Startup OnScale to Offer SaaS CAE

Having secured $3 million seed funding, startup OnScale today emerges from stealth mode. It plans to offer computer-aided engineering (CAE) on demand, urging customers to “run as many simulations and optimization studies as you need and pay only for the Cloud HPC core hours that you use.”

OnScale is a spinoff of Thornton Thomesseti, a consultant providing engineering design, investigation, and analysis.

The company is coming out of the gate as a new SaaS, but it’s not Software as a Service—it’s Solver as a Service, the company clarifies in its press release.

“OnScale has built the world’s first Solver-as-a-Service platform with advanced computer-aided engineering (CAE) multi-physics solvers that are seamlessly integrated with a scalable Cloud High Performance Computing (HPC) platform,” according to the company’s announcement. “OnScale’s solvers have been carefully architected and maintained by the same expert team over decades. Combined with the limitless HPC resources available on the cloud, OnScale’s platform breaks cost and performance barriers for engineers solving tomorrow’s toughest engineering challenges.”

OnScale showing results from complex weld structures used in high-pressure vessels and pipelines in modern power plants (image courtesy of Onscale).

OnScale analysis of SAW-based GHz RF filter showing electrodes, materials and surface wave behavior (image courtesy of Onscale).

Solver as a Service

The solvers currently available from OnScale are structural and electromechanical. Users are charged per core-hour, a calculation based on the number of computing cores deployed to tackle the simulation job.

“These are solvers our team has developed for different engineering projects over a decade,” explained Ian Campbell, OnScale’s CEO. “We’d never offered them publicly, but now we are.”

OnScale offers a free trial account with 10 core-hours per month for those who wish to try out the solvers, Campbell explained. “We’ll also be offering subscription models, similar to the way cellphone plans work,” he added. “So you get a set number of core-hours each month, but if you use more than the allotted number, you pay additional per core-hour fee.”

OnScale solver displaying stress field in high pressure jet engine turbine blade with complex materials and internal cooling structure (image courtesy of Onscale).

Pay Per Usage

OnScale’s approach is different from its near-rival Rescale, which offers HPC resources on-demand but the user must bring its own simulation software licenses and solvers to the transaction. In contrast, Onscale is offering the solver as part of the on-demand package.

Each approach has different appeals and advantages. Engineering firms that have invested in simulation software licenses and prefer to remain in the familiar CAE environment may favor Rescale’s offerings. On the other hand, startups and new companies that don’t have simulation licenses or aren’t interested in purchasing them may prefer OnScale’s solver-as-a-service package.

“When the customer hits run, we create just enough virtual HPC power to solve that specific problem,” said Campbell.

This approach may also prove appealing to project-based teams. Whereas a subscription is an ongoing cost whether the team uses the platform or not, OnScale’s pay-per-use style limits the cost to the required computing power for the job at hand.

OnScale is now officially in public beta phase. The company is initially targeting the following verticals:

• driverless car systems
• biomedical
• IoT
• RF filters and switches

“We’re starting out with these because we think they are underserved by established simulation vendors,” said Campbell. “But ultimately we want to work our way up to the planes, trains, and automobiles—the heavy simulation user market.