SaaStr's Jason Lemkin drops 3 urgent AI governance lessons for revenue teams

The Gist

• ClickHouse's AI spend surged 60x since February 2026, creating a new cost center overnight
• Unmanaged token usage leads to $600 spent to save $500, with no accountability until quarter-end
• AI vendors tout cheap base tokens but hide expensive reasoning token costs in volume
• Talent acquisition is legal but IP theft isn't: leave laptops behind when hiring domain experts

• The Apple lawsuit against OpenAI highlights the importance of hiring expertise legally without engaging in theft, which can lead to lawsuits.
• AI spend can escalate rapidly without governance, potentially becoming a significant cost center.
• Cost per completed task is a more valuable metric than cost per token when evaluating AI models.
• Net new logo growth is a more critical metric than retention for predicting the success of public B2B companies.
• Losing the bottom of the funnel can have long-term detrimental effects on a company's growth.
• Switching costs are collapsing rapidly due to AI, making renewals less secure.

• Implement a spend governor on AI to prevent unchecked cost escalation.
• Focus on net new logo growth as a primary metric for company health.
• Evaluate AI models based on cost per completed task rather than cost per token.
• Defend the bottom of your funnel to ensure long-term growth.

• 60x (Increase in ClickHouse's AI spend since February.)
• 15% (Annual net new logo growth rate that predicts success for public B2B companies.)
• $250 billion (Total developer wages in the US.)
• 10% (Gross-margin spend on AI that companies will tolerate.)