Successfully developed academic inventions have the potential to spawn new technological domains, form the basis of thriving business ventures, and improve the well-being of society. However, evaluating whether an early-stage scientific invention truly has such potential is extremely difficult, and financially backing such inventions is highly risky. And yet, organizations and their evaluators still back some of these inventions with resources for further development. We investigate this puzzle to pinpoint how and why evaluators decide to offer resource commitments at early stages, despite the red flags raised using standard evaluation criteria. Many academic inventions need these initial resources to dispel concerns regarding their commercial feasibility, so evaluators need to take a leap of faith with their support to prematurely avoid eliminating high-potential opportunities. We tested our theory using text analysis on nearly 700 invention evaluation reports written by a university's technology transfer experts. Our results revealed that evaluators backed inventions based on their feasibility (overcoming doubt and assessing maturity) and desirability (background familiarity and scientific complexity). Using the context of the research laboratory, our study insights can be applied to many management situations in which early-stage opportunities are assessed for resource commitments under high uncertainty.