The core message of Peng’s founder-focused master class is that hard-tech startups can build better physical products faster by using digital twins not only as engineering tools, but also as commercialization tools. After Elisa introduced the session as part of a broader look at hardware, physical products, and the organizations needed to build them, Peng grounded the talk in her own experience: she founded Transformix Engineering in her basement in 1995, ran it as CEO for more than two decades, took it global, exited successfully, and later helped take another technology from zero to more than $100 million in sales in under four years. Her stated aim was to share what worked, what failed, and the lessons she “paid for” as a founder and operator.

Peng’s central story began in 2012, when Transformix nearly went bankrupt after losing 95% of its revenue from one contract. Rather than shut down and distribute profits, the team invested in a new high-speed automation technology that a former VP called the “holy grail” of the industry. The problem was the classic hard-tech chicken-and-egg dilemma: they needed money to build an MVP, but customers wanted proof before paying. Instead of asking customers to trust them, Transformix built a virtual machine, designed the customer’s product inside it, and simulated the assembly process. This let the company demonstrate performance before cutting metal, sell its first MVP to an early adopter, and effectively have that first customer fund part of the R&D.

From there, Peng framed hard-tech founders as fighting a “three headed dragon”: long development cycles, heavy capital requirements, and the difficulty of finding a first customer willing to take a risk. She emphasized that even after finding early adopters and proving the technology, Transformix still almost went bankrupt because the gap between a working technology and a scalable business is large. Hardware makes assumptions expensive: each physical build-test-break-redesign cycle burns cash, while digital iteration can answer questions about performance, collisions, failures, manufacturability, and cost before ordering parts. Physical prototypes still matter because “physics does get the final vote,” but founders should preserve cash by validating as much as possible digitally before building.

Peng then showed how the same lesson applied after ATS acquired Transformix’s IP. She joined ATS on the day the world shut down because of COVID, when customers could no longer travel to see automation systems and trade shows stopped. Returning to the playbook she had used at Transformix, she had engineers develop a digital twin of the acquired technology, used it to demonstrate the concept remotely, and helped move the business from zero to over $100 million in sales in less than four years. Her concluding argument in the available transcript was that founders should reduce how much imagination they ask from customers and investors: showing a credible digital representation can make complex hard-tech products easier to understand, easier to trust, and faster to commercialize.