In the race for AI market share, companies are seen going in two routes: First are the proprietary AI vendors, such as OpenAI/Microsoft, Anthropic/Amazon and Google. These firms have technologically unique approaches for their foundation model training. The second option pursues an open-source model, and includes firms like Bytedance, Alibaba Cloud, and DeepSeek.
Understanding which route offers the upper hand rests on first exploring the paradox of being a technologically unique company: Yes, unique tech has a competitive advantage by making it difficult for competitors to duplicate it. But at the same time, the firm is unable to enjoy spillover effects from similar tech, and equity analysts may not understand it — causing them to drop coverage of the firm.
A new study published in Strategic Management Journal provides empirical evidence of this cost-benefit paradox, detailing the pay-offs and the double penalty of being a contrarian company. The findings can help organizations weigh the factors in deciding whether to invest in unique tech.
The research team — including Yang Fan of Colby College, Lubomir Litov of the University of Oklahoma, Mu-Jeung Yang of the University of Colorado, and Todd Zenger of the University of Utah — worked around the central question of whether technological uniqueness leads to better firm performance. To explore the topic, they designed a novel measure of technological uniqueness by way of patents: The less correlated the types of technology of a focal firm are with the technology types pursued by its competitors, the more unique the focal firm’s patent portfolio is. Then, the team looked at how the uniqueness of a firm’s patenting vector affects the scope of knowledge spillovers it enjoys, the cost of capital it pays, and the financial performance it generates.
The study determined that greater technological uniqueness is associated with better firm performance — and actually that less technological uniqueness results in underperformance; however, their work also highlighted the costs of technological uniqueness. Such firms enjoy fewer spillovers from their competitors, and equity analysts struggle to recognize value in these technologically unique firms, making analysts more likely to drop coverage of the firm.
Yang provides a few factors to consider before investing in unique technology. First, he suggests firms determine the importance of technological knowledge spillovers: How much progress will there be in the common industry technology? Another factor to consider, he says, is capital intensity: “If you are a technology company in a capital-intensive industry, you should think twice about going for a technologically unique path,” he says. “On the other hand, if most of your computing needs can be met by cheap cloud computing, and you do not need to buy hardware, the need for raising expensive capital is lower.”
So where does that leave our AI companies?
“From the perspective of our paper, the proprietary vendors such as OpenAI/Microsoft and others are facing a double challenge: higher cost of capital and loss of incoming learning opportunities,” Yang says. “So their bet really rides a lot on their internal AI models becoming better quicker than free open-source models — a bet that looks shakier, especially since ChatGPT-5 was such a disappointment.”
