Gartner Predictions: 30% Failure Rate Expected for Generative AI Projects by 2025

Recent research by analyst firm Gartner predicts that by the end of 2025, around 30% of companies experimenting with generative AI will abandon their projects after the proof of concept stage. These findings were shared at Gartner’s Data & Analytics Summit in Sydney, Australia. The study revealed that early adopters of generative AI are facing challenges related to rising costs, with deployments ranging from $5 million to $20 million.

For instance, customizing a generative AI model, such as fine-tuning a Llama model with industry-specific data, could cost a company $5 million to $6 million upfront and up to $11,000 in recurring expenses. Building a model from scratch could cost as much as $20 million. Even implementing a basic document search feature using retrieval augmented generation (RAG) can cost upwards of $750,000.

According to Gartner, businesses are finding it difficult to justify these substantial investments in generative AI. The firm noted that generative AI demands a higher tolerance for future financial investments rather than immediate returns on investment (ROI). Rita Sallam, Gartner's distinguished vice president analyst, highlighted the impatience of executives to see returns on generative AI investments post the hype of last year. However, organizations are struggling to prove and realize the value of these investments.

While early adopters of generative AI are experiencing mixed results, a recent Gartner survey of 822 business leaders revealed that only a small percentage reported revenue growth, cost savings, and productivity enhancements following generative AI deployments. Sallam emphasized the importance of analyzing the business value and total costs associated with generative AI technologies to ascertain both direct ROI and future value impact, enabling informed investment decisions.

Gartner recommends that businesses evaluating generative AI technologies should carefully assess the business value and costs to uncover potential returns and make strategic decisions about the scalability and expansion of generative AI initiatives. Ultimately, leveraging generative AI effectively can lead to significant business innovation and positive outcomes across various departments.