Gartner doesn’t publish numbers like this casually. When their analysts put a percentage on how many enterprise initiatives are going to get canceled, it’s because they’ve watched the pattern show up across enough client conversations that it’s become predictable. Over the past two years, they’ve published two separate forecasts about AI project abandonment, and read together, they tell a more useful story than either one does alone.
The first came out in the summer of 2024. Gartner predicted that thirty percent of generative AI projects would be abandoned after the proof of concept stage by the end of 2025, citing poor data quality, inadequate risk controls, rising costs, and unclear business value as the leading causes. The second came a few months later, forecasting that through 2026, sixty percent of AI projects would be abandoned because organizations lacked the data infrastructure to support them properly. Different numbers, different framing, same underlying diagnosis. In both cases, the projects that survive aren’t necessarily the ones built on the best technology. They’re the ones somebody could still justify.
I think a lot of executives read forecasts like these and hear a warning about technical risk: bad data pipelines, immature tooling, integration debt. Those things are real, and they’re part of it. But sit through enough of these conversations yourself and a simpler pattern emerges. A project survives its second budget review when someone in the room can say, with a number attached, what it’s actually doing. A project gets quietly abandoned when the honest answer to that question is a shrug. Data quality problems and unclear business value aren’t really two separate failure modes. They’re the same failure wearing different labels, because nobody can produce a defensible business case out of data they don’t trust.
This is why we built Nymbral around the contract and the telemetry rather than around a survey. A project that’s actually delivering has a trail: spend that’s traceable, adoption that’s measurable, outcomes that map back to the workflows the tool was meant to change. A project that’s coasting on momentum has none of that, and the absence usually becomes visible right around the same time a CFO starts asking pointed questions at renewal. Gartner’s abandonment numbers are what happens next. Ours is the tool built to keep a good project off that list, and to get a bad one off it faster.