The Seven Powers in the Age of AI

In our last post, we took a look at Hamilton Helmer’s 7 Powers, the definitive lens for understanding durable advantage. We argued that when comparing Saas with Marketplaces, the latter consistently build deeper, more durable moats based on deep network effects, reputational switching costs, vast data resources, and process power.   

We ended with a critical question: how does AI change this framework?    

This week, we take a shot at answering that question. AI is impacting SaaS businesses and Marketplaces asymmetrically. AI is fundamentally weakening the moats that have protected SaaS businesses for two decades, while simultaneously reinforcing the competitive advantages of well-designed marketplaces. This is not hyperbole. It is playing out in real time, reshaping billions in enterprise value and will in our opinion rewrite the venture capital playbook in time.

This asymmetric impact is our central thesis. We are witnessing a recalibration of defensibility, where AI is not simply a new feature but a structural force. AI threatens to commoditize what SaaS sells (codified workflows, bespoke information access) while helping to better solve what marketplaces must manage (human friction, complex data, operational risk).

This analysis examines each of the seven powers, comparing the impact of AI on the SaaS model versus the marketplace model.

1. Scale Economies

A business where per-unit costs decline as production volume increases.   

For SaaS companies, scale has historically meant spreading fixed costs such as cloud infrastructure, R&D, and support costs across more customers, driving margins from to 70% or higher. AI disrupts this advantage by democratizing software creation itself. Tools like GitHub Copilot, Cursor, and Replit now allow small, nimble teams to ship software at velocities that once required armies of engineers. When a three-person startup can prototype and deploy functionality that would have taken Salesforce six months and fifty engineers, scale no longer guarantees a defendable cost structure.

For marketplaces, scale works differently. More users do not just reduce costs, they improve the product. AI supercharges this dynamic by improving operational leverage across trust and safety, verification, fraud detection, matching quality and shipping. Each additional transaction feeds more data into AI powered systems that make the platform better, not just cheaper. Consider Mercor, an AI-first talent marketplace. Instead of relying on static resumes and human recruiters, Mercor uses AI-powered interviews to pre-vet candidates at scale. Every interview transcribed, every evaluation made, every successful hire completed feeds the system, making it better at matching candidates to roles. This is not about cost reduction, it is about creating a data flywheel that gets smarter with scale. The more candidates Mercor interviews, the more accurate its assessments become, the better its matches, and the stickier its marketplace.

2. Network Effects

The product becomes more valuable as more users adopt it

SaaS companies have relied on intra-company network effects (think Slack or Figma), where value grows as more colleagues adopt the tool. AI agents are now undermining this by enabling cross-platform interoperability. With protocols like Model Context Protocol (MCP) and Agent-to-Agent (A2A) , an AI agent can understand user intent (e.g., "draft a Q3 marketing plan") and execute the task by pulling data from multiple, disparate systems (e.g., Salesforce, Google Analytics, Notion) without the user ever opening the app. This directly attacks "same-software" network effects. The underlying SaaS app becomes an interchangeable, commoditized backend, and user loyalty shifts to the agent brand.

For marketplaces, network effects have always been stronger than for SaaS, and AI makes them even more powerful. The two-sided dynamic remains: more buyers attract more sellers, and vice versa. AI enables marketplaces to enhance their network effects. Uber's dispatch algorithms, Airbnb's dynamic pricing, and Amazon's recommendation engine all use AI to increase liquidity, getting more drivers on the road during peak hours, more hosts listing during high-demand periods, and more products in front of the right buyers. Better matching drives more transactions, which generate more data, which improves matching further while amplifying liquidity.

3. Counter-Positioning

Adopting a superior business model that incumbents avoid because it would harm their existing operations.

For SaaS, AI creates a classic innovator’s dilemma. Incumbents feel constrained by their focus on serving the needs of existing customers. Rewriting offerings to be more AI-native could undermine the core product and make it easier for upstarts to challenge them with leaner, pure agentic models. For a company like Salesforce, retooling the entire platform to be AI-native would require massive investment, customer re-education, and potentially lower revenues if AI-powered automation reduces the need for per-seat licenses. Meanwhile, AI-native challengers like Harvey (legal AI) and Cursor (coding AI) are capturing market share. 

For marketplaces, counter-positioning is far less threatening. A new AI-first marketplace cannot simply show up and replicate the supply side. Airbnb's hosts, Uber's drivers, and Zillow's property data took years, sometimes decades, to aggregate. An AI interface improves the experience, but it does not solve the cold-start problem of building liquidity.​ Moreover, marketplaces can adapt to new segments without cannibalizing core revenue. Zillow can add AI-powered search and valuation tools without undermining its agent subscription model. Hemnet can layer AI-enhanced marketing packages on top of its seller-pays pricing. There is no innovator's dilemma because AI makes the existing model better.

4. SWITCHING COSTS

Switching costs are the expenses (in time, money, or effort) a customer incurs when moving from one product to another.

SaaS switching costs have historically come from workflow integration, data lock-in, and team familiarity. Migrating from Salesforce to HubSpot meant months of data extraction, API reconfiguration, and user retraining. AI is lowering these barriers dramatically.​ AI-powered data migration tools can now map schemas, transform datasets, and automate integrations in days instead of months. If an AI assistant can pull customer data from Salesforce, sync it with a new CRM, and maintain workflows across both, the cost of leaving Salesforce drops. Enterprises are also rethinking the build-versus-buy decision; with AI coding tools, custom internal solutions that are perfectly tailored become feasible for companies that previously relied on off-the-shelf SaaS. 

For marketplaces, switching costs are structural, not technical. They are rooted in trust, reputation, and liquidity. AI amplifies this stickiness by making these reputational assets more valuable and trustworthy. Zillow’s two decades of behavioral data makes its recommendations more accurate than any newcomer, even if both use the same LLM at the interface layer. Similarly, DoorDash leverages AI to predict ordering preferences and delivery timing so precisely that customers rarely defect even when prices rise. Airbnb reinforces stickiness with AI-verified reviews and personalized matching based on prior stays. 

5. BRANDING

Customers are willing to pay more or choose one business over the other because of trust, identity, or reputation

SaaS branding has historically signaled reliability and ROI. If you buy Salesforce or HubSpot, you know it works. But brand loyalty in B2B software is shallow. Purchase decisions are rational, feature-driven, and price-sensitive. If a competitor offers comparable functionality at a lower cost, switching is justified. AI threatens to erode SaaS branding further by decoupling workflows from interfaces. If users increasingly interact with AI agents that orchestrate tasks across platforms, the underlying software becomes invisible. Whether the backend is Notion, Asana, or a custom tool, matters less if the agent handles everything. Brand equity built on user interface familiarity weakens when the interface is a conversational AI.

For marketplaces, branding is synonymous with Trust. AI reinforces this by enabling trust at scale. Marketplaces can now use AI to more deeply verify listings, detect fraud, moderate content, and resolve disputes faster and more accurately than human teams ever could. This enhances brand reputation, making the platform synonymous with safety and reliability. Rightmove, for example, commands over 80% of UK property search time because agents and buyers trust it; it uses AI to flag potential fraudulent listings and enhance overall platform security, strengthening brand reputation.

6. Cornered Resource

A cornered resource is preferential access to a coveted asset that competitors cannot easily obtain.

Few SaaS companies hold a true data moat. Most rely on usage data which has been called "the empty promise of data moats" because it's often replicable and lacks deep, proprietary signals. There are exceptions: companies with unique datasets (like Palantir in defense) or proprietary algorithms, can still defend their position. But for most SaaS businesses, the cornered resource significantly weakens. 

Marketplaces, by contrast, sit on irreplaceable data assets. Every transaction, search, click, review, and interaction generates proprietary behavioral data that AI can now unlock at scale. This is not just transactional data, it is intent data, preference data, and reputation data, all of which become more valuable with AI. AI also enables marketplaces to capture multi-modal data that was previously inaccessible. Image recognition on fashion marketplaces can extract attributes from product photos (e.g., "V-neck," "floral print," "midi length") without manual tagging. 

7. Process Power

Process power exists when a company's embedded processes enable lower costs or superior products that competitors cannot easily replicate.

SaaS companies have long relied on institutional knowledge, the accumulated expertise of working with thousands of customers, understanding their workflows, and embedding best practices into software. Salesforce knows how enterprise sales teams work. HubSpot knows inbound marketing. This knowledge justified premium pricing and sticky relationships. AI undermines this by codifying and democratizing institutional knowledge. If a startup can train an AI on best practices and deploy a product that automates what SaaS companies spent years learning, the institutional knowledge moat evaporates. Custom configurations that once required six months of professional services can now be generated by AI in days.

For marketplaces, process power is about operational sophistication, the ability to onboard supply, manage liquidity, prevent fraud, and maintain quality at scale. These processes are learned through experience, not copied from competitors. Used car marketplaces like Motorway use AI to learn the right price to buy and sell vehicles based on hundreds of thousands of past transactions. The more cars they trade, the more accurate their pricing models become. Logistics marketplaces use AI to learn optimal delivery routes based on historical traffic, weather, and demand patterns. Every completed delivery makes the system smarter.

Conclusion: What Defensibility Looks Like Now

Viewed through the AI lens, the 7 Powers reveal which moats endure, which ones dry out, and where new leverage emerges.

For a decade, SaaS moats looked unassailable. With AI, software becomes faster, leaner, and more universal, but its advantages are now shared rather than defended. Marketplaces are the mirror opposite. AI is a compounding force for them, solving their oldest and hardest operational problems at scale. It's the key that finally unlocks the immense, latent value in their cornered resources, the proprietary data from transactions, searches, and chats.   

For investors, the playbook is shifting. The search for durable advantage may now be moving away from the SaaS models that defined the last decade, and toward the AI-first marketplaces that will define the next.