AI Failed to Deliver Expected Business Returns: 95% of Companies Saw No Results

It has been revealed that generative AI technologies, which have been rapidly gaining popularity worldwide in recent years, have failed to deliver the expected financial results for many enterprises. According to Ixbt.com, citing a report by the MIT NANDA organization, 95% of the studied companies failed to achieve significant and measurable returns from implementing AI. Nevertheless, only 5% of the analyzed projects succeeded and managed to generate millions of dollars in profit. This was reported by Ixbt.com reports .

According to the data, this research was conducted against the background of $30–40 billion in investments directed toward GenAI technologies by corporate structures on a global scale. Experts studied over 300 publicly announced AI initiatives and conducted in-depth interviews with representatives of 50 organizations. Furthermore, the opinions of 153 senior executives who participated in four major industry conferences were compiled.

The Problem Is Not the Technology Itself, but Its Implementation

The authors of the study attribute this sharp gap between expectations and real results not to poor quality of AI models or strict legislative restrictions, but directly to difficulties in practical application. Among the main problems noted are unstable workflows, limited capabilities of systems to retain context, and account for user feedback. In addition, the fact that the developed solutions were not sufficiently linked to employees' daily tasks also had an impact.

According to the report's results, over 80% of the inspected organizations explored or tested universal tools such as ChatGPT and Copilot. However, only 5% of specialized corporate systems designed for specific narrow tasks reached the full industrial-scale exploitation stage. This shows that universal solutions cannot always be easily adapted to the specific needs of a business.

Difference Between External Partners and Internal Developments

The study also paid special attention to differences in approaches to creating software solutions. Within the studied group, projects created with the participation of external partners reached the full implementation stage in approximately 67% of cases, while developments made based on companies' own internal resources showed results around 33%.

At the same time, experts emphasize that success criteria are defined differently in each company. It is reminded that involving external contractors cannot automatically guarantee high results, and the identified correlation reflects only the cases within the selected research sample.