Returns on AI investments may be harder to achieve than the hype suggests

NVIDIA has conducted a survey of various telecoms operators (including mobile, fixed and cable companies), as well as their hardware and software suppliers, for the past 3 years.¹ NVIDA is one of the most significant vendors in the world in regards to AI and generative AI (GenAI) and is keen to promote the ongoing growth of AI, but there appears to have been a minor pause in some aspects of AI roll-outs when compared to last year’s survey results. The same observations have also been made by a researcher at MIT, who noted that only 5% of AI projects have exhibited any return on investment, meaning that 95% are given no returns for business.² Nonetheless, the results of NVIDIA’s survey show plenty of positives. For example, the number of respondents using or implementing AI increased by 8 percentage points between 2023 and 2024, which means that 97% of respondents are now engaged with AI and 49% are actively using it. This aligns with the research from MIT, which noted that 40% of its respondents are deploying AI and 80% are investigating it.

The two core business justifications for the use of AI are cost savings and increased revenue. However, there is a realisation that returns on AI investments are not that straightforward and are harder to come by than the market hype suggests.

¹ NVIDIA, State of AI in Telecommunications.

² Aditya Challapally conducted research based on 150 interviews with technology leaders, a survey of 350 employees and an analysis of 300 public AI deployments and found a clear divide between success stories and stalled projects. MIT NANDA (2025), The GenAI Divide: State of AI in Business 2025.