A Germany-based Artificial Intelligence (AI) expert and founder of Scaling Intelligence, Opetunde Adepoju, has raised alarm over the growing rate of failure in enterprise AI projects globally, warning that over $40 billion invested in such initiatives has yielded little or no returns.

Adepoju, a former AI Programme Manager at SAP, stated this in a report titled: “The $40 Billion Black Hole: Why Enterprise AI Keeps Failing,” where she blamed poor organisational readiness, weak strategic alignment, and rushed implementation for the massive losses.

She noted that about 95 per cent of enterprise AI projects fail to generate any measurable value, despite the increasing pressure on organisations to adopt AI.

“The problem is not the technology. It is how organisations approach AI. They see it as a quick fix or IT upgrade, rather than what it truly is, a strategic transformation,” she said.

Adepoju, who has a background in data science and business strategy, cited the example of a Fortune 500 company that invested $8 million in a chatbot project meant to improve customer service. However, six months later, the project was abandoned after it caused a 23 per cent drop in customer satisfaction.

“The tech worked, but employees didn’t use it. That’s not a technical failure; it’s an adoption failure. And it’s happening across industries,” she lamented.

According to her, the global rush to embrace AI is similar to earlier tech bubbles like the dot-com crash, the electricity panic of 1882, and the railroad collapse of 1873. In each case, early hype led to massive investments, followed by widespread failure before value was finally realised.

Adepoju listed the three major types of AI failure as technical, adoption, and value failures.

She stressed that while technical issues are easier to fix, adoption and value failures are more damaging and harder to detect.

“It’s not enough for AI to function. It must be used, and it must make an impact on the business,” she warned.

In the report, the expert also explained the five levels of AI maturity organisations must pass through to succeed, including manual, digitalised, analytical, intelligent, and AI-Native.

Adepoju warned that no company can skip stages and expect success, adding that using AI without proper data, processes and user engagement is a recipe for failure.

To avoid future losses, Adepoju urged company executives to reframe their questions from “How do we deploy AI?” to “What must be true for AI to work here?”, start with one high-impact use case, focus on usability and staff trust, and treat AI deployment as a learning journey.

She added that her next publication, titled User-First AI Adoption Framework, will provide practical steps for organisations to avoid common AI pitfalls and turn their investments into real value.