Startups Report Spending More on AI Compute Than Human Salaries

Tokenmaxxing as a Business Strategy

Several AI startups are publicly sharing that they spend more money on AI compute than on human employee salaries. This practice, referred to as 'tokenmaxxing,' treats high AI spending as a vanity metric and marker of growth.

Specific Examples from Source

• Swan AI CEO Amos Bar-Joseph reported a $113,000 monthly AI bill for a 4-person team, stating: 'I've never been more proud of an invoice in my life.'
• Bar-Joseph explained his company spends on Claude usage bills instead of human salaries, aiming for $10M ARR with a sub-10 person organization.
• He described the AI spend as covering functions that would typically require human teams: 'That $113K bill? A part of it IS our go-to-market team. our engineering, support, legal.. you get the point.'
• Andrew Pignanelli of General Intelligence Company reported spending $4,000 on Claude Opus tokens in a single day, noting: 'We've started spending more on tokens than on salaries depending on the day.'
• Chen Avnery of Fundable AI commented that their AI processes loan documents that would normally require 15 people, claiming: 'The math works when your AI spend generates 10x the output of equivalent human cost.'

Industry Context and Metrics

The source mentions Meta's internal 'Claudenomics' dashboard that tracks AI token usage by employees, with the narrative suggesting higher token usage indicates greater productivity. Salesforce has responded with 'Agentic Work Units' to measure whether AI spending translates to actual work output.

Business Models and Scale

• Startups are using AI to justify never hiring human workers initially, unlike larger companies using AI to reduce existing headcount.
• Medvi, a GLP-1 telehealth startup with 2 employees and 7 contractors built largely using AI, is reportedly on track for $1.8 billion in revenue this year.
• The industry is pursuing the 'one-person, billion-dollar company' concept, with venture capital pushing founders toward 'autonomous' companies with few or no employees.

Unanswered Questions

The source notes that these entrepreneurs don't address whether the AI compute spending is actually worth it, whether the money would be better spent on human employees, what types of disasters could occur, or whether this approach is financially sustainable.