The global economy is currently in the grip of a $1.3 trillion contradiction. Since the dawn of the 2020s, organizations have poured astronomical sums into digital transformation, yet the failure rate remains a haunting 70% to 80%. We are living through what I call the "Transformation Trap"—a period where the rapid irruption of technology is mistaken for the deep reorganization of the institutions that use it.
As a historian of large technical systems, I see a pattern today that is eerily familiar. When we ask engineers what the AI-embedded society of 2050 will look like, they describe faster algorithms and more GPUs. But history tells us that technology is never the bottleneck. The bottleneck is us: our hierarchies, our incentives, and our refusal to let go of the "central drive shafts" of a previous era.
To understand why your AI initiatives might be failing, we have to travel back to the late 19th century and visit the ghost of the steam engine.
The Victorian Drive Shaft: A Lesson in Inertia
In 1990, the economic historian Paul David published a seminal paper, "The Dynamo and the Computer" (David, 1990). He wanted to solve a mystery: why did the introduction of the electric motor in the 1880s fail to improve industrial productivity for nearly forty years? (David, 1989).
Factories in the steam age were masterpieces of mechanical complexity. They were built around a single, massive central drive shaft that ran the length of the building. Power was distributed to individual machines through a dangerous and inefficient web of leather belts and pulleys (Author, 2026). When factory owners first bought electric motors, they did exactly what many executives are doing with AI today: they "bolted them on." They simply replaced the steam engine at the end of the existing central shaft with a large electric motor (Author, 2026).
The power source had changed, but the organizational logic remained Victorian. The machines were still tethered to the shaft. The factory floor was still cramped and inflexible. Productivity stayed flat because the "installation" of the technology was not accompanied by the "deployment" of a new organizational model (David, 1990; Perez, 2002).
Productivity only soared in the 1920s when a new generation of managers moved to a "unit drive" system. They realized that because electricity could be distributed through wires, every machine could have its own small, dedicated motor (David, 1990). This allowed them to tear out the central shaft and reorganize the entire factory based on the logical flow of materials rather than the physical constraints of a steam pipe (David, 1989; Author, 2026). This was a "quantum jump" in organizational principles—and it is exactly what the AI era demands (Stratrix, 2025).
| Technological Era | Legacy "Central Shaft" | "Unit Drive" Reorganization |
| Steam/Railways | Small owner-led firms | Professional managerial hierarchies (Chandler, 1977) |
| Electricity | Motors on steam shafts | Flexible assembly lines (Fordism) (David, 1990) |
| Computer Age | Automating paper forms | Distributed knowledge work (David, 1990) |
| AI Era | Chatbots on legacy CRM | Agentic AI & Autonomous flows (Schram, 2026) |
The AI "Bolting-On" Phase
Most organizations today are in the "bolting-on" phase of AI. They are adding Large Language Models (LLMs) to legacy customer service departments or using generative AI to draft emails within a 1990s-style hierarchy. They are essentially putting a high-performance electric motor at the end of a rusty steam-era drive shaft.
Why is this a trap? Because "bolting-on" creates a "dosage curve" where more technology often leads to worse outcomes (Brown, 2026). We see this clearly in education, where $165 billion has been spent on EdTech, yet test scores have collapsed alongside device saturation (Brown, 2026). Schools gave every student a tablet (the bolt-on) without changing the instruction (the logic). This created the "Distraction Externality"—a state where the cognitive cost of students resisting non-educational apps exceeds the learning benefit of the software (Brown, 2026).
True transformation is not about doing the old things faster; it is about doing new things that were previously impossible.
The Three-I Framework: Infrastructure, Institutions, Incentives
To escape the trap, leaders must look beyond the "Infrastructure" layer and address the "Institutions" and "Incentives" (Author, 2026).
Infrastructure (The Material): This is the easiest part. You buy the GPUs, you license the LLMs. But without the next two layers, this is just an expense.
Institutions (The Rules): As Nobel laureate Douglass North argued, institutions are the "rules of the game" (North, 1991). If your organization’s rules reward information hoarding and manual oversight, AI will fail. You cannot run an AI-driven company with a Chandlerian hierarchy designed to manage 19th-century railway telegraphs (Chandler, 1977; North, 1990).
Incentives (The Humans): Every technological revolution redistributes power. Middle management has historically been the "perennial bottleneck" because they have the most to lose from transparency and automation (Author, 2026). If a manager’s status depends on "coordinating" information that an AI agent can now synchronize in milliseconds, that manager will subconsciously (or consciously) sabotage the transformation (Author, 2026; BPPE Consulting, 2025).
Case Study: Siemens and the "New Fabric"
Siemens provides the gold standard for moving from "bolting-on" to "deep reorganization." Under CEO Roland Busch, the company is executing the "ONE Tech Company" program (Busch, 2025). They aren't just "using" AI; they are building a "ONE Data Fabric" that unifies information across every business unit, from rail to healthcare (Busch, 2025; Siemens, 2025).
Siemens is also addressing the "Problem Decomposition" gap—the reality that most managers don't know how to break down complex goals into chunks that autonomous AI agents can execute (CPO Strategy, 2025). By creating a "ONE Software Engineering System" that enables company-wide code-sharing, they are essentially installing the "unit drives" of the AI era (Busch, 2025). They are willing to deconsolidate legacy units like Siemens Healthineers to focus on the high-velocity "Industrial AI" of the future (Siemens, 2025).
The Path Forward: Managing the Structural Crisis
We are currently in the "Frenzy" phase of the AI revolution, defined by financial speculation and irrational exuberance (Perez, 2002; Author, 2026). But a "Turning Point" is coming. Gartner predicts that 30% of generative AI projects will be abandoned by 2025 as the reality of institutional inertia sets in (Raghavan, 2025).
For decision-makers, the survival strategy involves three imperatives:
Move from Automation to Reorganization: Stop asking "How can AI do this task?" and start asking "If this information was free and instantaneous, how would we build this department from scratch?"
Invest in Human Judgment: AI is excellent at structured tasks but achieves only ~68% accuracy in emotional responsiveness compared to 92% for humans (BPPE Consulting, 2025). Your "unit drive" managers must evolve from being "monitors" of data to "mentors" of judgment (Author, 2026; Bardeen, 2025).
Acknowledge the 30-Year Horizon: History suggests the full impact of AI won't be visible until the 2040s or 2050s (Author, 2026; David, 1990). The winners are not those who "win" the 2024 hype cycle, but those who are building the institutional foundations for the mid-century.
The Victorian factory owners who clung to the central drive shaft eventually went bankrupt, out-competed by those who embraced the flexibility of the unit drive. The "Transformation Trap" is real, but it is avoidable. Stop bolting-on the future to the past. Create a strategy based on the possiblities of the new technology, and adapt your structure accordingly. Tear out the central shaft. Rebuild the fabric.
References
Author. (2026). The Transformation Trap: Institutional Inertia and the Great Productivity Paradox of the AI Era..
Bardeen, L. (2025, January 20). Cutting Through the AI Noise. Stanford SIEPR.
https://siepr.stanford.edu/news/cutting-through-ai-noise BPPE Consulting. (2025). AI-ready University 2.0: AI Tutoring: What it can replace, what it absolutely can't.
https://bppe.consulting/blog/ai-ready-university-20-ai-tutoring-what-it-can-replace-what-it-absolutely-cant Brown, N. B. (2026, February 8). What The Economist Got Right (and Terribly Wrong) About Education Technology. skepticism.ai.
https://skepticism.ai/p/the-165-billion-question-what-the Busch, R. (2025, November 13). ONE Tech: The Next Stage of Growth [CEO Presentation]. Siemens AG.
https://assets.new.siemens.com/siemens/assets/api/uuid:93238d21-b0b0-4fd6-a97c-7269527a445f/251113-ceo-presentation.pdf Chandler, A. D. (1977). The Visible Hand: The Managerial Revolution in American Business. Harvard University Press.
https://www.hup.harvard.edu/books/9780674940529 CPO Strategy. (2025). Siemens: ONE Tech strategy vs bolting-on AI 2025.
https://cpostrategy.media/blog/topic/ai-in-procurement/ David, P. A. (1990). The Dynamo and the Computer: An Historical Perspective on the Modern Productivity Paradox. The American Economic Review, 80(2), 355-361. [suspicious link removed].
David, P. A. (1989). Computer And Dynamo: The Modern Productivity Paradox In A Not-Too Distant Mirror. University of Warwick.
https://ideas.repec.org/p/wrk/warwec/339.html North, D. C. (1990). Institutions, Institutional Change and Economic Performance. Cambridge University Press.(
https://www.cambridge.org/core/books/institutions-institutional-change-and-economic-performance/AAE1E27DF8996E24C5DD07EB79BBA7EE North, D. C. (1991). Institutions. Journal of Economic Perspectives, 5(1), 97-112. [suspicious link removed].
Perez, C. (2002). Technological Revolutions and Financial Capital: The Dynamics of Bubbles and Golden Ages. Edward Elgar.(https://books.google.com/books/about/Technological_Revolutions_and_Financial.html?id=FNW5RriDOGAC).
Raghavan, S. (2025, October 21). Creating AI That Matters. MIT News.
https://news.mit.edu/2025/creating-ai-that-matters-1021 Siemens. (2025, November 13). Earnings Release Q4 FY 2025.(
https://assets.new.siemens.com/siemens/assets/api/uuid:3948cdd4-35e0-4c1d-8412-9aed4097b3d0/HQCOPR202511117277EN.pdf Stratrix Strategy Lexicon. (2025). Productivity Paradox Quick Definition.
https://www.stratrix.com/strategy-lexicon/productivity-paradox
No comments:
Post a Comment
Note: Only a member of this blog may post a comment.