B2B software faces AI cost controls and faster churn risks

B2B software companies are being pushed to manage AI as a major operating discipline rather than an experimental feature. Hiring domain experts remains legitimate, but taking files or proprietary materials creates legal exposure, with Anthropic cited as proof that talent can build a company worth more than $50 billion without carrying employer data out the door.

Token usage is now a meaningful cost center. ClickHouse’s AI spend is up 60x since February, underscoring the need for spend governors and a shift away from judging vendors by price per token. The better benchmark is cost per completed task, because reasoning tokens and model behavior can make cheap base pricing misleading.

Growth risk is also shifting. Net new logo growth above 15% a year is framed as the key survival signal, while retention can mask a weakening funnel. AI tools may take low-end users before incumbents notice, and faster migrations are eroding traditional switching costs. Salesforce reportedly used an LLM-powered lift to leave Marketo in one day, showing how sticky revenue can become less durable.

AI costs are expected to become a permanent margin item, with token spending modeled around 10% of revenue for software companies. Debt is treated as especially dangerous for slower-growth companies after Constellation bought TouchBistro, a $70 million ARR company, for $70 million, wiping out common equity through senior debt dynamics.