Enterprise Redefinition Observed

Enterprise Redefinition Observed (ERO) — 18 companies, plus 5 that failed

​

A multiple-case analysis measuring AI-era companies on one scale

​

The strongest finding in this paper is not what could be measured. It is what could not.

We rated 18 companies on five dimensions — 90 cells — from public information only. 27 cells (30%) came back "unobservable," and not at random: 11 in Organization, 7 in Leadership.

This page explains VURA Working Paper No.3; the full paper (English original, 38 pages; Japanese edition, 33 pages) is linked at the end.

This paper in three minutes

• 18 companies, one scale. NVIDIA to Costco to TSMC: 5 dimensions x 18 = 90 cells, rated from public information alone.

• 30% could not be rated, and were not filled in. Guessing would have made a tidy table. The paper wrote "unobservable" instead, and that honesty became its biggest finding.

• The blind spots have a pattern. Business and Capital are visible from outside; Organization and Leadership are not. What the market can price and what actually stalls redefinition are not the same dimensions.

• None of the 18 replaced its core Purpose. They rebuilt businesses and balance sheets, never the core. Once what is protected is written down, what can be dropped becomes clear.

• Value originated before "Creation" — in 17 of 18, at Redefinition or upstream. A plan that starts from financial targets runs the chain backwards.

• The five failures were still learning. Kodak built the first digital camera. What was severed was the connection from learning to capital allocation.

 

The core of redefinition is invisible from outside

The study did not guess at cells it could not settle from public information; it recorded them as "unobservable." That discipline became its most original contribution.

 

What gets priced and what stalls capability are different dimensions

Business redefinition is visible: products launch, segments change, the revenue mix shifts. So is Capital: capex and acquisitions appear in the allocation table. But how did the decision structure change? Who was given which decisions? How was the executive team's mental frame updated? Organization and Leadership are almost unreadable from outside.

Proposition P18 | Observational asymmetry The public observability of redefinition is systematically higher in Business and Capital than in Organization and Leadership. As a result, assessments of redefinition capability based on public information (including market prices) overweight the former and underweight the latter. Disclosure that narrows the asymmetry moves enterprise value without any change in underlying capability. — Falsification condition: rejected if it is shown that there is no systematic difference in observability across dimensions, or that expanded disclosure (with capability held constant) does not move enterprise value.

In plain terms — the more visible a dimension, the more readily it is priced. So a company can move its share price simply by becoming visible, without changing any capability.

Observation asymmetry means visibility is skewed across dimensions, and skewed the same way for every company — hence "systematic."

 

An analogy: you cannot see the kitchen from the dining room

An analogy, because this is the hardest idea on the page. Imagine judging a restaurant from your table. You see the dish, the price, and when the menu changed. That is Business and Capital. You cannot see the kitchen: who runs which station, when the chef changed the prep routine, who may try a new ingredient. That is Organization and Leadership.

And whether this restaurant can rewrite its menu next year is decided in the kitchen, not the dining room. A rating from the dining room misses what matters most. That is observation asymmetry.

(End of analogy. The real subject is the public record of 18 companies.)

 

The asymmetry deepens with time

In the contrast set, Organization and Leadership maturity is "—" (unobservable) for all five failed or exited companies. As historical distance grows, what can be seen narrows to Business and Capital. Time makes the invisible more invisible.

"It cannot be measured because it cannot be seen" is what this paper demonstrates. That absence became the starting point for disclosure design in WP5 (Brain Capital Management) and WP7 (Redefinition Capitalism).

 

What was measured, and how

The 18: NVIDIA, Microsoft, Amazon, Apple, OpenAI, Tesla, Google, Meta, Netflix, Sony, IBM, TSMC, ASML, Arm, Costco, BYD, SpaceX and Anthropic.

Each is measured on the five dimensions of Enterprise Redefinition: Purpose, Business, Organization, Capital and Leadership. 18 x 5 = 90 cells. For each, from public information only, the paper records:

• which dimensions moved

• the maturity level (ERMM), 1 to 5 — whether redefinition there was a one-off event or a mechanism that runs repeatedly

• where value originated on the Future Value Chain, the causal order Purpose → Learning → Redefinition → Creation → Enterprise Value

• which redefinition patterns were used

 

The sample was deliberately mixed

Founded over a century ago (IBM) and under 15 years ago (OpenAI, Anthropic). Listed, newly listed in 2026 (SpaceX), and private. The US, Europe (ASML) and East Asia (TSMC, Sony, BYD). Platforms, but also a warehouse retailer (Costco), a lithography-equipment maker (ASML) and a contract manufacturer (TSMC). One reason: if "redefinition capacity" were just US big-tech economics relabeled, Costco and TSMC should have been unmeasurable. They were not.

 

The five failures are a contrast set, not a control group

Kodak, Nokia (devices), BlackBerry (devices), Yahoo and Sears are measured on the same scale, and the difference read. A methodological caution: the paper does not call them a "control group" but a "contrast set." A control group is matched to the successful companies on size, industry and period, so a difference cannot be blamed on those conditions. This study did no such matching, and says so. The five are material for contrast, not a device for proving causation.

Their ratings also follow a time anchor: each case is rated only on documented actions within its "decisive time window," not its whole history — fixing which version of the company, when, is rated. Windows: Kodak 1990s–2012; Nokia devices 2007–2013; BlackBerry devices 2007–2016; Yahoo 2007–2017; Sears 2005–2018.

 

Zero companies replaced their core Purpose

Among 18 companies that rewrote their businesses, organizations and capital structures, not one replaced its core Purpose. That is not trivial: firms this bold might easily have swapped their reason for existing as casually as a product line.

The finding changed the framework. Two layers (replacement / redefinition) were not enough, so the paper recodes Purpose in three: replacement / re-articulation / re-expression. Six companies newly articulated or rewrote their core on a datable occasion (Microsoft 2015, Sony 2019, TSMC 1987, Costco 1983, among others); the other 12 changed only wording, scope or field of application.

Proposition P12 | Purpose stability In sustained enterprise redefinition, the core Purpose is retained or re-articulated rather than replaced. Enterprises that replace their core Purpose will show lower cross-dimensional coherence and lower subsequent future value creation than enterprises that confine themselves to re-articulation or re-expression.

In plain terms — a reason for existing is restated, not swapped. Companies that swapped theirs did worse.

Re-articulation is the hinge that authorizes subtraction. Once what must be protected is written down, what can be discarded is decidable.

 

Business and Capital move together

Counting the dimensions that moved: Business 16 of 18, Capital 15 of 18, and the two moved together in 14 companies.

Proposition P13 | Capital co-movement Redefinition of the business is accompanied by redefinition of capital. A declared business redefinition without a corresponding movement in the capital allocation table predicts the non-realization of the declared redefinition.

In plain terms — a "new strategy" that leaves the money where it was will not get executed. A declaration becomes a redefinition only when it reaches the capital allocation table.

The sample keeps its own counter-example (Meta); these are not convenient cases. And only two companies moved all five dimensions at once — not, the paper argues, a sign of capability.

Proposition P14 | Simultaneity threshold The relationship between the number of dimensions redefined simultaneously and future value creation is inverted-U-shaped. Simultaneous redefinition beyond a threshold indicates not superior redefinition capability but risk exposure — the dilution of learning capacity and capital-allocation capacity.

In plain terms — changing everything at once is not proof of strength. It looks more like overstretch.

 

Value originates before "Creation"

On the Future Value Chain, 17 of 18 companies originated value at Redefinition or further upstream, most often at the Learning → Redefinition boundary (7 companies). Zero originated value at Enterprise Value.

This gives a practical test. If a medium-term plan begins with financial targets, the chain is running backwards. The paper's example is IBM in the 2010s: 22 consecutive quarters of revenue decline. IBM reversed course and survived; Sears ran the same inversion to the end.

Patterns showed the same regularity. No company redefined itself through a single pattern; all 18 combined two or more. The most frequent, the Layer Shift (13 companies) — moving where value is generated to another layer of the industry structure — always appeared with another pattern.

 

Measure coherence, not frequency of change

The spread is the gap between a company's highest and lowest maturity level across the five dimensions. Three companies had a spread of 2 or more (Amazon, Meta, Costco), and the paper reads them differently. Amazon's and Meta's spread is incoherence. Costco's is a "settled deep redefinition." A dimension redefined deeply long ago, whose premises still hold, may sit at a low number and stay there.

Proposition P17 | Coherence over frequency Cross-dimensional coherence predicts sustained future value creation better than frequency or recency of change. Wide dispersion in maturity predicts fragility — except where a low-maturity dimension has been settled by a past deep redefinition whose premises still hold.

In plain terms — the company that changes most often is not the strong one. The strong one is the company whose five dimensions mesh. A low number is not always a weakness.

 

The five failures were still learning

The contrast set's most important finding: four of the five failed companies had learning.

Kodak built the world's first digital camera: in 1975 its engineer Steven Sasson assembled the first self-contained prototype. In 1976 Kodak held 90% of US film sales. Revenue peaked near $16 billion in 1996. On January 19, 2012, it filed for Chapter 11.

The company that saw the future first fell first. What was severed was not foresight or learning but the connection from learning to redefinition to capital — exactly the boundary where value origination clustered among the 18 successes. The same seam created value on one side and was cut on the other.

Proposition P20 | Failure signature Redefinition failure is predicted by the co-occurrence of two or more of four markers: (a) declaring redefinition while capital allocation remains moored to existing profit pools; (b) the absence or churn of an articulated core capable of authorizing subtraction; (c) an inverted Future Value Chain with financial targets placed first; and (d) redefinition begun only after an overt loss of position.

In plain terms — failure has a shape. "Announce but do not move the money." "Never settle what to protect." "Start the plan from the numbers." "Move only once the loss is public." Two at once is an alarm.

 

The two groups do not overlap above Level 3

The table shows how the 18 successes and 5 failures are distributed across maturity levels (rated cells only).

Level18 successes (63 rated cells)5 failures (12 rated cells)

Level 1.  0.  5

Level 2.  4.  5

Level 3.  3.  2

Level 4 46.  0

Level 5 10.  0

What the table says: the two groups barely overlap. The failures have nothing at Level 4 or above; the successes have nothing at Level 1. They meet only at Levels 2–3, and that overlap is the diagnostic core: top and bottom are obvious, judgment is needed only in the middle.

The same "2" carries opposite meanings. Costco's Business Level 2 records a settled deep redefinition whose premises still hold. Kodak's records a model never questioned while its premises expired. Same number, opposite content.

Nokia is more telling still: the company that executed a large-scale redefinition in the 1990s rated only 1–2 in its window 15 years later. Maturity is not a trait a company possesses but a state it maintains — not a qualification earned once and kept.

 

Do not average these numbers

A second methodological caution: the paper states that averaging or ranking these levels is a misuse. Maturity is a diagnosis, not a rank order; the numbers name kinds of state, not points. "The five average 3.2" or "this company ranks 4.1 overall" erases what was just shown: the same 2 can mean opposite things. These numbers are not for adding up or lining up.

 

Capability can be bought. Time cannot

Every high-maturity case had a structural mechanism that secured time. Five types:

1. A parallel revenue pillar (11 companies) — an existing business that feeds the redefinition

2. Design of ownership structure and legal form (4)

3. Synchronized counterparty commitments (4) — in 2012, before EUV was finished, ASML secured EUR 1.38 billion in R&D funding from Intel, TSMC and Samsung, plus EUR 3.85 billion for 23% of its equity: customers jointly buying time for a machine that did not yet exist

4. Prepayment (3) — Costco's membership fee delivers trust up front, as working capital

5. Advance contracts with investors and disclosure design (2) — Amazon's shareholder letters handed over, years ahead, the criteria that would later justify its allocations

In the failures these were absent or ran in reverse. Kodak's and BlackBerry's rich core businesses delayed the test instead of funding redefinition: type 1's stall risk, run to completion. Sears' ownership structure consumed time rather than defending it. Yahoo's investor relations delivered pressure, not time.

Time is secured only by designing who evaluates you, on what cycle, against what promise.

 

Strength and fragility come from the same structure

One more regularity held in all 18: each company's strongest feature and its most fragile one came from the same structural choice. NVIDIA's published roadmap synchronizes an industry; one missed delivery would break the synchrony. TSMC's centrality is also its single point of failure. Costco's thin margins align interests and leave no cushion. BYD's vertical integration is speed uphill and fixed cost downhill.

BlackBerry is the mirror image. A membership model that rewards constancy was Costco's strength; an installed base that rewards constancy is why BlackBerry waited until too late. Same trait, opposite sign. So one diagnostic question suffices:

What dependency does your strength stand on?

The answer writes the real risk register, the one no external rating shows.

 

What the paper admits it cannot yet show

The paper names five limits.

1. The case materials and the analysis have the same author, so no inter-rater reliability — whether another reader would reach the same ratings — can be claimed

2. Relying on public information only, 27 of 90 cells are unobservable

3. The 18 are prominent firms that succeeded at redefinition at least once: survivorship bias, seeing only survivors (mitigated, not removed, by the contrast set)

4. Observation is a single window as of August 2026

5. Maturity levels and pattern assignments are judgment codings, not psychometrically validated scales — judgments against criteria, not a statistically validated instrument

It also leaves a discrepancy in place: an earlier coding table listed 26 unobservable cells; the paper's recoding found 27. The one-cell difference is kept as a note, not reconciled.

As case research, whether its regularities generalize to a large sample cannot be settled here. That is why all nine propositions (P12–P20) carry falsification conditions. Writing down what would prove the theory wrong, alongside the theory, is this series' discipline.

 

Implications

For executives. Check the order: if your medium-term plan begins with financial targets, the chain is running backwards. Then sort what your company says it "cannot change." Most of it needs a board resolution, not a change in the law.

For investors. Three points. The ordering test can be run on public information alone. Time-securing mechanisms can be disclosed, yet almost never are. And disclosure completes redefinition: the paper's example is Amazon, whose separate segment reporting let its redefinition show up in enterprise value. The capability existed before; what changed is that it became visible. That is P18, read from the practitioner's side.