A simple AI tool from Anthropic triggered a $285 billion market loss by revealing how easily AI can replicate expensive legal and financial software, threatening the traditional per-user SaaS licensing model. While Nvidia’s Jensen Huang argues AI won’t kill software, the real disruption is in how software is priced and sold, forcing companies and workers to rethink their value in an AI-driven world.
A recent event in the tech industry saw $285 billion in market value wiped out in just 48 hours, triggered not by a revolutionary product, but by a simple 200-line markdown prompt released by Anthropic for its Claude Co-work AI tool. This prompt, which automates legal contract review, exposed how easily AI can replicate tasks that previously required expensive software and human labor. The open-source nature of the plugin made it clear that much of the value in established legal and financial software platforms could be replicated with basic AI workflows, leading to a massive sell-off in companies like Thomson Reuters, RELX (LexisNexis), and LegalZoom, as well as private equity firms invested in these sectors.
However, the markdown file itself wasn’t the true cause of the crash. Instead, it revealed a deeper, ongoing structural problem: the per-seat SaaS licensing model that has underpinned the enterprise software industry for decades is fundamentally threatened by AI. As AI agents become capable of performing tasks without human intervention, the justification for charging per-user license fees collapses. The market had been slow to recognize this shift, but the Anthropic plugin made the threat visible and undeniable, accelerating a repricing that had already begun due to declining revenue forecasts and compressed valuations across the software sector.
Jensen Huang, CEO of Nvidia, argued that AI will not kill software because AI runs on software infrastructure, and more AI means more demand for databases, APIs, and middleware. While this is true, the real issue is not the need for software itself, but the way software is priced and sold. The analogy to print media is apt: the internet didn’t destroy the value of news content, but it did destroy the business model of selling bundled newspapers. Similarly, the proprietary data and accountability provided by enterprise software companies remain valuable, but the per-seat access model is rapidly becoming obsolete.
The impact of AI is not limited to stock prices; it is already changing real-world business negotiations. For example, KPMG successfully pressured its auditor, Grant Thornton, to lower audit fees by citing AI-driven cost savings, even though no actual automation was implemented. This demonstrates how the mere existence of AI capabilities is being used as leverage to renegotiate service fees across industries, from legal to consulting to design. The shift is less about immediate automation and more about the changing expectations and bargaining power in knowledge work, as clients demand pricing that reflects the new economics of AI.
Ultimately, the survival of SaaS companies depends on their ability to pivot from traditional, user-based licensing to agentic, value-based models that charge for data access and accountability rather than headcount. This transition is urgent, as the cost of building custom software with AI is plummeting, threatening the core value proposition of buying generic SaaS tools. The same lesson applies to individual knowledge workers: simply adding AI as a bolt-on to existing workflows is not enough. Both companies and individuals must fundamentally rethink how they operate in an AI-driven world, or risk being left behind as the pace of change accelerates.