The global pandemic has accelerated conversations about the importance of technology in streamlining workflows, simulating onsite capabilities, and keeping teams connected. But Dillon Erb, M.Arch ’14, and Daniel Kobran, M.Arch ’14, have been having those conversations for a while.
The result of those conversations — as students in studio, as research assistants, as recent alumni who knew deep down that the traditional path of practicing architecture wasn’t for them — is Paperspace, the software-as-a-service (SaaS) company that they launched soon after graduation.
Through Paperspace, named as a nod to early AutoCAD, Erb and Kobran say they are unlocking the next generation of accelerated cloud applications “by providing effortless access to powerful cloud computing and developer-first tooling.” Their flagship product is Gradient — a Kubernetes-based platform that enables more efficient operation of machine learning and artificial intelligence.
Paperspace’s client list includes HBO, Deloitte, and Dropbox; they’ve raised nearly $30 million in funding from investors; and their team is more than 30 engineers, salespersons, and marketers.
It’s a far leap from the early days when it was just Erb and Kobran — with a great idea but no real clue how to run a startup — doing everything from assembling office furniture to providing customer support.
“We were feeling that boundless creative high of being in architecture school, but we also were endlessly naïve,” Erb says of launching their business. But, Kobran adds, “We had this sense that it was a once-in-a-lifetime opportunity to pursue something risky.”
Although it is now ubiquitous, cloud computing was still a relatively nascent idea when Kobran and Erb were students at Michigan. As they looked at the work that Associate Professor Sean Ahlquist and other faculty were doing with CAD and 3D, the duo wondered how cloud computing could make it more efficient and more powerful.
“There was a lot of good engineering and a lot of good technology, but it was hard to access. We realized that was a design problem just as much as anything else,” Erb says. “And we thought, great, that’s in our wheelhouse.”
Erb and Kobran saw that people were using graphics processing units (GPUs) — the technology that enables high-performance computing — for data processing, not for rendering or visual effects. “Today it’s called machine learning, basically using GPUs to build out intelligent applications,” explains Erb, whose M.Arch thesis explored machine learning. “Our insight was that the same thing that we were building for architects also would be good for this data science group doing machine learning.”
Seven years later, “visual effects and GPU rendering and machine learning have collapsed into a new industry,” Erb says. “And they’re not totally dissimilar. People are using machine learning to generate architecture models. They’re using machine learning to generate videos and media.
I think we were lucky in timing, but we also worked hard to orient what we were building to what people wanted” — or maybe what they didn’t yet know they wanted because they didn’t realize it was possible.
It’s a lesson that Kobran and Erb learned at Y Combinator, the Silicon Valley–based startup accelerator: make something that people want. Erb and Kobran participated in Y Combinator, whose other participants include Airbnb, DoorDash, Dropbox, Twitch, and Reddit, just months after graduating from Michigan.
“It sounds dumb because it’s simplistic,” Erb says. “But it’s deceptively complicated to make something that someone really, actually cares about.”
Based on Paperspace’s success, though, Erb and Kobran did just that.
“One of the main things we learned at Y Combinator was how to take customers’ needs and wants and synthesize that information into an actual product roadmap. Then how does that product actually get built? Dillon and I spend a lot of time whiteboarding, still using a lot of the same creative design tools we used at Michigan,” says Kobran, who oversees Paperspace’s business operations as COO while Erb is the company’s primary external face as CEO.
New York–based Paperspace’s community of more than 400,000 users range from individuals to startups to universities. “We were fortunate in that this thing that we were building ended up exploding in the world, and a lot more people were using GPUs,” Erb says.
Their clients also are universities across the world — including Taubman College, which is using Paperspace machines and software to bridge the gap for students who don’t have onsite access to the college’s technology during the pandemic.
It’s been a tough year for most businesses, and Erb and Kobran have navigated challenges with Paperspace, too. At the same time, “the pandemic has accelerated the shift to remote work and distributed teams, which is right in our wheelhouse,” says Kobran. “Although we of course wish it hadn’t happened in such an unfortunate way.”
Kobran, who studied environmental design as an undergrad, and Erb, who studied philosophy, have always had a strong interest in technology. “Michigan provided an amazing opportunity to work at the intersection of technology and design, so that’s how we both ended up there,” Kobran says.
They got to know each other through studios and by working on adjacent projects in Taubman College’s Liberty Research Annex, and they realized they had a shared interest in the fusion of technology and design. They engaged in spirited debates as to which of them was the nerdiest. And now they’ve channeled that competitive fire outward as they’ve grown their business.
“The best way to find a co-founder is to stay up all night in an architecture studio and work with someone to achieve the impossible with zero time,” says Erb. “Launching our company was scary, but once we realized that we could work harder and smarter than anyone else, because we had just spent infinite hours in school designing projects and then pitching them, we became fearless.”
— Amy Spooner