From SaaS to AI-First: How Companies Are Reshaping Innovation

What It Covers

Elad Gil and Sarah Guo examine whether AI is genuinely killing SaaS or whether market panic is misreading short-term signals. They analyze AI revenue growth velocity, token cost collapse, vendor durability, and how founders should think about exits and defensibility in a rapidly shifting competitive landscape.

Key Questions Answered

• •SaaS Displacement Reality: Vibe-coding replacing enterprise software is overstated for large organizations. A Fortune 500 company will not rebuild its CRM over a weekend, and enterprise sales, change management, security compliance, and multi-stakeholder workflows create structural barriers that internal AI-generated code cannot realistically overcome in the near term for complex, scaled deployments.
• •Revenue Velocity Benchmark: AI labs moved from $1B to $10B revenue in roughly one year — compared to 20+ years for Adobe and 8-9 years for Salesforce. Projections show labs reaching $100B in 3-5 years versus 27 years for Microsoft. Founders should recalibrate what "late stage" means given this compression.
• •Token Cost Collapse: GPT-4-level inference dropped from $37 per million tokens to $0.25 in 21 months — a 150x reduction. O1-equivalent models fell from $26 to $0.30 per million tokens in 11 months — an 88x drop. Founders building on AI should model aggressive cost reduction curves into their unit economics and pricing strategy.
• •Code Quality as Unsolved Problem: Abundant AI-generated code creates a production fragility risk when no engineer deeply understands the codebase. Testing, smart review, formal verification, and agent-assisted auditing are all partial solutions. This gap represents an open market opportunity for tooling that manages human attention allocation across AI-generated codebases.
• •Exit Timing Framework: Schedule a dedicated board meeting once or twice annually specifically to evaluate exit opportunities — removing emotion from the decision. Most companies have roughly a 12-month window of peak valuation. Competitive dynamics, lab forward-integration, and capability jumps can reset category leadership rapidly, making pre-scheduled, analytical exit reviews a structural necessity.