How Corporate Leaders Escaped "AI Purgatory":Three Phases of AI Maturity

The Paradox That's Blocking Your Budget!

Why is implementing artificial intelligence in large companies so frustrating?

Let's start with "WHY". And the "why" is that leaders today are drowning in contradictory data. On one hand we hear alarming reports from centers like MIT and BCG, which claim that 90% or even 95% of corporate Generative AI initiatives end in failure and bring no measurable return on investment (ROI) (MIT).

And then, almost immediately, we read the latest Wharton study (October 2025), which shows something completely different: 75% of business leaders in large companies (1000+ employees) see a positive ROI from AI (Wharton)

What is true? Is AI a revolution or a bottomless pit?

As an analyst who follows this data not just month by month, but day by day, I'll tell you: both reports are true.

They don't describe two different types of companies. They describe two different phases of the same journey. What we are seeing is a great stratification of the market. Most companies are stuck in the painful phase of experimentation ("AI Purgatory"). But a handful of leaders — the group described by Wharton — have cracked the code and moved on to the next stage.

A corporation's journey with AI resembles the history of electrification. It's not about buying a new tool. It's about building an entirely new power plant in a city that has run on steam engines for a hundred years.

Phase 1 (2023) – "The Wonder Phase" (Exploration)

Analogy: "The Discovery of the First Light-bulb"

Remember that winter of 2022/2023? The premiere of ChatGPT caused a shock, kicking off the "Wonder Phase" for corporations. Boards and executives around the world saw "magic" and were overcome with powerful FOMO (Fear of Missing Out).

The corporate reaction was immediate and top-down: "We must have this!". Magical thinking dominated. The focus was on possessing the technology. Investments were driven by the fear of being left behind and competitive pressure, not by strategy.

It was like admiring Edison's first light bulb at a presentation and announcing that the electrical revolution was complete, all while the entire city still ran on steam.

Phase 2 (2024) – "The Chaotic Installation Phase" (Experimentation)

Analogy: "Hooking Up Electricity to Steam Engines"

We enter 2024. Companies are massively buying licenses for AI tools — handing out Copilots, Geminis and hundreds of other applications. The era of chaotic experiments begins.

This is the heart of "AI Purgatory".

Managers, under pressure from boards, try to "shove" AI wherever they can. But they do so without changing the foundations. They force AI into old, broken, bureaucratic processes.

As in our analogy: they started installing electric lighting in factories that were still powered by steam engines. The result? It was brighter, but the machines didn't work any faster. The workers were only more frustrated.

It is in this phase that the statistics of 90-95% failure are born. Why?

• Automating Chaos: Implementing AI into a flawed process doesn't fix the process. It just creates faster, more expensive chaos.
• Lack of Goals: Projects were launched without clear business metrics and without owners who would be responsible for them.
• Burnout and Resistance: Employees were expected to implement AI "on the side" of their normal, already demanding jobs. This bred frustration, cynicism, and burnout. People didn't trust the outputs of the "black boxes" and the technology, instead of helping, created more work.

Most corporations are right here today — in a room full of expensive AI tools, surrounded by frustrated employees and processes that still operate like they did a hundred years ago.

Phase 3 (2025) – "The Grid Rebuild Phase" (Accountable Acceleration)

Analogy: "Replacing Steam Engines with Electric Motors"

And here we come to 2025 and the data from the Wharton report. Who are these 75% of leaders who see ROI?

These are the companies that were the first to understand that the technology is not the problem. The old factory model is the problem.

They understood that AI is not a tool you "add," but a new power source that requires rebuilding the foundations. They stopped plugging electricity into steam engines. They started tearing out the old production lines and installing new, efficient electric motors in their place (new processes based on AI).

The evidence for this is visible in the Wharton report data itself:

• Accounting (Measuring): A full 72% of these leaders "systematically measure ROI". This is a key change. They stopped measuring activity (the number of pilots) and started measuring real business impact (productivity, profit, efficiency). Instead of hundreds of chaotic experiments, they focused on a few key, high-value business initiatives.
• Scaling (Rebuilding): They understood that AI doesn't work without changing processes (so- called organizational rewiring). AI success is 70% dependent on changing processes and people, and only 10% on algorithms. Instead of just "illuminating" the steam factory, they began to demolish the old production lines and build new ones powered by AI.
• Investing in People and Rules: A full 89% claim that AI strengthens employee competencies. This doesn't happen on its own. It's proof that these companies invested in what 90% of others skipped: in people (training, upskilling) and in processes (guardrails). They created safety frameworks, giving employees the trust and psychological safety to use these new, powerful tools.

Conclusion: Are You Still Buying Light bulbs or Building a Power Plant?

Looking at the data over time the evolution becomes clear. The year 2025 has brutally separated the "AI tourists" from the "AI builders."

The tourists, which includes most companies, are still stuck in Phase 2 ("AI Purgatory"). They are excited by the light bulbs but frustrated that their expensive AI licenses aren't powering their old, steam-driven processes.

The builders — the leaders from the Wharton report — are in Phase 3. They understood that a true AI transformation is not an IT project. It is a strategic decision to rebuild the company's operational foundations.

The question every leader must ask themselves today is not: "Did we buy AI?". The real question is: "Do we have the courage to stop the old factory so we can fundamentally rebuild it around a new, more powerful energy source?" The answer to that question separates those still stuck in "AI Purgatory" from those who have just started to build the future.