The company that decided running a simulation shouldn't require filing an IT ticket. Los Altos, 2014 to now - photographed in the act of turning supercomputers into a button.
Here is a slightly absurd fact about modern engineering. A person can design a car, a heart valve, or a jet engine bracket entirely on a screen - and then, when they want to know whether the thing actually works, they hit a wall made of infrastructure. The simulation that answers the question needs a cluster of machines, specialized software licenses, a scheduler, and usually a ticket to an IT team who will get to it eventually. The physics is hard. The plumbing around the physics is, somehow, harder.
Simr - a Los Altos company that spent its first decade called UberCloud - exists because two people decided the plumbing was the real problem. Their pitch is not that they invented a better solver, or a faster chip. It is that the annoying middle steps between "I have a design" and "here is the answer" can be automated away. They gave the idea a name that sounds like it was reverse-engineered from DevOps, because it more or less was: SimOps, short for Simulation Operations Automation.
It is a good name because it tells you the whole thesis in one word. DevOps took the drudgery out of shipping software - the provisioning, the environments, the "works on my machine" - and turned it into automation that developers barely think about. SimOps proposes the same trade for the people who run engineering simulations. Focus on the design decision. Let the software find the compute, stage the job, and clean up afterward.
Simr gives engineers a single platform for using any compute resource with any leading simulation tool.
Launch CFD, FEA and multi-physics jobs in leading tools - Ansys, Abaqus, COMSOL, Siemens - without hand-configuring clusters or schedulers. The engineer picks the study; SimOps handles the machinery.
The same workflow runs on public cloud, private cloud, or the hardware already sitting in your building. Burst to the cloud when a deadline hits; fall back on-premise when it doesn't.
Simulation data - often the crown jewels of a product roadmap - stays under the customer's control. You borrow the supercomputer without handing over the blueprints.
The founding pair reads like a setup to a joke that turns out to be a strategy. One spent his career in the deepest end of high-performance computing. The other spent his running the unglamorous operations that keep software companies alive. Cloud HPC needs both halves - the physics and the plumbing - which is roughly the point.
An industrial engineer by training with an MBA from Koc University, Yenier has been building Software-as-a-Service since the early 2000s and held development and operations leadership roles before co-founding the company. His specialty is the operational side: making HPC and cloud actually usable at scale.
A PhD in applied mathematics, former professor, and one-time Senior Director of Grid Computing at Sun Microsystems. He served on the US President's Council of Advisors on Science and Technology from 2004 to 2008. In January 2025 he shifted to a semi-retired role as SimOps project manager.
The interesting thing about selling automation is that you are selling the absence of work, which is a genuinely hard thing to put on a purchase order. Simr's answer is a number: simulations that go, on average, about ten times faster, with a claimed ceiling around forty times more engineering productivity. Whether those numbers hold in your particular shop is the sort of thing you'd want to test - but the direction is clear enough. Every hour an engineer spends babysitting a scheduler is an hour not spent on the actual product.
There is a reason a corporate venture arm like BMW i Ventures wrote a check. If you build cars, the loop of "design, simulate, test, repeat" is your product-development metabolism. Speed it up and you build fewer physical prototypes, book less wind-tunnel time, and get to market sooner. The investment thesis is not really about software. It is about compressing the calendar.
The rebrand from UberCloud to Simr, in the same breath as the SimOps launch, was a quiet promotion. UberCloud sounded like a place you rented machines. Simr wants to be the operating layer that sits above whatever machines you happen to have - a category, not a vendor. That is an ambitious repositioning for a roughly 66-person company, which is either the fun part or the risky part depending on your temperament.
Burak Yenier and Wolfgang Gentzsch launch UberCloud to move engineering simulation workloads to the cloud.
The company develops award-winning software containers that package engineering apps for portable, reproducible cloud HPC.
Stanford's cardiac simulation runs in UberCloud and wins HPCwire Editors' Choice and Hyperion awards at SC'17.
The company becomes Simr and introduces Simulation Operations Automation as a new category.
Uncorrelated Ventures leads the round, joined by BMW i Ventures and Earlybird Venture Capital.
Co-founder Wolfgang Gentzsch moves to a semi-retired SimOps project-manager role.
Before it powered factories, it simulated a beating human heart.