Legal AI Harvey's margins went from +50% to -50% in 6 months on frontier costs (finance.yahoo.com)

Legal AI startup Harvey, valued at $15.6 billion, has recently experienced a significant shift in its financial landscape, with gross margins plummeting from +50% to -50% within six months due to rising costs associated with reliance on major AI providers like OpenAI. This spike in operating costs prompted Harvey to pivot towards open-weight AI models, which are generally more affordable and allow for custom solutions tailored to their specific needs. In August, Harvey launched its own model using Moonshot AI's Kimi K3, achieving performance comparable to leading systems at markedly lower costs, thereby restoring its profitability. This trend reflects a broader movement in the AI/ML community as startups across various sectors—such as healthtech and fintech—embrace open-weight models to cut exorbitant expenses and maintain greater control over their technology. While this shift poses risks to established AI giants' revenue and raises debates around the economic viability of custom model-building, it also emphasizes the importance of optimizing operational costs. However, challenges remain, including data privacy concerns and the need for specific expertise, which may hinder adoption for some firms. The ongoing evolution of AI usage continues to reshape the competitive landscape, as companies balance the benefits of independence against the efficiency of established models.