Startups brag about spending more on AI than human employees

A growing trend among tech startups, dubbed "tokenmaxxing," involves CEOs boasting about spending more on AI compute than on human salaries. Amos Bar-Joseph, CEO of Swan AI, posted on LinkedIn that his four-person team spent $113,000 on AI tokens in a single month, calling it a point of pride. He stated the goal is $10M ARR with fewer than ten employees, using AI for roles traditionally filled by humans. Meta has an internal dashboard called "Claudenomics" tracking employee AI token usage as a productivity metric. Salesforce introduced "Agentic Work Units" to measure AI spend effectiveness. Chen Avnery of Fundable AI commented that the $113k is not a cost but a reallocated headcount budget. Medvi, a GLP-1 telehealth startup with two employees and seven contractors built largely via AI, is reportedly on track for $1.8 billion in revenue this year. Andrew Pignanelli of General Intelligence Company said his AI compute spend now sometimes exceeds human salaries. Critics question the sustainability and actual value of this spending, noting that AI companies like OpenAI and Anthropic are losing money, and AI failures often require human cleanup.