Authored by SC Moatti, the distinguished founding managing partner of Mighty Capital and board chair at Products That Count, this incisive analysis cuts through the hype surrounding AI to reveal the only two strategic advantages that genuinely fortify a business against the relentless tide of well-funded competitors and rapidly evolving models. Moatti, a venture capitalist lauded on the Kauffman Top 30 Index and Power100, has a formidable track record, having invested in pioneering companies like Amplitude, Netskope, and Groq. Her expertise, honed by developing products used by billions at giants like Meta and Siebel Systems and chronicled in her award-winning bestseller on product development, provides a unique lens through which to examine the dynamics of the modern tech landscape. She holds a master’s in electrical engineering and a Stanford MBA, and is a Kauffman Fellow and member of Young Presidents Organization, underscoring the depth of her strategic insight.

Today, if your entrepreneurial pitch still begins with the phrase “we use AI,” you’re likely describing a component of your infrastructure, not a unique business differentiator. The sheer ubiquity of AI integration across industries underscores this shift; a staggering 97% of products nominated for this year’s Products That Count Product Awards already incorporate AI deeply into their functionalities. This statistic alone signals the definitive end of AI as a standalone competitive edge. The market has moved beyond novelty; AI is now a baseline expectation, much like having an internet connection or a mobile app. For investors, the real challenge lies in identifying a sustainable moat, a barrier that can withstand the onslaught of a generously capitalized competitor who could launch tomorrow with a superior AI model. The question isn’t whether your product uses AI, but what structural advantage prevents others from simply building a better version of it.

To uncover these elusive moats, a rigorous analysis was conducted, leveraging Crunchbase data on 576 venture-backed, AI B2B companies that have successfully raised $50 million-plus rounds since the beginning of 2025. This rich dataset was then meticulously evaluated through the lens of Hamilton Helmer’s seminal “7 Powers” framework, a strategic blueprint for understanding enduring competitive advantages. Layered onto this quantitative analysis were qualitative insights gleaned from Products That Count’s expansive community of over 600,000 product leaders, providing a ground-level perspective on what truly resonates and endures in the market. The findings are unequivocal: in an environment where the cost and complexity of building AI-powered solutions are rapidly approaching zero, the only moats that truly stand the test of time are those that a mere model cannot generate or replicate. These are the structural advantages woven into the very fabric of the business model or network architecture, rather than residing in the ephemeral quality of an algorithm or the transient volume of data.

The analysis starkly reveals that only two of Helmer’s seven powers offer a defensible path without demanding the colossal capital expenditure required to compete directly with giants like OpenAI. Conversely, two other commonly perceived moats are revealed as strategic traps, offering little long-term protection. The remaining power, scale economies, is largely beyond the reach of most startups, a game played by a select few titans.

Counter-positioning: The Moat That Costs Nothing to Defend

Counter-positioning emerges as a potent and often underutilized moat, one that offers formidable defense at virtually no ongoing cost. This power manifests when a newcomer pioneers a business model so fundamentally different from that of established incumbents that the incumbents simply cannot replicate it without severely undermining their own existing economic structures. The classic narrative of Netflix versus Blockbuster perfectly illustrates this: Blockbuster, deeply reliant on late fees, could have adopted a subscription model, but doing so would have cannibalized its primary revenue stream, ultimately leading to its demise. Its rational inaction to protect its existing model inadvertently created Netflix’s impenetrable moat.

In the fast-paced AI era, counter-positioning is rare, appearing in only 5% of the companies in the dataset. However, its scarcity is directly correlated with its value. Investors recognize this unique strategic advantage, pricing it at an impressive median enterprise value of 5.3x per dollar raised—the highest multiple observed across all powers in the analysis. This premium reflects the profound difficulty an incumbent faces in adapting to a counter-positioned competitor.

Consider a vertically integrated AI insurer that sells directly to employers, leveraging AI for hyper-personalized risk assessment and real-time claims processing. This model fundamentally disrupts the traditional brokerage system and its associated commissions. Legacy insurers, deeply embedded with broker relationships and reliant on established underwriting margins, cannot simply pivot to this direct-to-employer, AI-driven model without gutting their existing distribution channels and revenue streams. Their network of brokers, once an asset, becomes a liability in the face of this new paradigm. Similarly, AI-native revenue management systems, designed from the ground up to automate complex pricing and inventory optimization, pose an existential threat to legacy vendors. These traditional players often derive substantial revenue from high-margin consulting services that help clients implement and manage their outdated, complex systems. Adopting an AI-native solution would eliminate the need for these services, effectively cannibalizing their own consulting revenue.

The brilliance of counter-positioning lies in the incumbent’s rational decision not to respond in kind. They see the threat, understand the new model, but are structurally incapable of adopting it without self-destruction. This inertia, born of their own success and legacy systems, becomes the newcomer’s unassailable moat. For founders, the diagnostic question is simple yet profound: "Could a well-resourced incumbent copy your model if they wanted to?" The optimal signal isn’t an outright "no," but rather, "technically yes, but the cost of doing so, in terms of cannibalized revenue, disrupted relationships, and fundamental business model overhaul, would far exceed our cost to build it from scratch." This strategic paradox is the essence of counter-positioning.

Network Economies: The Moat That Builds Itself

Network economies represent another exceptionally powerful moat, characterized by a self-reinforcing dynamic where a product’s value to each user increases proportionally with the number of other users who join. At scale, this phenomenon naturally leads to winner-take-all or winner-take-most outcomes within the network’s defined boundaries, whether geographic, professional, or industry-specific. LinkedIn stands as the quintessential example: as more recruiters join, the platform becomes more attractive to candidates. More candidates, in turn, draw more professionals seeking career opportunities, which then attracts even more recruiters, creating a virtuous cycle of increasing value and lock-in.

In the AI landscape, network economies, much like counter-positioning, are rare, appearing in only 5% of the companies in the dataset. However, they command an impressive 4.2x multiple, making this the most capital-efficient pathway for founders to achieve a substantial valuation premium today. The beauty of network effects is that once the initial "cold-start problem" is overcome, the moat tends to deepen organically, making it progressively harder for competitors to challenge.

The B2B variant of network economies is particularly undervalued and potent. Here, the nodes of the network are not individual users but entire companies. Imagine a platform connecting brands directly with factories, streamlining the entire production cycle from design to delivery, or an advertising exchange linking niche advertisers with highly specific audiences, or a collaborative design platform where partners share resources and expertise. Every new participant – be it a brand, a factory, an advertiser, or a partner – makes the network more valuable for every existing participant. A new factory joining offers more production capacity and specialized capabilities to brands. A new brand joining provides more business opportunities for factories. This interconnectedness creates immense value.

Furthermore, the data generated and accumulated across these myriad interactions—details on costing, production cycles, audience behavior, design iterations—compounds over time. This rich, real-time, and context-specific data becomes an increasingly invaluable asset. While individual data points might be replicated, the intricate web of relationships, the trust built within the network, and the aggregate intelligence derived from millions of interactions become progressively harder to replicate or buy. A well-funded competitor with a superior AI model might be able to process data more efficiently, but they cannot simply buy the network of established relationships and the historical interaction data that underpins its value.

The cold-start problem—getting both sides of a two-sided market to commit simultaneously—is undoubtedly challenging. It often requires significant upfront investment in attracting early adopters and demonstrating initial value. However, founders who successfully navigate this initial hurdle gain ownership of something profoundly powerful: a self-reinforcing system that a competitor, no matter how well-funded or technologically advanced, cannot simply purchase or quickly recreate. This inherent stickiness and compounding value make network economies an unassailable fortress in the AI era.

What Looks Like a Moat But Isn’t: The Strategic Traps

While counter-positioning and network economies stand as robust defenses, several other commonly perceived moats are proving to be less effective in the AI era, often leading founders into strategic traps.

Cornered Resources (Proprietary Data, Unique IP, Exclusive Access): This category, encompassing proprietary datasets, unique intellectual property, or exclusive access to certain resources, accounts for the second-highest prevalence in the dataset at 44%. However, despite its widespread adoption, it yields the worst multiple: a mere 2.6x. The reason for this dismal performance is clear: investors have witnessed countless proprietary datasets rapidly diminish in value due to the advent of powerful foundation models and the rise of synthetic data generation. An AI model, especially a large language model, can often infer, generate, or quickly learn patterns from vast public datasets, effectively eroding the uniqueness of many "proprietary" data advantages. If a data advantage doesn’t compound in ways that become progressively harder to replicate over time – for instance, if it’s not embedded within a network effect or protected by extreme regulatory barriers – it simply cannot be considered a true moat. A static dataset, no matter how large, is increasingly vulnerable to commoditization.

Switching Costs: This is the most crowded power, present in 37% of companies. It often looks like a formidable moat because customers, once integrated, are genuinely reluctant to leave. The sheer effort, time, and risk involved in migrating from one complex enterprise system to another—re-architecting workflows, retraining staff, transferring institutional memory, and mitigating data loss—creates significant inertia. However, the cost of building these switching costs is the problem. Deep enterprise entanglement, which creates this stickiness, requires expensive, long sales cycles and intensive integration efforts. The multiple achieved (4x) is comparable to network economies, but the capital required to reach that valuation is roughly 10x higher. This disproportionate capital requirement makes switching costs a less efficient path for most startups. It remains a viable strategy for founders who can engineer a product-led growth (PLG) motion to drastically reduce customer acquisition and integration costs, or those who can cleverly design their platform to encourage user collaboration, thereby converting passive switching costs into active network economies over time. For example, a project management tool that allows seamless external stakeholder collaboration might start with switching costs but evolves into a network effect as more partners join the platform.

Scale Economies: While intuitively appealing, scale economies are largely not the game most founders are playing. This moat arises when a company’s unit costs decrease significantly as its production volume increases, making it incredibly difficult for smaller competitors to match its pricing. The median multiple for this category, excluding behemoths like OpenAI and Anthropic, collapses from 6.1x to a modest 3.2x. A staggering 88% of the category’s capital belongs to these two AI giants, illustrating the extreme concentration of this power. Believing your unit economics improve with growth is a fundamental business principle, but it does not equate to possessing a true scale moat. The gap between these two concepts is measured in billions of dollars—capital that most startups will never raise. True scale moats in the AI era are reserved for those building foundational models or infrastructure that demands immense computational resources and R&D investment, a realm largely inaccessible to the vast majority of venture-backed companies.

The Only Question That Matters

Every single company within the analyzed dataset, by definition, incorporates AI into its product. Yet, the stark disparity in their valuations and long-term defensibility highlights a critical truth: the companies commanding premium multiples have built something profoundly structural beneath the AI layer that a model cannot generate or replicate on its own.

This "something" is inherent in the business model design or the network architecture. It transcends mere model quality, feature sets, or the sheer volume of proprietary data. It directly addresses the pivotal question every founder must be able to articulate concisely: "What about my business would survive a competitor who starts today with more capital and a better model?"

If a founder cannot answer this question with a clear, structural advantage—whether it’s an uncopyable business model (counter-positioning) or a self-reinforcing user base (network economies)—they are, by definition, building a product. The founders who are truly commanding 4x to 5x multiples in the AI era are not just building products; they are building enduring power, strategically embedding moats that will withstand the relentless forces of innovation and competition. This shift in perspective, from merely integrating technology to architecting defensible power, is the ultimate determinant of success in the AI-driven future.

(Illustration: Dom Guzman)

Stay up to date with recent funding rounds, acquisitions, and more with the Crunchbase Daily.