Brew Bytes called this failure rate a year before Gartner did
Today's lead cites Gartner's finding that most AI projects don't deliver. Brew Bytes guest Charles Gabriel called the same number over a year ago. On Episode 13, "95% of Enterprise AI Projects Fail," Gabriel cited MIT research showing 95% of enterprise AI experiments failing, and gave the same diagnosis two different research houses have now confirmed. In his words: "They don't always know the problem that they're solving. They see cool AI tech and they want to bring it in and they want to pitch it."
What it actually means
Gabriel scopes AI deals for a living, and his read wasn't about the technology. It was about enterprises reaching for AI before anyone had written down the problem it was meant to solve. A year and a different research house later, Gartner landed on the same failure pattern, just with a lower headline number.
When the same finding turns up twice from two unrelated sources a year apart, it has stopped being a blind spot and become a pattern nobody has fixed. Before your next AI project gets signed off, make someone write the problem it solves in one sentence, with no model or vendor name in it. If nobody can do that, you already know which side of the number you're on.
The same failure rate, from two unrelated sources a year apart, isn't a blind spot. It's a pattern nobody's fixed.
Source: Brew Bytes — Ep13, Charles Gabriel · Sunday 27 September 2026