SaaStr predicts AI tokens will replace 25% of GTM headcount by 2027

The Gist

• Jason estimates AI tokens will consume 33% of engineering budgets by 2027, replacing marginal roles
• Uber already caps engineer token spend at $1,500/month, signaling the new cost benchmark
• EDA software's 13% tool spend ceiling will shatter as AI workflows dominate operations

• AI tokens will replace 25% of GTM headcount by 2027, as companies shift budgets from human roles to AI-driven efficiency.
• The ratio of dollars spent on engineers to dollars spent on tokens is the most critical metric in B2B right now, shaping the future of AI market pricing.
• Autonomous AI engineers (like Devin) are emerging as a transformative force, potentially replacing human roles in software development.
• The bar for startup funding has risen dramatically, with investors now seeking 'billion-dollar positions' rather than just 'billion-dollar outcomes.'
• Public software companies are bifurcating into two growth trajectories: traditional human-per-seat models (slow growth) and AI-driven agentic models (fast growth).
• The shift from capex-light to capex-heavy business models in AI companies raises questions about long-term stock performance and sustainability.

• Start tracking token spend as a percentage of fully loaded engineering costs to anticipate future headcount adjustments.
• Evaluate whether your product is consumed by AI agents (growth opportunity) or only by humans (potential decline).
• Raise capital aggressively when well-funded competitors are raising, as the cost to compete is increasing dramatically.
• Ensure your pitch demonstrates credible potential for a 'billion-dollar position' rather than just a 'billion-dollar outcome.'

• 25% (Predicted replacement of GTM headcount by AI tokens by 2027.)
• $65B (Anthropic's funding round and IPO filing valuation.)
• $26B (Cognition's valuation after raising $1B.)
• 60% (Gartner's projected increase in AI software spend this year.)
• 1.5x (Output increase required to justify replacing human roles with AI tokens.)
• $300K (Approximate fully loaded cost per employee in product/engineering orgs.)