One of the most fundamental breakthroughs at Nvidia has been building processors that power and integrate with highly detailed, compute-intensive graphical simulations, which can be used in a wide range of applications, from games and industrial developments through to AI training. Now two of the engineers who helped build those physical simulations for Nvidia and its customers have raised a sizeable seed round of funding for a new startup as they strike out on their own.
Manchester-based Vsim, as their startup is called, is developing a new physics simulation framework. It has now raised $21.5 million from EQT Ventures, along with Factorial Fund, Samsung Next, Tru Arrow, Xora (a fully owned subsidiary of Temasek), IQ Capital, Koro Capital, Concept Ventures, Lakestar Scout Fund, and Carles Reina. Vsim had previously raised around $1.4 million, so this latest round brings the total raised by Vsim to $24 million. This latest capital injection had previously been rumoured and is coming in at around a $100 million valuation.
Vsim has largely been operating in stealth so far (complete with sparse website), so there is not a lot of evidence of what it’s building just yet. Michelle Lu, who co-founded the company with Kier Storey, told TechCrunch that it is initially aiming to target opportunities in robotics training.
That is not, however, the limit of what the tech might be able to do. One of the key reasons why Vsim has raised a a bigger-than-usual seed round is because what they are building has the potential to be used for a lot more.
The opportunity in the market that Vsim is targeting is that, while simulation technology has been around for years, improvements in processing power are leading to more efficient algorithms, as well as tooling aimed at more specific applications.
Lu and Storey not just the co-founders of Vsim; they are also a couple that have been working side-by-side for nearly two decades, going back to their days as physics students at Newcastle University.
The pair’s first foray into startup life was a short-lived effort when they were fresh out of their PhD program. It might have had the right talent and ideas but it came at the wrong time, and maybe the wrong place, too: it was 2007, and there was little money to be raised for startups building simulation technology in Northern England, even from promising PhDs.
So the pair moved, together, to a new pair of jobs, working for Bizarre Creations, a games studio based out of Manchester. Bizarre found some success and got acquired by Activision, and there they built physics engines for a number of titles. Activision eventually shut down Bizarre and the pair moved to Nvidia, where they worked as engineers focusing on building simulation technology for more than a decade.
Lu said that she and Storey have honed in on simulations for robotics in particular as a first use case because of what they see as a gap in the market: while robots, particularly in industrial settings, have existed for years now, we now seem to be at an inflection point, thanks to advances in processing, mechanics and AI, to enter a new phase in terms of what is getting build and where that will be used.
“Back then, we were just some PhD graduates,” Lu said of the pair’s earliest efforts. “Now we have 20 years of experience and have talked to a lot of potential customers. That’s how we have found the target for our product.”
While companies like Nvidia are also building robotics simulations — indeed if you look into what the founders were doing there, it included a lot of work on exactly that — what Vsim has built and is building takes tech to a new level.
“The reason we’re excited about simulation is that it’s fundamental to a lot of different sectors, ranging from research to entertainment to manufacturing, pharma and robotics,” Sandra Malmberg, the partner at EQT Ventures who led the round, said in interview. Simulation tools today are built for a world with a limited degree of freedom and for fixed environments, but as ambitions for applications grow, robots (and other autonomous machines) will need to make decisions in real time and in the real world, which is dynamic and unpredictable. “There is no simulation platform for that today,” Malmberg continued. “This calls for a high performance simulation where you can act with speed, in real time, with accuracy, taking the room and the world into consideration. And this is what Vsim is building. The can enable robots initially and later, they will do more.”