Chapter 082　What Did Microsoft Redefine?

Microsoft is the only case among the enterprises this series treats that stalled once and redefined itself out of the stall. Forty years from founding, it became one of the largest companies in the world. Then, for roughly a decade, it visibly lost momentum. And then it came back to the center of the industry. The question worth asking is not why it succeeded. It is what was happening in the middle of the stall, and what made the turn possible. This chapter follows that turn along the five dimensions of Enterprise Redefinition.

1 What makes this enterprise worth a question

Most enterprises either keep growing on the strength of success or disappear in failure. Few come back from a stall. That is what makes Microsoft a singular object of observation for management theory. Confirm it with numbers, using the company’s own disclosures available as of August 2026. For the fiscal year ended June 2014, full-year revenue was $86.83 billion and net income was $22.07 billion (FY14 Q4 earnings announcement, July 2014). The same disclosure put the annualized revenue run rate of the cloud business at $4.4 billion. About 5 percent of the total. Set that against the fiscal year ended June 2026 (FY2026). Fullyear revenue was $331.8 billion, up 18 percent year over year. Operating income was $155.2 billion, up 21 percent (Microsoft, July 29, 2026). Azure’s annual revenue passed $100 billion for the first time, growing 41 percent for the year. In twelve years the company grew roughly 3.8 times. But this chapter is not about the multiple. The multiple is a result, not a cause. What we have to look at sits on the side of the cause. Microsoft during the stall was not incompetent. It had the technology, the people, the capital, and the customer base. What it did not have was the will to rewrite its own definition. That is where this case earns its question. The seven-stage process of Enterprise Redefinition begins with Recognize. Its central question is, “What assumptions about our enterprise are becoming obsolete?” A stall is the state of being unable to pose that question. It is not the state of answering it wrongly. And the reason the question cannot be posed is usually not a lack of capability. It is having too much to protect.

2 Conventional answers and their limits

Three explanations of Microsoft’s return circulate widely. Each is partly right. None translates into a managerial decision. The first answer: “It changed because the CEO changed” On February 4, 2014, Satya Nadella became chief executive officer. The change that followed was dramatic. Therefore, the account runs, the succession was everything. It is a fact that the succession was the starting point. But this explanation is not reproducible. “Bring in a good executive” is not an answer in management. And who counts as a good executive can only be known afterward. The explanation also drops part of the record. In his first-day company-wide email, Nadella wrote that Microsoft had to succeed in a “mobile-first, cloud-first world” (Microsoft, February 4, 2014). That recognition was shared by the previous leadership as well. The difference was not one of recognition. The second answer: “It bet on the cloud and won” The second is an explanation by business selection. It moved from packaged software to the cloud, and so it won. This too is correct. But the order is told backwards. Microsoft did not move because the cloud was profitable. In order to move, it first discarded something. On July 8, 2015, the company recorded an impairment of $7.6 billion in its phone hardware business. Alongside it, it announced a reduction of up to 7,800 jobs (Microsoft, July 8, 2015). Nadella said at the time that the company was moving from a strategy to grow a standalone phone business to a strategy to grow and create a vibrant Windows ecosystem. An impairment is the admission, in the financial statements, of an error in past capital allocation. That was the substance of the turn. Before the business selection came the recognition of the loss. The third answer: “It caught the AI boom” The third is that the company got onto the generative AI wave of 2023 and after, earlier than anyone. But the Microsoft–OpenAI partnership began on July 22, 2019. According to the announcement at the time, Microsoft invested $1 billion. OpenAI’s services would move to Azure. Microsoft would become the preferred partner in commercializing new technologies (Microsoft, July 22, 2019). Four years before the boom. So it did not “catch” anything. It had prepared the place to stand. It could prepare it because by that point the company’s definition of itself had already been rewritten. The defect common to all three conventional answers is the same. Each narrates a result of the turn as the cause of the turn. What we need is an explanation of the layer beneath the result.

3 What was redefined — an analysis across the five

dimensions We take the five dimensions of Enterprise Redefinition in order: Purpose, Business, Organization, Capital, and Leadership. We cannot assert anything about decisions made inside the company. Everything below is what can be read from public information.

3.1 Purpose — the criterion for what to protect changed

Microsoft’s founding objective is widely known. A computer on every desk and in every home. That objective was achieved. An achieved objective binds the enterprise from the moment it is achieved. In 2015 the company revised its mission: “to empower every person and every organization on the planet to achieve more” (Microsoft’s official site, the same wording as of August 2026). The revision was made public through an internal email reported by GeekWire in June 2015. What matters here is the direction of the rewriting. The old objective named a product. The new objective names no product. What it names is a result occurring on the customer’s side. The difference is enormous in practice. Under an objective that names a product, customer value that does not pass through your product is a betrayal of the objective. Under an objective that names a result, it becomes one option among several. As the Enterprise Redefinition paper states, the five dimensions do not change at the same frequency. A Core Purpose can stay stable while its expression and its means of realization evolve. In Microsoft’s case, the core — give people tools and amplify what they can do — remained. What changed was the condition that the tool had to be Windows.

3.2 Business — not what was sold, but what was charged for

Redefining a business is not the question of “what do you sell.” It is the question of “what value do you provide.” A packaged software business is complete at the moment of sale. The buyer purchases a license and uses it thereafter. The seller’s attention gathers on the release date of the next version. Whether the customer actually uses the product does not connect directly to revenue. A cloud business is different. If the customer stops using it, next month’s revenue falls. The provider is therefore forced to care continuously about the customer’s outcome. The structure of billing determines the structure of attention. In the FY2026 fourth-quarter disclosure, Microsoft Cloud revenue was $59.3 billion, up 27 percent year over year. Intelligent Cloud in the same quarter was $39.3 billion, up 32 percent. Productivity and Business Processes was $37.8 billion, up 14 percent. More Personal Computing, by contrast, was $12.9 billion, down 4 percent (Microsoft, July 29, 2026). And now a third turn is under way. AI agents. In the third quarter of FY2026, Nadella referred to the annualized run rate of AI-related revenue. It exceeded $37 billion, up 123 percent year over year (Microsoft, April 2026). The company describes this as “the age of agentic computing.” As of the fourth quarter of FY2026, paid seats of Microsoft 365 Copilot passed 30 million. Commercial remaining performance obligations (RPO) stood at $678 billion, up 84 percent from the prior year. That figure is close to a reservation of future revenue.

3.3 Organization — the criterion of evaluation changed

On July 11, 2013, the leadership of the time announced a companywide reorganization titled “One Microsoft.” It was a declaration that the company would become “a devices and services company” (Microsoft, July 11, 2013). The reorganization tried to merge the walls between divisions. But as far as public information allows us to read, the org chart changed while the criterion of decision-making did not. The assumption that Windows sits at the center remained. Redefining an organization is not redrawing boxes. It is changing who is evaluated on what. The turn in that sense appears in a series of judgments during the Nadella period. Providing Office for iPad. Treating Linux as a customer environment rather than an enemy. Acquiring GitHub, the developer platform, for $7.5 billion on June 4, 2018 (Microsoft, June 4, 2018). Each of these delivers value without passing through Windows. Under the old assumption, none of them is executable. And in 2026 the unit of the organization itself is moving again. What the company presented at Build 2026 was a framework for governing AI agents as workers inside the enterprise. A specification for controlling agent execution declaratively, and a management platform for agent operations, have been published (Microsoft for Developers, 2026). The organization is ceasing to be a set of people only.

3.4 Capital — reallocated toward the future rather than the past

Redefining capital is the most financially painful of the five dimensions. The $7.6 billion impairment in 2015 is the type case. It was not the occurrence of a loss but the recognition of one. The substantive failure had happened earlier. Delay the recognition and you protect the current period’s profit. But the capital stays tied to the past. Line up where capital went afterward and the direction is consistent. The investment in OpenAI. Datacenter construction. Inhouse AI silicon. According to the FY2025 shareholder letter, the company has more than 400 datacenters across 70 regions, and 80 percent of the Fortune 500 use Azure AI Foundry (Microsoft, 2025). There is a heavy price here. The company is shifting from a software enterprise to a capital-intensive one. According to reporting, capital expenditure in the fourth quarter of FY2026 reached $41 billion (InfotechLead, July 2026). Software margins and infrastructure depreciation are different in kind. We should not write this up as a story of success.

3.5 Leadership — from judge to designer

Redefining leadership is the move from individual judgments to the design of a future. What can be read from public information is this. The company’s leadership is designing a layered structure rather than the win or loss of a single product. An infrastructure layer, a model layer, a development platform layer, and a business application layer. At each, models from other companies are handled as well. Azure AI Foundry carries models from multiple providers other than Microsoft. On June 2, 2026, Microsoft AI announced seven MAI models of its own. The company explained that these were trained from scratch rather than distilled, and referred to efficiency improvements in its in-house silicon, Maia 200. Mustafa Suleyman, who leads the unit, calls the goal “Humanist Superintelligence.” And he stated that people — that is, you — must always remain in control (Microsoft AI, June 2, 2026). This overlaps with the thinking behind Human-on-the-Loop Management. Human beings do not approve individual outputs from inside the loop. From above the loop, they design the system as a whole. But the correspondence between a stated philosophy and its implementation cannot be verified from outside. Here too we avoid asserting.

4 Structure — maturity, the value chain, and the

equations

4.1 How to read it against the Enterprise Redefinition Maturity

Model The Enterprise Redefinition Maturity Model (ERMM) has five levels: Reactive Enterprise, Improvement Enterprise, Transformation Enterprise, Continuous Redefinition Enterprise, and Future Value Enterprise. Microsoft during the stall matches the description of an Improvement Enterprise closely. Operations were excellent and products kept improving. The existing business model was rarely questioned. In the source paper’s phrasing, the state is one of becoming increasingly efficient while remaining fundamentally unchanged. From 2014 to about 2019 the company reads as a Transformation Enterprise. Enterprise-wide initiatives emerged, and Purpose evolved. The limit of that stage is continuing to view redesign as a project rather than a permanent organizational capability. As of August 2026, behavior corresponding to a Continuous Redefinition Enterprise is observable — the level at which an organization redesigns itself before external disruption requires it. But the path needs care in the reading. The company did not move to AI after the cloud matured. Nor did it move to AI before the cloud matured. Cloud and AI did not arrive in sequence; they overlapped. In 2014, the recognition of a mobile-first, cloud-first world was stated. In 2019, $1 billion went into OpenAI. The period of growing the cloud business and the period of preparing an AI platform sat on the same stretch of time. Redefinition proceeded as parallel processing, not sequential processing. Read on the assumption that stages are cleared one at a time, and these twelve years cannot be understood. There is a second characteristic, still in progress. The search for a balance between dependence and building in-house. The company started its AI business without models of its own. It secured the capability through an external partnership. In the October 2025 revision of that partnership, technical exclusivity loosened while the commercial tie strengthened. Then, in June 2026, the in-house MAI models were announced. A movement to thin the dependence and a movement to deepen it are running at once. Not being committed to either is what the present picture is. Assertion is not available. And an important reservation applies. As the source paper notes, levels do not line up across dimensions. The company’s capital allocation and business look Level 4, while the consumer-facing domain is harder to read that way. More Personal Computing declining may be one sign of it. Maturity is assessed across all five dimensions in balance; exceptional capability in one dimension does not raise the whole. We do not call Microsoft a Future Value Enterprise. That level names a state of actively designing the shape of an industry rather than reacting to external change. The company’s present position depends heavily on partnership relationships with others. And maturity is not a target to be reached. The appropriate level differs by industry and environment.

4.2 Where in the Future Value Chain the value was created

The Future Value Chain has this ordering. Purpose → Learning → Redefinition → Creation → Enterprise Value Microsoft’s twelve years read along that ordering. Purpose was rewritten in 2015. Learning was embedded in the organization as participation in open source, acceptance of external models, and the winning back of the developer community. Redefinition was executed as the abandonment of the phone hardware business and the movement of capital into the cloud. Creation appeared as Azure, Copilot, and Foundry. And Enterprise Value came last. An enterprise that reverses the ordering cannot follow this path. Put enterprise value first and the 2015 impairment is not choosable.

4.3 Giving the skeleton to the equations

The Future Value Creation Capability formula reads as follows. FVCC = Purpose × Learning × Redefinition × AI Integration × Ecosystem × Capital Allocation × Trust This is multiplication, not addition. If one term goes to zero, the whole goes to zero. Which term was near zero for Microsoft during the stall? Not Learning, and not Capital Allocation. Redefinition was as close to zero as it gets. However high the other terms, the product stayed small. And at present the most fragile term is Trust. On September 12, 2025, the European Commission made the company’s commitments regarding the tying of Teams legally binding. The general obligations run for 7 years; those on interoperability and data portability run for 10 (European Commission, September 12, 2025). Trust grows faster than capital. It is also lost faster than capital. Look at the leadership formula as well. Leadership = Purpose × Question Design × Capital Allocation × System Architecture × Trust Read against this formula, the most visible term in the company’s leadership is System Architecture. It designs layers and places others, including rivals, within them. But strength in this term does not compensate for weakness in the others. It is multiplication.

5 What it looks like in practice — the turning points, and

what was given up Set out in sequence. July 2013, “One Microsoft.” A declaration of becoming a devices and services company. The org chart changed. The assumption remained. February 2014, the change of chief executive. The first-day email states the recognition of a mobile-first, cloud-first world. Full-year revenue for the fiscal year ended that June was $86.83 billion; the annualized cloud revenue run rate was $4.4 billion. 2015, the revision of the mission. In July of that year, a $7.6 billion impairment in the phone hardware business and a reduction of up to 7,800 jobs. June 2018, the acquisition of GitHub for $7.5 billion. A move to take the developer ecosystem. July 2019, a $1 billion investment in OpenAI. Azure as the exclusive cloud, and Microsoft as the preferred partner in commercialization. On October 28, 2025, the two companies restructured the relationship. Microsoft’s investment is stated as worth about $135 billion, roughly 27 percent on an as-converted diluted basis. Intellectual property rights were extended through 2032. Rights to research IP hold until an independent expert panel verifies AGI, or until 2030, whichever comes first. At the same time, the right of first refusal as cloud provider was eliminated, and OpenAI became able to use other clouds for non-API products. OpenAI committed to purchase an incremental $250 billion of Azure (Microsoft, October 28, 2025). This revision is not the dissolution of dependence. It is a change in the form of the dependence. Technical exclusivity loosened; the commercial tie strengthened. June 2026, the announcement of the in-house MAI model family. Full-year revenue for the fiscal year ended that June was $331.8 billion and operating income was $155.2 billion. Azure passed $100 billion for the year. What was given up Redefinition is not addition. It is the act of deciding what to protect and what to let go. Set out what the company let go. First, the design of taking the phone market with its own hardware. Released with an impairment attached. Second, the assumption that value delivered outside Windows is not permissible. Office for iPad, the acceptance of Linux, the independent operation of GitHub. None is executable under the old assumption. Third, the capital structure of a software enterprise. The combination of low capital expenditure and high margin is something the company is giving up by its own hand. The third has not yet produced a result. It is a bet in progress. The conditions under which this advantage breaks We avoid praise. Four conditions under which the company’s present position breaks. First, a slowdown in demand. Capital expenditure occurs first; depreciation takes effect afterward. If AI demand grows more slowly than assumed, margins can deteriorate rapidly. Second, the asymmetry of the partnership. The relationship with OpenAI has loosened its exclusivity while containing an enormous purchase commitment. Variation in the counterparty’s financing or business plan reaches the company’s revenue as well. Third, trust and regulation. The European decision runs for 7 to 10 years. Bundled selling of enterprise software sits close to the core of the company’s revenue structure. If that remains a continuing object of regulation, the freedom to design prices falls. Fourth, self-destruction. The more agents substitute for work, the more the premise of billing by number of users is shaken. The company holds both seat-based and consumption-based billing, but nothing guarantees it can control the speed of the shift between them. About the future, little can be said with confidence. What is written here is not a forecast but a set of conditions to watch.

6 What transfers, and questions for the executive

What transfers from this case, and under what conditions? Four points. First, a stall does not appear as a problem in the results. Microsoft during the stall was profitable. Revenue was growing. The substance of a stall is the criterion of decision-making becoming fixed to past success. Watching the financial statements will not detect it. Second, reallocating capital begins with recognizing a loss. The $7.6 billion impairment was a judgment to take the pain first. Many incumbents defer this to the very end. Deferred capital cannot be moved toward the future. As the third of the Ten First Principles states, Capital Exists to Create Possibility. Capital exists in order to create possibility. Not in order to protect the correctness of the past. Third, Purpose can keep its core and change its expression. An objective that names a product will eventually bind the enterprise. The rewriting from “a computer on every desk” to “every person and every organization achieving more” was a translation, not an abandonment. It is worth checking whether your own objective contains a product name. Fourth, can you permit customer value that does not pass through you? Office for iPad and the acceptance of Linux both hang on this single point. Closed supplier groups, own-brand products, in-house-first doctrine. Much of what an incumbent protects has the same structure. From here, three questions. Not abstract ones. Questions that can be answered at your next executive meeting. Question 1 — What share of your budget is attached for the same reasons it was attached five years ago? Capital allocation is the decision about which future to allocate possibility to. A budget that continues for the same reason means the same future is being chosen again. If the reason it cannot be cut is “we already invested,” it is not capital. It is a chain. Question 2 — Do you provide even one instance of customer value that does not pass through your own product? If there is not one, the enterprise’s objective still names a product. As long as the objective names a product, redefining the business is structurally impossible. Question 3 — What assumptions about our enterprise are becoming obsolete? This is the central question of Recognize, the first stage of Enterprise Redefinition. An enterprise that cannot answer it has not begun to redefine. And the answer does not come from AI. AI can test an assumption. Deciding which assumption to doubt belongs to human beings. What Microsoft redefined, in one line. It changed how it named its objective — from a company that protects a product to a company that helps customers achieve. Business, organization, capital, and leadership all moved from that single point, in that order. The order was not reversed. And what stood at the entrance to the turn was not a new design. It was the recognition of an old failure. As the sixth of the Ten First Principles states: Enterprise Exists to Redefine Itself. An enterprise exists in order to redefine itself. What a stalled enterprise needs is not a story of revival. It is a meeting that decides what to discard.

In brief

What Microsoft redefined is how it names its objective — moving what it protects from the product to the customer’s achievement.
Within what public information allows us to read, the stall period is close to an Improvement Enterprise and the present close to a Continuous Redefinition Enterprise.
Value arose in the reallocation of capital that begins with recognizing a loss through impairment — that is, at Redefinition.
If a slowdown in demand, the asymmetry of the partnership, and regulation coincide, this position can break.

Key concepts

Enterprise Redefinition / the Enterprise Redefinition Maturity Model / the Future Value Chain / Future Value Creation Capability / the Ownership-to-Usage Pattern (→ Vol. VI, Ch. 059)

The chain of ideas

Purpose → Learning → Redefinition → Future Capital → Enterprise Value

Related first principles

Principle 3 — Capital Exists to Create Possibility. Principle 5 — Learning Is the Ultimate Competitive Advantage. Principle 6 — Enterprise Exists to Redefine Itself. Principle 8 — Trust Compounds Faster Than Capital.

Related chapters

Vol. V, Ch. 041 “What Is Enterprise Redefinition?” — the criterion separating improvement from redefinition sits in that chapter
Vol. V, Ch. 046 “What Does It Mean to Redefine a Business Model?” — why the structure of billing determines the structure of attention
Vol. VI, Ch. 056 “What Does It Mean to Redefine Investment?” — how to read the recognition of a loss as capital reallocation
Vol. VI, Ch. 059 “Cases of Enterprise Redefinition” — the definition of the Ownership-to-Usage Pattern and its typical failure

Papers and companion volumes

Kadowaki, N. (2026a). Future Value Theory: A Management Framework for Enterprise, Capital, and Society in the Age of AI. VURA Working Paper Series. SSRN: https://ssrn.com/abstract=7120980 / Zenodo: https://doi.org/10.5281/zenodo. 21255662
Kadowaki, N. (2026b). Enterprise Redefinition: Toward an Enterprise Evolution Theory for the Age of AI. VURA Working Paper Series. (Published on Zenodo; under review at SSRN)
100 Questions on Management in the Age of AI, #059 “Can Large Enterprises Survive in the Age of AI?”

Chapter 082 What Did Microsoft Redefine?