The Defensibility Imperative: How Public Market Repricing Is Influencing Early-Stage VC Investing
Public market investors are gradually resetting valuations for many recurring-revenue software businesses. This shift is driven not only by macro or regulatory factors, but from a growing conviction that AI will make many traditional software products easier to replicate, automate, or displace.
Growth still matters, but it is no longer sufficient on its own. Investors are placing greater emphasis on durable differentiation through business model durability, pricing power, deep customer dependency, and long-term relevance in an AI-accelerated world.
As a result, defensibility has become central to early-stage underwriting. Growth must now be evaluated alongside a company’s ability to maintain its advantage as competition intensifies, and development costs fall. The emerging imperative is clear: build businesses that are not only scalable, but enduringly hard to displace.
Reframing the Venture Return Equation
Public market comparables continue to anchor a large share of venture exit outcomes. As public software valuations have normalized, this has practical implications for how early-stage returns are underwritten.
At more modest public-market multiples, higher entry valuations at the early stage rely on a narrower set of outcomes to generate strong returns. These outcomes typically involve (a) strategic acquisitions at a massive premium, (b) companies that become clear category leaders, or (c) shifts in public market sentiment on AI that support higher long-term multiples.
None of these paths are unrealistic, but they are inherently less common. As a result, valuation discipline has become more consequential than in prior cycles, not as a constraint on ambition, but as a tool for risk management.
The focus is shifting. Rather than asking only whether a company creates value, investors increasingly need to assess whether that value can be sustained and protected over time.
Value Creation ≠ Value Defense
A recurring pattern is emerging in diligence. Many companies show strong customer validation, clear pain points, and growing revenue. By traditional venture metrics, these businesses are compelling. Yet a deeper question often follows: how difficult would it be to replicate the core offering?
In some cases, modern development tools and foundation models significantly lower the cost and time required to recreate large portions of functionality. When a product’s differentiation rests primarily on an interface layer wrapped around broadly available models like GPT-4 or Claude – “what we call a thin AI wrapper” - barriers to entry can be thinner than they initially appear.
As we wrote in our 2023 piece Navigating the Rise of Generative AI, there is a critical distinction between companies that use AI and companies whose value is created and compounded by AI. The former treat AI as a feature. The latter build systems where AI gets smarter with every interaction, where data creates barriers that rise over time rather than erode.
This does not diminish the real problems these companies are solving. But solving a problem and defending the solution are distinct challenges.
Paths to Defensibility in an AI-First World
We see several forms of defensibility that allow companies to hold up as markets evolve. Not every company needs all of them. But companies without any credible path to defensibility are increasingly difficult to underwrite at early-stage multiples.
1. The Technical Moat: When AI Compounds with Use
“AI” is not a moat. Proprietary data and novel architecture are. For example, our portfolio company Cytrix aggregated the proprietary knowledge and actions of over 5,000 white hat hackers, training proprietary models that orchestrate a human-like pentesting. Cytrix has built a dataset that general-purpose LLMs cannot replicate. That data trains models that determine the next-best-action in security testing. Time and time again Cytrix’s customers mention how accurate, fast, and scalable Cytrix is, citing that they have never seen something like Cytrix in applicative security.
Another example is portfolio company Augmented Intelligence (AUI). AUI’s Apollo-1 build is a prime example of this technical wedge, it uses a neuro-symbolic architecture to combine the linguistic fluency of neural networks with the deterministic reliability of symbolic AI. This “white-box” approach allows for traceable decision-making and policy compliance that standard black-box LLMs simply cannot guarantee in high-stakes B2B environments.
This is what AI defensibility looks like. Not AI as a feature, but AI as a compounding asset.
2. Economic & Distribution Defensibility
Defensibility does not always come from what you build. Sometimes it comes from how you capture value and reach customers.
Traditional per-seat SaaS pricing is under existential pressure as AI agents automate human work. When revenue scales with user count, efficiency gains can compress growth.
The emergence of AI agents has fundamentally redefined the landscape, moving us from simple chatbots to autonomous systems capable of executing complex workflows. When these agents are fueled by high-quality, structured, and proprietary data, the value proposition shifts from incremental to exponential.
More resilient models align pricing with enduring economic flows—transactions, volume, or outcomes that persist regardless of automation. In parallel, companies with privileged distribution—embedded partnerships, ecosystem-level integrations, regulatory channels, or network-driven sales loops—can develop advantages that are difficult to replicate even when products are comparable.
Our portfolio company TULU exemplifies this form of defensibility. TULU provides curated, hyper convenient on-prem goods and services to multi-unit apartment buildings across the United States and Europe. By partnering directly with property owners, TULU is often able to “nest” its platform usage fees within existing rental leases. Landlords get a valuable economic model that increases existing tenant stickiness and enhances property attractiveness to new tenants; TULU becomes a “friendly Trojan horse” and saves time, costs and focus on renewing annual user contracts. Tenants also arguably benefit the most - for a few dollars per month they have access to products that meaningfully enhance their day-to-day quality of life.
In an AI-accelerated world, pricing models tied to durable economic flow and distribution channels that are structurally difficult to access materially reduce risk.
3. Workflow Design and Integration: Owning the Operational Core
Not all defensibility requires proprietary data or revolutionary pricing. Sometimes it comes from solving deeply fragmented, operationally complex problems in industries that resist change.
Our portfolio companies, such as Optibus (public transit optimization) and Wisor (freight forward management optimization, operate in more traditional markets. These industries already have hundreds of point solutions – scheduling tools, route planners, compliance trackers, CRM systems – but almost no end-to-end workflow integration.
This isn’t a moat in the classic sense. A competitor could theoretically replicate the functionality. But the switching costs, integration complexity, and operational risk of migration create real stickiness.
4. The Distinctive Team: Finding the F1 Drivers
The final line of defense isn’t just about the machine or the business; it’s about who is behind the wheel. You can give a mediocre driver the fastest car on the grid, but they will still lose the race when the rain starts to fall, or the track conditions shift. In the current market, we are looking for F1 drivers; founders who possess the “crazy instincts” and full-out passion required to react to changing conditions in real time.
Even with the best technical “car,” a startup is only as good as the driver’s ability to navigate high-speed volatility. This is why we prioritize two specific traits:
Extreme founder-market fit: We back specialists who have spent a decade in their vertical. They don’t need to learn where the turns are; they lived the problem for years and can feel the traction of the market before the data even shows it.
Fiscal agility and adaptability: F1 drivers don’t just go full throttle; they know exactly when to brake and when to accelerate out of a corner. We look for teams that treat their burn rate as a dynamic lever. If unit economics hit a specific threshold or they see a tremendous pull from the market, they must have the instinct to double down. The moment conditions slip, they must have the discipline to pivot and protect the machine, cutting their burn rate to survive to fight another day. This level of agility is the only way to win in a market this fast.
Defensibility Is a Spectrum, Not a Binary
Very few early-stage companies have all of these characteristics. Expecting perfect defensibility at Seed or Series A is unrealistic.
What has changed is the necessity of having a credible path to defensibility. Five years ago, we could invest in a company with strong product-market fit and reasonable growth, trusting that execution and market dynamics would create defensibility over time.
Today, that bet is much harder to justify. As software becomes faster to build with AI and competition increases, and as public markets place greater emphasis on durable advantages, relying solely on emergent defensibility carries more risk than it once did.
Perfection is not required, but clarity is. How does the product or solution get harder to replicate as it scales? What improves with usage? What compounds over time? What elements of the business strengthen barriers rather than lowers them?
The Existential Question Every Founder Must Answer
SaaS isn’t going away, but undifferentiated, feature-driven, per-seat SaaS without clear differentiation carries more risk than it used to be.
The winners in this environment will be companies that can answer one question convincingly:
“Why will this still matter - and still be hard to replace - five years from now?”
Answers based on speed, polish, or being first are rarely sufficient on their own. More durable answers point to advantages that strengthen with scale (e.g., data that compounds, business models aligned with enduring economic flows, or deep integration into workflows that are costly to unwind).
For early-stage investors, this means explicit underwriting of defensibility is no longer optional. It’s existential.
The companies that survive and thrive will be the ones that don’t just create value but compound it in ways that competitors can’t easily replicate.