Jason Lango and Juan Alonso
The spark that launched Luminary Cloud came when watching Nobel Prize lectures. Jason Lango had returned to Sutter Hill Ventures (SHV) in 2018 as an entrepreneur-in-residence after the first company he co-founded, Bracket Computing, had been acquired by VMware. Searching for the next big thing, Jason looked to the world’s top scientists for inspiration.
“Watching the lectures on physics simulation and multiscale modeling in Chemistry by 2013 Nobel Prize laureates Martin Karplus, Michael Levitt and Arieh Warshel was a reminder of the impact of computing power on our ability to simulate physics at high levels of accuracy and realism,” explains Jason. “It reminded me of my early career in supercomputing. It was a call to action to explore modern supercomputing with an application in mind.”
Even in the six years since that Nobel Prize, computing had radically advanced. Graphics processing units had been repurposed from their original task of running display screens and turned to handling huge amounts of data. Cloud computing offered the ability to scale up and down significant processing power. Leveraging both these technologies, modern computing infrastructure offered the ability to answer profound questions about physical systems and products in real time. And that power could truly change engineering, and as a result, the world.
“Computer simulations are going to be the main tool to develop new physical products in the future,” says Juan Alonso, a Stanford University professor who would co-found Luminary Cloud with Jason. “The more we progress, the more we’re going to rely almost exclusively on very high fidelity simulation.”
Jason and Juan’s stories run parallel as far back as the late 1990s. After graduating from Brown University with a bachelors in computer science, Jason moved out West in August 1998 to work for supercomputing juggernaut Silicon Graphics in Mountain View, California. A year earlier, in March 1997, Juan had also made the trip to the Bay Area to start as an assistant professor at Stanford University focused on physics simulation for practical engineering design. Independently they were starting a journey to build the company they would co-found two decades later.
Originally from Madrid, Spain, Juan had moved to the U.S. to study aeronautics and astronautics at MIT in 1988, and went on to pursue a Ph.D. at Princeton University, where his advisor was Antony Jameson, one of the founding fathers of computational fluid dynamics. When he took up his position at Stanford, Juan’s interest wasn’t just in solving theoretical and technical problems, but in finding ways to apply them to the real world. That interest also brought him back East to NASA headquarters in Washington, D.C., from 2006 to 2009, where he ran NASA’s Fundamental Aeronautics Program, leading research in supersonic, hypersonic, fixed-wing and rotary-wing projects across the U.S.
Luminary Cloud is in many ways Juan’s third big attempt at building a CFD tool that will change the world. The first, part of the U.S. Department of Energy’s Accelerated Strategic Computing Initiative (ASCI), focused on modeling complex turbulent flows through entire jet engines, and leveraged Design of Experiments high-performance computing resources such as Blue Gene/L at the Lawrence Livermore National Laboratory in Livermore, California. The second, launched soon after Juan’s return from NASA and took a more collaborative approach, building an open-source multi-physics analysis and design framework called SU2 that has worldwide usage in both academic and industry settings. The ASCI project pioneered the use of GPUs for CFD simulations and SU2 embraced the application of CFD to design optimization.
Jason, meanwhile, gained experience as a software engineering leader at high-performance infrastructure technology companies, including Silicon Graphics, NetApp and Cisco. He saw NetApp, a data infrastructure business, grow from a medium-size company with a workforce of around 900 people through to a large corporation of 7,000 and an annual revenue of more than $1 billion. He joined security firm IronPort Systems when it was still a late-stage startup, and experienced its acquisition into technology giant Cisco.
Feeling that he had missed out on Silicon Valley’s dot-com heyday around the turn of the century, and nostalgic for his time at IronPort, Jason began looking for ideas to start his own company. A connection with SHV Managing Director Mike Speiser would lead to joining SHV as an entrepreneur in residence with a goal to search for opportunities where new technology could disrupt large, established industries.
The thread that connected Jason and Juan’s journeys was high-performance computing. In his time at Silicon Graphics, Jason worked on the IRIX operating system for the company’s Origin series of computers. As a graduate student, Juan helped maintain the computers in his lab at Princeton, and when he moved to Stanford, he used part of his startup funding to buy a 16-processor SGI Origin 2000.
When one of Juan’s squash partners, Doug Mohr, invited him to meet Jason at the SHV offices in Palo Alto in early summer 2019, Juan assumed he was going to be pitched an investment idea. But for two hours, Jason merely asked lots of questions: about CFD, about GPUs, about everything Juan had worked on over the previous 20 years.
In fact, SHV had brought the two together to see whether they had chemistry, whether they could work well together, and whether there was mutual interest in starting a company.
Following that first meeting, Jason and Juan had a series of “founder dates” throughout the summer, meeting up at each other’s homes as well as the SHV offices. As they talked, they began to refine the ideas that would become “Realtime Engineering”. They knew that they could run simulations many times faster than the existing players in the CAE space, all tied to older, CPU-based, on-premises architectures. “Nobody had effectively leveraged GPUs. Nobody had leveraged the elasticity of the cloud,” explains Juan. But they still didn’t know how far they could take this.
Without realizing that Jason was a movie buff, Juan made the analogy that he wanted to build the artificially-intelligent engineering assistant, J.A.R.V.I.S., from the Iron Man series. The concept immediately clicked. Luminary Cloud could make CFD simulations (among the most challenging and expensive of all computer tasks) automated and interactive. Such a platform would empower the world’s engineers to design, analyze and innovate far faster and better than they ever could before.
“Computation had to become a commodity. You had to get results very quickly, with the assurance that you were going to have guarantees that they were accurate,” says Juan. “Virtual prototyping is the future. It’s going to take thousands of simulations per product, tens of thousands of simulations. And when we can do them in two to three seconds, and we can run hundreds at a time, you’re going to do them in an afternoon.”
Increasingly sure of the potential of their idea, and that they now had a clear path forward, Jason and Juan went back to SHV to talk to Mike Speiser. And in October 2019, in the same conference room where they had first met a couple of months earlier, they founded Luminary Cloud.
