Alexander Kardos-Nyheim, an angel investor writing in a guest commentary for Crunchbase News, argues that the most significant long-term value in AI will come from startups tackling deep technical challenges at the model and infrastructure level. This stands in contrast to companies building application-layer products on existing AI platforms.
Kardos-Nyheim shares the specific processes and questions he uses to evaluate the investability of an AI startup. His framework prioritizes technical depth and foundational work over product speed-to-market.
The commentary suggests that as AI platforms become commoditized, the barriers to entry for application-layer startups shrink, potentially diluting their competitive advantage. Infrastructure and model-level companies, by contrast, may build moats through proprietary technology and research.
This perspective challenges the prevailing venture capital trend of funding AI applications that ride on top of large language models. Founders may need to rethink their pitch to emphasize technical differentiation rather than just market timing.
Kardos-Nyheim’s own experience includes selling his AI startup before it generated revenue, giving him direct insight into what investors may overlook in early-stage AI companies.