Chapter 1 — From Disc to Cloud

A disc says goodbye

In May 2013, Adobe made a decision that looked, at the time, close to reckless. It announced that its most profitable and best-known products — the Creative Suite that housed Photoshop and Illustrator — would no longer be sold as a perpetual license. Users could no longer pay once for a disc and a serial number and use the software forever. From then on, to use Photoshop you had to subscribe to Creative Cloud, by the month or by the year.

Today this is the default form of almost all software, and we take it for granted. But put yourself back in 2013 and you can feel how bold the decision was. Adobe was then a mature software company with roughly $4.4 billion in annual revenue, stable profits, and the deep trust of professional creatives.^[1] Many of its users were freelancers and small studios, plenty of whom upgraded only once every few years — some running a single disc for five or ten. Now Adobe was telling them: you must keep paying every month, or the software stops working.

The backlash was fierce. Angry users organized petitions accusing Adobe of “holding hostage” the tools they earned a living with; more than a few analysts worried that forced subscription would push swaths of unwilling users toward piracy or competitors. Financially, the short-term cost was immediate: because the way revenue was recognized had changed, the large one-time payment of an outright purchase was now spread out into small monthly subscription revenue, and Adobe’s revenue actually fell in the first year or two of the transition — from around $4.4 billion before the switch down to just over $4 billion at one point.^[1] What Wall Street fears most is falling revenue, and Adobe had chosen it deliberately.

Then time delivered its verdict. As more and more users were “locked” into a monthly paying relationship, Adobe’s revenue stopped being one-time, pulse-like version sales and became a smooth, predictable, ever-accumulating curve. By this book’s data cutoff, Adobe’s annual revenue had crossed a scale that would have been hard to imagine before: roughly $19.4 billion in fiscal 2023, $21.5 billion in fiscal 2024, and $23.8 billion in fiscal 2025.^[2] In a little over a decade, revenue had grown to more than five times its pre-transition level — and the quality of that growth, in predictability, retention, and margin, far surpassed what came before.

Adobe’s gamble is a microcosm of the whole software industry’s move from “disc” to “cloud.” It tells us the one thing this chapter wants to make clear: the migration from perpetual license to subscription was not merely a change in how you charge — it was a deep transformation in how risk is allocated, how value accumulates, and who captures that value. That transformation is what produced the “SaaS compounding machine” the market later worshipped; but it is also within the mechanics of that transformation that the distortions and traps this book returns to again and again were first planted.

“No Software”: a starting point that was different from the beginning

If Adobe is a mature company turning toward subscription midway, then the company that wrote subscription into its DNA from the very start was another — Salesforce.

In 1999, Marc Benioff founded Salesforce in a small San Francisco apartment under a deliberately provocative banner: “No Software.”^[3] The slogan did not mean they made no software — quite the opposite; software was exactly what they made. Its real meaning was: users no longer need to buy, install, or maintain any local software. You don’t buy a disc, don’t have IT deploy, tune, and upgrade it across server after server; you just open a browser, log in, and use a complete customer relationship management (CRM) system. The software runs on Salesforce’s servers, Salesforce maintains and upgrades it, and you pay by the seat, by the month.

This model, which looks self-evident today, was pure heresy at the turn of the century. Enterprise software was then a “heavy” business: a company paid hundreds of thousands, even millions, to buy the perpetual right to use a piece of software, paid a large sum more to have it deployed and implemented, and then paid a substantial annual “maintenance fee” for updates and support. The vendor recognized most of the revenue as current-period revenue the moment the contract was signed; whether the customer actually used it well or got real value out of it afterward had little to do with the vendor’s current results.

Salesforce turned that whole logic upside down. Instead of collecting one large license fee up front, it broke the charge into a long stream of subscription revenue that flowed out over time. This seemingly mere “installment plan” actually reshaped the relationship between software vendor and customer — and in the next section we’ll see exactly what it changed.

Three layers of restructuring: delivery, sales, revenue recognition

The migration from disc to cloud happened concretely across three interlocking layers. Understand these three and you understand the bedrock of the SaaS business model.

Layer one, delivery: from “you deploy” to “we host.” In the license era, software was an asset the customer had to install and operate. In the SaaS era, software is a service the vendor hosts in the cloud and delivers over the network. The technical precondition for this shift was the maturing of cloud infrastructure — above all the arrival of public clouds like Amazon’s AWS around 2006, which for the first time let a software company serve tens of thousands of customers elastically without building its own vast data centers.^[4] The change in delivery brought a crucial economic consequence: multi-tenant architecture (many customers sharing the same underlying system) made the marginal cost of serving “one more customer” extremely low. This is the technical source of SaaS’s high gross margins, and the physical foundation of what was later hailed as a “compounding machine.”

Layer two, sales and pricing: from “buy once” to “subscribe continuously.” A license is a one-and-done sale; a subscription is an ongoing relationship that must keep renewing, must keep proving its worth. The significance of this change runs far deeper than “collecting in installments.” It means the vendor’s revenue no longer depends on “how many new deals we signed this quarter” but, to a large degree, on “of the customers we’ve already signed, how many are still renewing and how many are willing to use more.” The center of gravity in sales shifted from “get the thing sold” to “keep the customer, and get them to use more.” This shift gave rise to the most important set of operating metrics in the later SaaS world — retention, net revenue retention, and the “land and expand” growth paradigm — which we’ll unpack one by one in Chapters 2 and 3.

Layer three, revenue recognition: from “booked now” to “amortized over time.” This is the most technical layer, and the one investors most often misread. Accounting rules require subscription revenue to be recognized gradually, over the period the service is actually provided. That is, when a customer prepays a full year of subscription at the start of the year, the vendor cannot count it all as revenue the moment the cash arrives — it can only recognize one-twelfth each month; the remaining unrecognized portion sits on the balance sheet as a liability called deferred revenue.

This arrangement creates a phenomenon that is extremely important to investors yet often overlooked: in a SaaS company, the inflow of cash and the recognition of revenue are separated in time. The customer’s prepaid cash arrives early (a huge boon to the company’s cash flow — effectively the customer providing interest-free financing for the company’s growth), but the corresponding revenue is recognized slowly over the coming year. This is also why, looking at a single quarter’s revenue, you often underestimate the true business scale of a fast-growing SaaS company; and to see clearly how much “signed but not yet recognized” future revenue it is holding, you need to look at two other figures — deferred revenue, and the broader remaining performance obligations (RPO). These two metrics are the keys this book uses again and again when it later dissects SaaS financials, and we’ll cover them in detail in Chapters 3 and 8.

【Key concept】SaaS, subscription, and deferred revenue

SaaS (Software as a Service): software delivered over the network by subscription and hosted and maintained by the vendor in the cloud. Users need not buy and deploy local software; they pay by time (month/year) to use it.

Subscription vs. perpetual license: a perpetual license is a one-time purchase of the right to use software, with revenue essentially recognized all at once at signing; a subscription is ongoing periodic payment, with revenue recognized gradually over the service period. The former makes revenue pulse-like and hard to predict; the latter makes it smooth and cumulative.

Deferred revenue: the portion a customer has already paid for but the vendor has not yet delivered the corresponding service on. It is a liability on the balance sheet (representing the “service obligation” the vendor owes the customer) and is recognized as revenue gradually as the service is provided month by month. Growth in deferred revenue is often an early signal of a healthily expanding subscription business.

What problem did SaaS actually solve?

Any business model that wins does so because it creates value for both sides of the transaction at once. SaaS’s success is no exception — it solved a real pain point for both customers and vendors.

For customers, SaaS lowered the threshold of trying and the risk of owning. In the license era, buying a piece of enterprise software was a major, hard-to-reverse capital-expenditure decision: pay a large sum up front, endure a long and expensive deployment, and if the software turned out to be bad, the sunk cost was too high to admit failure. SaaS turned that heavy, one-time capital expenditure into a light operating expense you could stop at any time. You could start with just a few seats, expand if it worked, stop renewing if it didn’t. Nor did updating the software remain the customer’s worry — the vendor upgrades continuously in the cloud, and every user is on the latest version at all times. This “low barrier to enter, expand on demand, update continuously” experience massively unlocked demand for software, especially among the small and mid-sized businesses that could not afford large up-front outlays.

For vendors, SaaS traded predictable, cumulative revenue for a higher-quality business. A license software company starts each quarter as if from zero — last quarter’s big deal does not automatically become this quarter’s revenue, and the sales team must prove itself all over again, year after year. A SaaS company, so long as its customers don’t churn, sees last year’s revenue automatically recur this year, with newly signed customers stacked on top. Revenue thus takes on the character of a “stock”: it is a reservoir that keeps filling, not a series of rains that must fall anew each time. Better still, customers’ prepaid cash provides cheap funding for the company’s expansion, and multi-tenant architecture keeps the cost of serving each additional customer extremely low. Predictable, cumulative, high-margin, and self-financing — these four properties stacked together form the narrative the whole market fell for: software is a compounding machine.

But remember that stretch of falling revenue in Adobe’s early transition. It reminds us that starting this machine has a cost, and that the “compounding” once it is running is far from automatic. That is exactly the question we need to press on in the next chapter.

【Company field notes】Adobe: a transition that traded short-term pain for long-term compounding

Adobe’s subscription transition is the classic textbook case of “how a license company becomes a SaaS company.” Its value lies not in the outcome — “the transition succeeded” — but in how clearly it lays bare the cost structure of such a transition:

Short term: the change in revenue recognition spread one-time large license revenue into monthly subscription revenue, and reported revenue fell in the first year or two. This is the “J-curve” trough almost every company moving from license to subscription must pass through.

Long term: once the subscription relationship is established, revenue becomes smooth, predictable, and steadily cumulative as users expand and prices rise. Adobe’s annual revenue grew from roughly $4.4 billion before the transition to about $23.8 billion in fiscal 2025.^[1][2]

Investment takeaway: when a license company announces a shift to subscription, the short-term drop in revenue and profit may be precisely the prelude to long-term value — provided its product has enough customer stickiness to survive the transition trough. The same transition story reappears in Chapter 12 when we discuss the insurance-software company Guidewire moving from license to cloud subscription, though the difficulty and the ending are not necessarily the same.

Risk was transferred — and transferred to the vendor

We said earlier that SaaS is a “reallocation of risk.” Now we can put it more concretely.

In the license model, risk was borne mainly by the customer: the customer paid a large sum, and whether the software was good and worth the price was the customer’s risk; the vendor, having received the cash, was essentially “home free.” In the subscription model, the scales of risk tip heavily toward the vendor: the customer has prepaid only a short stretch of time, and if the software is bad or fails to keep creating value, the customer can simply stop renewing. The vendor must continuously and repeatedly prove its value over the life of the relationship to keep it going and expand it.

This risk transfer is the hinge for understanding every strength and weakness of SaaS.

It is the source of all of SaaS’s strengths: precisely because the vendor must keep proving value, a good SaaS company is forced by the market to keep improving the product, staying close to customers, raising the experience; precisely because revenue depends on renewal, a SaaS company with excellent retention can enjoy that enviable, self-accumulating revenue curve.

But it is also the root of all of SaaS’s risks: if a SaaS company’s product isn’t sticky enough and switching costs aren’t high enough, then “the customer can leave at any time” turns from a slogan into a cruel reality — high churn leaks like water from a bucket, draining away every hard-won customer and forcing the company to run forever on a treadmill of burning cash to acquire new ones. And because the cost of acquiring a customer (the sales and marketing outlay) is often incurred up front in one lump, while the revenue that customer brings is recovered only slowly over many future years, a rapidly expanding SaaS company is very likely loss-making on paper for a long time — it is using today’s losses to buy tomorrow’s recurring revenue. Whether that trade is worth it depends on how long the customer stays and how much they expand — which is exactly the core question Chapter 2’s “compounding machine” presses on.

【Mistakes investors make】Mistaking “recurring” for “high quality”

Many investors see the words “recurring revenue” and automatically stamp them “high quality, predictable, deserving of a premium multiple.” This is one of the most common — and most dangerous — errors in SaaS investing.

Recurring revenue describes only the form of the revenue (that it repeats periodically); it says nothing about the quality of the revenue. A company that loses 20% of its customers a year and must keep burning cash to acquire new ones to sustain growth also has “recurring” revenue — but the quality of that revenue is worlds apart from a company whose customers are highly sticky and whose existing customers naturally expand each year. To judge revenue quality you have to look at retention (how many customers left), net revenue retention (whether the ones who stayed expanded or shrank), and the cost paid to acquire that revenue — all of which we can only truly dissect once we’ve built the full metric system in Chapters 2 and 3.

Reverse view: is subscription really better for everyone?

An honest book should not sing only one model’s praises. Behind the victory narrative of subscription lie downsides worth watching.

For customers, subscription is not always the better deal. A user who upgrades only once every few years, with stable feature needs, might have run a single disc for many years in the perpetual-license era; in the subscription era they must pay year after year, and the total spend over the long run may well exceed the one-time purchase price of the past. Subscription turned software from “an asset you own” into “a service you rent” — and once you stop paying, you may lose access even to the data and work you accumulated over years. This “lock-in” is a moat for the vendor but, for the customer, a cost in the form of lost control. It is exactly this tension that explains why the anger of Adobe’s users during the transition was not unreasonable.

For vendors and investors, beware too of treating “switch to subscription” as a cure-all. The subscription model can amplify a good company’s advantages, but it just as surely amplifies a mediocre company’s problems. If the product itself lacks stickiness, going subscription only exposes high churn nakedly in the renewal rate; if the market is already saturated and growth can be sustained only by relentless price increases, then the growth of “recurring revenue” may be fragile and borrowed from the future. Put differently, subscription is a magnifying glass — it makes good businesses better and leaves bad businesses nowhere to hide.

【Reading the financials】From license to subscription, look at these three first

When you study a company moving from license to subscription, or a pure SaaS company to begin with, open its 10-K/10-Q and look first at these three places:

Revenue composition: how much is subscription revenue, license revenue, professional-services revenue, and how fast is each growing? Is subscription’s share rising steadily?

Deferred revenue and RPO: deferred revenue on the balance sheet, and remaining performance obligations (RPO) in the notes, reflect “signed but not yet recognized” future revenue. Their growth often reveals the temperature of the business earlier than current-period revenue does.

Revenue-recognition policy: find the revenue-recognition accounting policy in the notes and confirm how subscription, service, and license revenue are each recognized. For a company mid-transition especially, look carefully at whether falling revenue means the business got worse, or merely that the recognition method changed.

How to read these three and what they reveal, we unpack hands-on in Chapters 3 and 8.

Chapter summary

The migration from perpetual license to subscription was not a simple change in how you charge, but a reallocation of risk, a restructuring of how value accumulates, and a redivision of who captures that value.
The migration happened across three layers: delivery (from customer self-hosting to vendor cloud hosting, bringing extremely low marginal cost and high margin), sales (from one-time purchase to ongoing subscription, shifting the focus from “get it sold” to “keep them and get them to use more”), and revenue recognition (from booked-now to amortized-over-time, creating the timing gap between cash inflow and revenue recognition, and giving rise to the two key metrics of deferred revenue and RPO).
SaaS won because it simultaneously lowered the threshold of trying and the risk of owning for customers, and traded predictable, cumulative, high-margin, self-financing high-quality revenue for vendors — these four properties stacked together form the bedrock of the “software is a compounding machine” narrative.
But the core of this machine is the transfer of risk to the vendor: the vendor must keep proving value to keep the revenue. This is both the source of all of SaaS’s strengths and the root of all its risks.
The single most important warning: “recurring revenue” describes the form of revenue, not its quality. Recurring does not equal high quality. Distinguishing the two is the starting point of the entire analytical framework this book is about to build.

Notes and sources

Note: this chapter is historical and conceptual narrative. All primary financial data cited are verified against SEC EDGAR; the historical events of the transition era (Adobe’s transformation, Salesforce’s founding/IPO, the launch of AWS) are all cross-verified against primary and multiple sources.

[1] Adobe’s subscription transition (announced in 2013, Creative Cloud moving fully to subscription): ✅ verified (2026-07-15) — FY2013 total revenue fell about 8%, subscription’s share of revenue rose from 15% in FY2012 to 28% in FY2013, paid subscriptions grew from roughly 300,000 to about 1.4 million, and Creative ARR rose from about $155 million to about $768 million. Pre-transition FY2012 revenue was roughly $4.4 billion, dipping to just over $4 billion in the early transition, consistent with the “~8% decline.” Source: Adobe Inc. FY2013 10-K (SEC EDGAR, CIK 796343).
[2] Adobe’s recent revenue: about $19.4 billion in FY2023 (period end 2023-12-01), about $21.5 billion in FY2024 (period end 2024-11-29), about $23.8 billion in FY2025 (period end 2025-11-28). Source: Adobe Inc. 10-K, verified against SEC EDGAR XBRL data, retrieval date 2026-07-15.
[3] Salesforce ✅ verified: founded 1999-03-08 by Marc Benioff and others, with the “No Software” brand positioning; NYSE IPO 2004-06-23, raising over $110 million. Source: Salesforce company history / prospectus (SEC EDGAR, CIK 1108524), cross-verified.
[4] Public cloud infrastructure ✅ verified: AWS launched S3 on 2006-03-14 and EC2 in 2006-08. Source: Amazon official announcements.
Supplement: Salesforce’s verifiable revenue history — about $4.07 billion for the fiscal year ended 2014-01-31, about $5.37 billion for FY ended 2015-01-31, about $6.67 billion for FY ended 2016-01-31, about $8.39 billion for FY ended 2017-01-31. Source: Salesforce, Inc. 10-K, verified against SEC EDGAR XBRL, retrieval date 2026-07-15. (Not cited individually in this chapter’s body; kept as source material for the Salesforce case in Chapter 9.)