The Cronus and Zeus Cycle in AI

The myth of Cronus and Zeus captures one of humanity’s oldest themes — the Cycle of Power. A ruler becomes afraid of being replaced → he tries to stop the next generation from rising → yet the successor inevitably emerges and overthrows him.

Cronus overthrew his father, Uranus. Zeus overthrew Cronus. And later, even Zeus feared being overthrown by his own children.

The Cronus and Zeus cycle of power embodies Fabrica’s current dilemma in investing in AI startups.

That same cycle now plays out in Silicon Valley. The power struggle isn’t between gods, but between the LLM labs and the startups built on top of them.

Developers like Cursor (coding), Harvey (law), and OpenEvidence (medicine) have thrived on the capabilities of OpenAI’s and Anthropic’s models — yet those same labs are beginning to compete with their own customers, a familiar cycle in which platform owners move downstream, using pricing power and control to squeeze the startups they once enabled.

The question isn’t if tensions will rise, but how long the uneasy peace will hold.

* AI labs, once the driving force behind the AI startup ecosystem, are now moving into the very markets they helped create. Cursor’s costs jumped after Anthropic raised prices, while both OpenAI and Anthropic launched in-house coding assistants, entering Cursor’s domain.

* Startups are betting that deep domain knowledge will drive true differentiation — building moats from proprietary data, specialized workflows, and hard-earned trust, advantages that general-purpose LLMs can’t easily match. Harvey focuses on complex legal work like M&A coordination, while Cursor retrains its model every two hours with real-time data.

* Unlike traditional SaaS, AI startups face rising costs as usage scales. Their defense lies in smarter routing and performance-based pricing — Harvey, for instance, can justify expensive models because law firms are willing to pay for precision.

Conclusion

Zeus defeated Cronus. And AI startups, along with the VCs backing them, hope their specialization and agility will let them do the same.

In cybersecurity, consolidation has been the rule rather than the exception, and the same will probably await AI — a future shaped by ever-larger Zeuses absorbing those who once nurtured them.