Harvey CEO says the company's AI usage jumped from 1 trillion tokens a month to 12 trillion

The surge in Harvey’s AI token usage from one trillion to twelve trillion per month isn’t just a business milestone—it signals a deeper shift in how corporations value artificial intelligence. Token consumption, a proxy for computational workload, reflects not mere adoption but a strategic reallocation of resources toward AI systems that demonstrably improve efficiency, reduce costs, or unlock new capabilities. For Harvey, a legal AI startup, this explosion likely stems from clients prioritizing high-value applications—contract analysis, regulatory research, or due diligence—where the cost of processing vast datasets is outweighed by the potential savings in time and risk mitigation. The fact that companies are tripling down on such heavy usage suggests a market consensus: AI isn’t an experimental tool but a core operational expense. What’s less obvious is the undercurrent driving this demand. Legal and professional services have long lagged in digital transformation compared to sectors like finance or logistics, where automation is already table stakes. The pandemic accelerated adoption, but Harvey’s figures reveal something more fundamental: businesses now see AI as a hedge against inflation, labor shortages, and regulatory complexity. The jump from one to twelve trillion tokens implies that firms aren’t just testing AI—they’re embedding it into workflows at scale, treating it as indispensable rather than optional. Yet this raises critical questions. Will the returns justify the investment as usage balloons? Legal AI, for all its promise, still grapples with accuracy and bias issues, particularly in niche or highly regulated domains. Harvey’s growth also exacerbates the broader AI infrastructure divide: only well-funded enterprises can afford such token volumes, potentially entrenching a two-tier system where AI’s benefits accrue to the already privileged. Meanwhile, regulators are playing catch-up, with little clarity on how to audit or tax AI-driven productivity gains. Long term, Harvey’s trajectory mirrors a broader trend: the commodification of AI. As costs plummet and capabilities rise, the real competition won’t be between startups but between industries racing to integrate AI into their DNA. The question is whether the legal sector—and others following its lead—can sustain this pace without losing sight of the human judgment that still underpins their work.