BeyondMath secured $8.5 million for the development of its digital wind tunnel

The advent of generative AI occasionally makes it difficult to remember that what is now commonly known as 'artificial intelligence' encompasses much more than text and media generators. In particular, the dominance of generative AI is partially responsible for the conflation of artificial intelligence and machine learning, especially considering that the latter's applications are not as attractive to a wide audience as generative AI's. BeyondMath's digital wind tunnel is a prime example, as we usually do not think much about how things like cars, planes, and boats are made.

When design teams plan a vehicle's feature, they think about what may work, in terms of the physics and corresponding math behind the desired feature. Then, since the math tends to be mind-blowingly complex, they usually verify their work by running a simulation on a powerful computer, which will take some time to determine if what the design team came up with is possible. If this is so, the next step is to test the feature under conditions closer to what one would expect in the real world. This is done with the help of a wind tunnel, and at times, it may turn out that the wind tunnel test does not agree with the computer simulation, which means that the team needs to go back to the drawing board. Because of the resource-consuming nature of this process, this loop can be repeated a small number of times, thus limiting the improvements that can be done on a single feature at once.

BeyondMath's digital wind tunnel leverages machine learning to overcome these issues. Due to the scarcity of simulation data, BeyondMath trained its model on how physical wind tunnels work, and real-world data of fluids interacting with solid objects. The resulting model not only runs standard simulations to verify that the math checks out; it can also provide close to real-time simulations of the physical phenomena modeled by the equations, thus accelerating the engineering process while saving valuable resources and cutting costs for organizations.

One of BeyondMath's early adopters is Formula 1, and the company has shared that some unnamed F1 teams approached it to speed up their aerodynamics testing and development cycles. The company has also publicized its $8.5 million seed round, led by UP.Partners, and supported by Insight Partners and InMotion Ventures. According to BeyondMath, the funding will be directed to grow its team and scale its computing resources as the company continues transforming the engineering process.