SC Moatti, founding managing partner at Mighty Capital and chair of the board of Products That Count, says that the phrase “we use AI” is no longer enough to distinguish startups. With AI integrated into 97% of the products nominated for this year’s Products That Count awards, AI has, in her view, become closer to available infrastructure than to a barrier preventing competitors from catching up with a company.
Moatti bases this conclusion on an analysis of Crunchbase data covering 576 business-focused AI companies that raised funding rounds worth $50 million or more since the beginning of 2025. The analysis used the “Seven Powers” framework developed by Hamilton Helmer, along with data from the Products That Count product leaders community, which has more than 600,000 members.
First Advantage: Counterpositioning
Moatti defines “counterpositioning” as building a radically different business model that an established competitor cannot imitate without damaging its existing economics. She cites the familiar comparison between Netflix and Blockbuster: Blockbuster could have adopted a subscription model, but doing so would have undermined the late-fee revenue on which its stores depended.
According to the data presented in the article, only 5% of the companies analyzed use this type of barrier, but they achieved a median enterprise value of 5.3 times every dollar raised, the highest multiple among the powers analyzed. The concept appears, according to the examples cited, in AI-based insurance companies that sell directly to employers and in revenue-management systems built on AI from the outset.
In the first case, traditional insurance brokers may find it difficult to imitate the model without damaging their relationships and underwriting margins. In the second case, moving to a new model could undermine the high-margin consulting revenue of traditional systems providers. The author argues that the established competitor’s inability to act, because doing so would harm its existing business, is itself the barrier.
Second Advantage: Network Effects
In a network-effects model, the value of the product increases as new users join it. Moatti cites LinkedIn, where having more recruiters attracts candidates, then more professionals, which in turn attracts more recruiters.
Network effects appear in only 5% of the companies studied, but they achieved a multiple of 4.2 times, making them, according to the analysis, the most capital-efficient path to a strong valuation premium. The author argues that their application in business markets is not limited to individual users; they may connect brands and factories, advertisers and audiences, or platforms and partners.
As pricing data, production cycles, and audience behavior accumulate through these interactions, imitating the network becomes more difficult. The practical challenge, however, remains persuading both sides of a two-sided market to commit at the same time. Moatti says that a company that succeeds in overcoming the cold-start problem already possesses an asset that a well-funded competitor, even with a better model, cannot buy directly.
Barriers That Appear Strong but Are Less Durable
Constrained resources, such as proprietary data, intellectual property, and exclusive access, rank high in prevalence, appearing in 44% of the companies studied. However, their multiple was the lowest, at 2.6 times. The author explains this by arguing that foundation models and synthetic data have weakened the ability of many proprietary datasets to remain an advantage that is difficult to copy.
Switching costs appear in 37% of the companies. They may indeed prevent customers from leaving, but achieving them generally requires deep integration with the organization, costly sales cycles, and the risks of redesigning systems. Their multiple is approximately 4 times, close to that of network effects, but the capital required to build them is about 10 times greater, according to the article.
Moatti argues that switching costs may become more efficient if a company relies on product-led growth or designs a reason for users to collaborate within the platform, gradually turning this barrier into a network effect.
What Does This Mean for Startups?
The article also questions whether economies of scale should be considered a strategy available to most founders. The median multiple for this category falls from 6.1 to 3.2 when OpenAI and Anthropic are excluded, and the two companies account for 88% of the capital in this category. The author therefore distinguishes between improving unit economics with scale and possessing a barrier that is actually based on scale.
Editorial reading from certi.news: This article does not provide evidence that counterpositioning and network effects are the only possible barriers in every market, because it is based on a specific sample and valuation multiples rather than a direct causal test. However, it raises an important practical question for AI companies: what will remain of the business if today’s competitor begins with more capital and a better model? According to the argument presented, the answer must lie in the design of the business model or the structure of the network, not in the mere presence of AI.