How AI can help Startups Bring Much-Needed Breakthroughs to Engineering Software

By Margret Dupslaff, BMW i Ventures

Engineering software — the digital toolbox that helps engineers design, analyze, and manage product development projects — has existed since the 1950s and 60s. Although advancements have been made since the days of PRONTO, SKETCHPAD, and rudimentary project management tools, artificial intelligence, cloud computing, and collaboration tools are now fueling many necessary improvements to the existing engineering toolset.

As products and requirements change, engineering software must keep pace. Development cycles are shrinking, and faster time to market is critical. At the same time, products must have better performance and meet other KPIs like lower drag, high stiffness, or low cost. Requirements such as sustainable materials and latest manufacturing technologies, such as 3D printing, need to be accounted for early on to avoid high costs and delays later in the process.

A mature industry in need of disruption

Incumbent players like Ansys, Siemens, and Dassault dominate the engineering software space, owning more than 75% of the market. They invest substantially in their solutions, but their merger and acquisition history shows they wait for the right solution to emerge. They then acquire and integrate it with their existing tools. There have been roughly 60 acquisitions of engineering software players since 2020. The most recent acquisition of Ansys by Synopsys is a great example of further consolidation in the market.

Despite incumbent players’ advancements, most of today’s engineering solutions suffer from inefficient manual processes and many iteration cycles, long simulation times, and inefficient collaboration among engineers of different stages in the development process.

Engineering software is a market ripe for disruption, and startups can play a significant role in providing solutions that reduce product development cycles, help engineers design better products, promote more efficient and automated processes, and lead to reduced costs. AI/deep learning, cloud computing, and more advanced collaboration tools are the most promising technologies for disruption.

AI-assisted engineering

In engineering, AI holds promise. Neural networks can predict real, complex physical phenomena in real-time and create multi-dimensional models. Large-language models offer a new way to interact with software, including computer-aided engineering and CAD tools. In practice, designers and engineers could find an optimal design much faster using an AI model which is trained based on all historical designs and their equivalent performance KPIs.

The designer could set the specific requirements (e.g., a tire rim with particular geometries and KPIs such as weight, stiffness, durability, drag, material, and costs) and the AI model could then create a design space based on the historical designs and expand it beyond the original boundaries, calculating confidence intervals for the target KPIs. The AI model doesn’t just replicate old designs; it learns from them and suggests new, potentially groundbreaking concepts.

In this instance, the AI model empowers the designer to explore innovative possibilities and make informed choices. The designer and simulation engineer join forces, leveraging AI-driven insights to discuss, refine, and ultimately create a tire rim that pushes the boundaries of performance, efficiency, and cost-effectiveness. This same technology could simulate structures, crashes, heating and cooling flows, aerodynamics, acoustics, electronic PCB design, etc. This is just one example of how AI can be applied to transform engineering. Celus, Neural Concept, Monolith AI, PhysicsX, and Extrality are all startups innovating in AI-assisted engineering.

Engineering simulations in the cloud

Today, only 10 to 20% of high-performance computing is done in the cloud, but the cloud offers many opportunities, including running more simulations with faster results, greater efficiency, and faster time-to-market. Barriers for companies in moving to the cloud are security concerns, latency, user experience issues, and the complexity of setting up cloud computing. Thus, simulations rely on on-premise infrastructure. This practice is becoming less effective as industrial players are under pressure to increase innovation, and products are becoming increasingly complicated.

Manufacturers need to accelerate development processes and run more simulations. They also need more compute power to train AI models. The more compute power is needed, the more benefits to the cloud versus on-premise computing.

New software tools enable existing simulation software to be used in the cloud. Most use a containerization algorithm that scales simulations to 1,000 CPU cores and 100s of GPUs in the cloud. No additional training is required. In the cloud, simulations can be spread across many CPU and GPU cores, so simulations can run in hours versus weeks, increasing operational efficiency by up to 30x. There is no need for expensive data centers that have capacity constraints.

By taking simulations to the cloud, simulation engineers can see results much faster, leading to increased efficiency and faster development cycles. Startups in the space include SimScale, Rescale, and UberCloud.

More advanced collaboration tools

Today, engineering is hampered by iterative development processes. Designers and engineers initially conceive a small number of designs and then incrementally improve them, considering requirements iteratively to reach the production stage. There is constant back and forth and data transfer among various software tools.

Startups like Synera, spread.ai, valispace, canvas GFX, and CoLab Software are taking a more holistic, connected, automated, and cyclical approach to the problem with single-point-of-truth platforms that integrate all other tools — AI, cloud computing, and existing design and simulation software tools — and eliminate discontinuities. The benefits in data management, process efficiency, knowledge-sharing, quality, and cost reductions are enormous.

Challenges and opportunities for startups

For startups wanting to enter the engineering software space, change management is the biggest challenge. Many engineers are reluctant to change their way of working, and only a few are early adopters who are open to new technologies that can significantly support their daily work. Startups will need to find early adopters that promote the advantages of the latest technologies within the company. Other challenges include long sales cycles and lengthy approval processes for new suppliers and budget cuts and freezes, especially because incumbents have relationships and access to customers, and they can often offer lower-priced entry packages.

Here are four tips for startups to excel in the industry:

• Integration is critical: Do not develop your software as a silo. Integration will make the ultimate difference in user adoption and activity and play a significant role in your exit opportunities. Integrating your solution with other products must be easy, not just at a high level in reading flexible input files or generating output formats needed for following steps but also deep down in your mathematical models.
• Make your ROI clear: Quantify the cost reduction, time saved, and product/performance increase to prove to the customer that your solution is essential to their competitiveness. Companies often struggle to quantify the business impact of new technologies and do not want to invest upfront for the software and implementation/training.
• Be a painkiller: Prove that you are a painkiller rather than a vitamin. You must solve a painful problem for customers and not just be “nice-to-have”. Otherwise, budget cuts will affect you harshly, and the sales cycle will be lengthy.
• Know your core (feature vs. product): Have a clear perspective on whether your product will remain a stand-alone product or be converted into a feature integrated into other existing products. If you offer a feature, what makes you confident you can build a business on it?

Parting thoughts

AI, cloud computing, and new collaboration tools are pivotal to much-needed evolution in the engineering software space. Startups that embrace these trends and technologies for growth and differentiation stand to win — if they can solve a painful problem, clarify their differentiation, demonstrate ROI, and be easily integrated into existing environments.

Engineering impacts all of us. It’s time for more startups to provide better tools. Let’s help engineers create products that meaningfully transform our quality of life.

‍