Nov. 2, 2025 /Mpelembe Media/ — An article from The Economist titled “OpenAI and Anthropic v app developers: tech’s Cronos syndrome,” examines the emerging competitive dynamic between large language model (LLM) providers, such as OpenAI and Anthropic, and the specialised AI application developers that build their businesses atop these models. The article uses the metaphor of Cronos devouring his children to illustrate the fear that the powerful, highly-valued AI labs may eventually usurp the profits of the smaller app-makers like Cursor and Harvey.
While the smaller companies aim for artificial specialised intelligence to create a competitive moat through unique data and outcome-based pricing, the article notes the labs are already developing rival tools and raising costs, posing an existential threat. However, the app developers’ long-term survival is suggested by an HSBC estimate predicting that software vendors will capture the majority of the AI-enhanced IT services market by 2030, limiting the LLM providers’ dominance.
The competitive advantages that app developers are building to survive the commoditization and dominance threats posed by AI labs (a dynamic referred to as the “Cronos syndrome,” where the titan tries to devour its children) centre on deep specialisation, data moats, and sophisticated cost management techniques.
Here are the key strategies app developers are employing to retain their edge:
1. Specialisation and Focusing on “Artificial Specialised Intelligence”
App developers believe their path to survival lies in focusing on “artificial specialised intelligence” (ASI), which is AI tailored to a specific field, such as law or medicine, rather than striving for artificial general intelligence (AGI) like the AI labs.
- Targeting Niche, Complex Business Processes: App developers are focusing on complex business processes that present significant challenges and opportunities but are not generalisable. Value is sought in “the most boring, mundane thing…hidden in the back of a company [that is] slow, expensive and consequential”.
- Tackling High-Value Complexity: They are aiming for tasks that general-purpose models will find harder to replicate. For instance, Harvey, an AI app for law firms, focuses on complex tasks like helping coordinate several law firms during a mega-merger, moving beyond boilerplate assignments such as generating non-disclosure agreements. This depth of focus is where they hope to retain a competitive edge.
2. Building Data Moats and Customer Stickiness
App developers are leveraging their longevity in business to build data-driven advantages that make their services “stickier” with customers.
- Acquiring Specialised Data: The argument is that the longer the app developers remain operational, the more specialised data their agents will acquire, consequently improving their performance. This effect is likened to self-driving cars becoming more reliable the more miles they travel.
- Creating a Competitive Moat: This continuous acquisition of specialised data creates a “competitive moat that the ai labs will struggle to cross”. For example, Cursor, an app that helps developers write code, uses real-time data to update its own Large Language Model (LLM) every two hours, which it believes enhances the coding experience for its customers.
3. Implementing Cost Management and Revenue Model Innovation
App developers must manage high marginal costs that increase as they grow due to reliance on LLMs. To offset these rising costs, they are experimenting with new techniques and revenue models.
- Diversifying Models: One technique involves using a variety of models, including open-source ones. This allows the app developers to route the simplest queries to the processing option that is cheapest.
- Outcome-Based Charging: Instead of charging customers based on usage of the underlying models, some developers are opting to charge based on outcomes. This allows companies like Harvey to justify opting for the largest and most expensive models because law firms are willing to pay for the resulting “perfect accuracy”.
By focusing on specialised AI and acquiring unique data that improves performance over time, app developers are trying to carve out defensible positions, contrasting with the general models provided by the AI labs. This strategy is based on the idea that the majority of the market share (70% by 2030) will belong to software vendors utilizing LLMs, while the LLM providers themselves will capture only about 30% of the overall market for AI-enhanced IT services.
In summary, the “Cronos syndrome” is like a seed company that sells specialized seeds to independent farmers. The seed company (the AI lab) realizes the farmers (the startups) are producing massive harvests (profits). The seed company then starts developing its own automated farming systems and sells the seeds at a higher price, hoping to push the farmers out and claim the entire agricultural value chain for itself. However, the farmers, by growing unique, specialized crops and knowing the local soil better, may still survive and prevent the seed company from monopolizing the entire market.