Why 80% of AI Initiatives Stall Before They Deliver

A top analyst from Yandex Uzbekistan dropped a sobering stat at the recent Digital Uzbekistan forum: roughly 80% of business AI projects fail. Alexander Merkushev, who oversees AI deployments at the company, says the root cause isn’t technical—it’s strategic. Companies are rushing to apply artificial intelligence everywhere they can, instead of focusing on where it actually solves a real problem.

Merkushev draws a direct parallel to the Big Data frenzy a decade ago. Back then, firms spent heavily on data lakes and analytics platforms, only to find that collecting data without a clear use case rarely paid off. The same pattern is repeating with AI. The hype cycle is peaking, and disappointment is already setting in as organizations realize that generative AI and large language models won’t magically fix broken processes.

Building an in-house enterprise AI system is a massive undertaking. Merkushev warns it demands hundreds of specialists, months or years of development, and millions in investment. Yet many companies jump in without a clear problem statement, then wonder why their ROI never materializes.

The path forward, he suggests, is to step back from the hype. Identify a concrete business need first, then explore whether AI is the right tool—not the other way around. That shift in mindset could turn the 80% failure rate into a much smaller number.