Businesses face AI scaling problems as report finds

A DDN report finds 65% of enterprises say their AI environments are too complex and 54% cancelled AI projects over two years because of infrastructure issues, while energy and sustainability concerns are rising.

DDN's State of AI Infrastructure Report finds many companies are having difficulty turning AI projects into long‑running production systems. The report says a majority view their AI environments as overly complex and that infrastructure failures have led to cancelled projects. It also highlights growing attention to energy and sustainability around AI operations. The study notes that most organisations expect cloud and hybrid approaches to be central to scaling AI. Key findings: - 65% of enterprises report their AI environments are too complex to manage, and 54% have cancelled AI projects in the past two years citing infrastructure issues. - 97% of respondents say cloud is essential to scale AI initiatives, with hybrid AI workloads expected to grow by about 162% over the next 12 months. - 72% rely on external expertise rather than only in‑house teams, and the report records widespread skills gaps (reported by 98% of respondents) and struggling teams (83%). - Energy and sustainability are prominent concerns: 93% are actively trying to reduce AI energy impact, and around 47% cite power and cooling as top inefficiencies; "tokens per watt" is mentioned as an emerging metric. Summary: The report attributes many AI failures to fragmented or legacy infrastructure and to integration problems rather than purely compute limitations, and it highlights a strong reliance on cloud and third‑party expertise. Many organisations also report skills gaps and mounting pressure to reduce AI energy use. Undetermined at this time.