Databricks co-founder: All software monopolies will collapse within 24 months

The Gist

• Databricks hits $6.9B revenue run-rate with AI products alone at $1.7B
• Tavakoli predicts no current software monopoly will survive the next two years
• Enterprises are 'token maxing' with no clear ROI measurement on AI spend
• Traditional BI is effectively dead as AI reshapes analytics

• All software monopolies will collapse within 24 months due to the rise of AI-native startups and reduced migration costs.
• Enterprise AI adoption is currently characterized by high token spend without clear ROI, creating opportunities for vendors who tie AI to measurable outcomes.
• The bottleneck for enterprise AI isn't model quality but context - maintaining live, updated business knowledge that agents can use.
• Databricks' Genie Ontology solves the context problem by continuously extracting and updating business knowledge from various sources.
• Traditional BI tools are being replaced by AI-powered systems that provide real-time answers to business questions for all employees, not just data analysts.
• The cost of switching software vendors has dropped dramatically due to LLMs' ability to understand and migrate legacy systems quickly.

• Focus on tying AI solutions to clear, measurable business outcomes rather than just selling AI capabilities.
• Invest in building or acquiring technology that maintains live business context for AI systems, as this is the key bottleneck in enterprise AI adoption.
• If competing against incumbents, leverage lower migration costs and price aggressively while demonstrating clear outcomes.
• If defending market position, accelerate AI-native product development as lock-in strategies are becoming ineffective.

• $6.9B (Databricks' current revenue run-rate)
• 80%+ (Databricks' year-over-year growth rate)
• $1.7B (Run-rate of Databricks' AI products alone)
• 140% (Databricks' net retention rate)
• 30 days (Time Databricks takes for enterprise-grade migrations)
• 70,000 (Number of users a car manufacturer loaded onto Databricks Genie)