GTM Horizons
AI’s Reality Check: Why 95% of Companies Are Missing the ROI
Originally published October 15, 2025
A Career’s Perspective on Tech Revolutions
Having witnessed the seismic shifts of the last two decades - from “e-Business” and the Internet to widespread adoption globally of mobile phones to the complete overhaul of enterprise architecture via Cloud computing, I am blown away by what we see with AI and Generative AI. It is incredible. The pace of change is accelerating at a rate that is almost hard to believe. And yet, the parallels are undeniable: transformation succeeds not by dragging along whatever technology “sticks,” but by powering a clearly defined business strategy. Without that, the hype cycle inevitably gives way to the dreaded “trough of disillusionment.” The bright spot today is the speed at which agile leaders are deploying, pivoting, and capturing value. For everyone else, 2025 is serving up a harsh reality check.
The Widening Chasm: Elite vs. The Rest
A small, elite group of AI-mature organizations – roughly 5% of the market are achieving astounding results. These firms are projected to see five times the revenue increases and three times the cost reductions from AI compared to their peers. Their secret? They were early movers who not only invested in AI but strategically reinvested those early returns into fundamental organizational capabilities.
This elite group is heavily leaning into Agentic AI, autonomous systems that handle multi-step tasks end-to-end. Agentic AI is on track to account for 29% of total AI value by 2028, and the leaders are already funneling substantial portions of their AI budgets into these transformative agents.
Meanwhile, the remaining 95% are pouring massive capital into AI (U.S. tech giants alone will spend over $350 billion on data center expansions in 2025), yet they're struggling to connect these investments to the bottom line. This confirms a crucial truth: spending on AI does not equate to value from AI.
The "Trough of Disillusionment" and Internal Friction
Generative AI has officially entered Gartner's "Trough of Disillusionment." The initial hype has given way to a more pragmatic understanding of its limitations, the complexities of governance, and the often-elusive hunt for tangible value. This phase is exposing a costly truth: as one expert notes, "AI is only as strong as the data feeding it," meaning AI projects often just amplify the chaos of decades-old data debt and fragile internal systems.
This operational friction isn't just external; it's tearing at internal structures. Nearly two-thirds of executives report significant internal division and tension arising from GenAI adoption, frequently pitting IT against business units. Disturbingly, 31% of employees admit to sabotaging AI strategies. This is a clear signal of misalignment, fear, or lack of engagement.
The core issue remains unchanged: simply acquiring AI tools is insufficient. The most significant factor driving tangible earnings impact (EBIT) from AI is the wholesale redesign of workflows to embed this technology. Without this fundamental operational rewiring which requires disciplined change management, AI remains an expensive, underutilized asset.
The Critical Role of Workflow Redesign
This chart visually demonstrates that workflow redesign is the organizational step with the biggest effect on a company's ability to see earnings impact (EBIT) from AI:
View the referenced research chart
Executive Priorities: Talent, Power, and Governance
Looking ahead, executive focus must sharpen on critical areas:
- Talent is the Ultimate Bottleneck: AI/ML engineer roles now take over 70 days to fill on average. The imperative is clear: companies must stop chasing the general AI label and focus on the precision of fit for their domain-specific use cases, fostering an AI-ready workforce from within.
- Infrastructure as Geopolitical Power: The AI arms race is no longer just about chips; it's about power grids and land. Nations and major corporations are locked in a global competition to build multi-gigawatt data centers, transforming compute infrastructure into a core industrial and geopolitical asset. Leaders must consider energy access and sustainability in their long-term AI strategies.
- Governance to Prevent Sabotage and Financial Ruin: Robust governance, transparent communication, and comprehensive change management are paramount. Furthermore, finance leaders must budget for the true cost of AI –the hidden expenses of data preparation, integration, and MLOps – to ensure that poor cost forecasting doesn't "kill the 'I' part of ROI."
The path to AI-driven enterprise value is clear, but it demands far more than just financial investment. It requires strategic courage, a commitment to fundamental operational redesign, and executive leadership that can navigate both technological innovation and profound organizational change.
Go Deeper: Key AI Reports for Executive Research
Value is driven by organizational change, not just technology adoption. Workflow redesign has the biggest effect on EBIT from AI use.
BCG: Are You Generating Value from AI?
A widening gap exists between "Future-Built" leaders and the rest of the market. Leaders expect 5x revenue increase and 3x cost reduction vs. peers.
Generative AI is moving into a phase of pragmatic disappointment due to governance and value challenges. GenAI is placed in the Trough of Disillusionment.
Poor change management is causing internal friction and resistance. 31% of employees admit to sabotaging AI strategies.
Technocratic: Why Your AI Initiatives are Failing
Failures stem from skipping fundamentals, not technology limits. AI failures are the byproduct of skipping fundamentals like Change Management and Customer Discovery.
Bloomberg / Analyst Data Investment in foundational infrastructure is becoming a critical financial and geopolitical constraint. U.S. tech giants projected to spend over $350B on data center expansions in 2025.
Additional Sources:
- Gartner: Hype Cycle for Artificial Intelligence, 2024
- EY - US: The CEO Imperative 2025
- Deloitte: State of AI in the Enterprise 2025
- Bloomberg: AI Data Center Spending to Surge Past $350 Billion in 2025
- Technocratic: Why Your AI Initiatives Are Failing
#AITransformation #ExecutiveLeadership #BusinessStrategy #GenerativeAI #DigitalTransformation #FutureofWork #ROI #2025AI
