Counter-Positioning and Network Effects: The Only AI Moats That Work

The differentiation crisis in AI

AI is no longer a competitive advantage. When 97% of products nominated for this year's Products That Count Product Awards integrate AI deeply, claiming "we use AI" describes infrastructure rather than a defensible business position.

Mighty Capital analyzed Crunchbase data on 576 venture-backed AI B2B companies that raised rounds of $50 million or more since early 2025, applying Hamilton Helmer's 7 Powers framework. The research, first reported by Crunchbase News, reveals which competitive moats survive when well-funded competitors can launch with superior models at minimal cost.

Two moats command premium valuations

Counter-positioning appears in just 5% of companies analyzed but commands the highest valuation multiple: 5.3x enterprise value per dollar raised. This power emerges when a newcomer builds a business model so structurally different that incumbents cannot copy it without destroying their own economics.

Vertically integrated AI insurers selling directly to employers exemplify this pattern. Traditional brokers see the threat but cannot replicate the model without cannibalizing broker relationships and underwriting margins. AI-native revenue management systems create similar dynamics—legacy vendors would gut high-margin consulting revenue by matching them.

Network economies, also present in only 5% of companies, earn a 4.2x multiple. The B2B variant connects companies rather than individual users: brands to factories, advertisers to audiences, platforms to partners. Each new participant increases value for existing participants, and the accumulated interaction data compounds in ways competitors cannot replicate by launching with better models alone.

Why it matters

Founders and investors face a capital allocation crisis. The research shows that what appears defensible often isn't: proprietary data and exclusive access, present in 44% of companies, earn just 2.6x multiples because foundation models and synthetic data erode data advantages. Switching costs, the most crowded moat at 37% prevalence, require 10x more capital to achieve comparable multiples to network economies. Scale economies collapse from 6.1x to 3.2x when OpenAI and Anthropic are excluded—88% of that category's capital belongs to those two companies.

The structural test

The diagnostic question for founders: What about your business would survive a competitor starting today with more capital and a better model?

Counter-positioning works when the answer is "they could technically copy us, but it would cost them more than it costs us to build." Network economies work when the answer involves compounding data and relationships that accumulate across connected participants.

Switching costs can work for founders who engineer product-led growth to reduce sales cycle expenses or who convert stickiness into network effects by making users collaborate on the platform. Scale economies remain viable only for companies raising billions.

The companies commanding 4x to 5x valuation multiples have built something underneath the AI that models cannot generate: structural advantages in business model design or network architecture rather than in model quality, feature sets, or data volume.

SC Moatti, founding managing partner of Mighty Capital and board chair at Products That Count, detailed these findings for Crunchbase News. The analysis drew on insights from Products That Count's community of over 600,000 product leaders.