breakthrough that would accelerate the timeline to useful quantum computing.
Today’s quantum computers tend to produce a lot of errors because their qubits are sensitive and easily disrupted by environmental noise.
This makes correcting errors vital when using quantum computing applications.
“We’ve been working on better error-correction codes,” Simmons said.
A traditional way of trying to correct errors was through what she called “surface codes”—a method that requires so many qubits that industry insiders were projecting reliable quantum computing to be far in the future.
“Excitingly, there are other ways,” she said, before explaining that instead of using surface codes, researchers could use codes called quantum low-density, parity-check codes, which need fewer qubits.
“We have unleashed upon the world a new family of these codes that are way more efficient,” she said, referring to Photonic.
The timeline for when quantum computing will be viable has been a matter of debate.
Speculation late last year was that it was on the near-term horizon, which led to an investment craze in quantum-computing companies.
Shares in D-Wave Quantum Inc., which has extensive operations in Burnaby, closed out October trading at US$1.06 per share. Two months later, it ended the year trading close to eight times higher, at US$8.40.
Other quantum-computing companies, such as IonQ Inc., Quantum Computing Inc. and Rigetti Computing Inc. saw similar parabolic share-price surges at the end of 2024.
All of those companies have seen share prices fall more than 40 per cent year-to-date.
Influential technology executive and Nvidia Corp. CEO Jensen Huang may have popped the quantum-computing investment bubble at the CES 2025 conference in Las Vegas in early January, when he said practical quantum computing use was decades away.
“If you kind of said 15 years ... that’d probably be on the early side,” he said in a speech at the conference.
“If you said 30, it’s probably on the late side. But if you picked 20, I think a whole bunch of us would believe it.”
Simmons said she believes practical quantum computing activity will take place far sooner than does Huang.
“We’re likely closer to five to eight years, when we start to see these larger-impact breakthroughs,” she said.
