Software Industry: How the Business Models, Distribution and Economics Work

1.Industry in 30 Seconds

A software company may sell:

a monthly subscription;
a per-user enterprise contract;
a consumption-based service;
a perpetual or term license;
maintenance and support;
an application platform;
cloud infrastructure;
professional services;
AI usage.

Two software companies can both report recurring revenue while having very different:

deployment costs;
gross margins;
sales cycles;
infrastructure requirements;
customer concentration;
switching costs.

The central question is:

How does a software company turn code into a product that can be sold repeatedly without rebuilding the entire solution for every customer?

A simplified software value chain looks like this:

CUSTOMER PROBLEMfinance · collaboration · security · operations · design · analytics

↓

SOFTWARE DEVELOPMENTcode · models · interface · workflow · integrations

↓

DELIVERYSaaS · cloud · on-prem · hybrid · edge

↓

IMPLEMENTATIONconfiguration · data migration · integration · training

↓

USEseat · transaction · workflow · API · compute · application

↓

MONETIZATIONsubscription · consumption · license · support · service

↓

RENEWAL / EXPANSIONmore users · more modules · more usage · higher tier
↺

The industry is attractive because software can often be reproduced at low physical marginal cost.

But that does not mean every software business is cheap to scale.

Enterprise software may require:

expensive sales teams;
solution engineers;
cloud compute;
customer success;
professional services;
security;
implementation.

The code can be copied cheaply.

Winning, deploying and retaining the customer may not be cheap.

2.What the Industry Actually Produces

Software does not produce one universal output.

It produces repeatable digital capability.

Workflow automation

Software can replace manual sequences with structured processes.

Examples:

approve invoice;
route support ticket;
schedule production;
close accounting period;
manage customer relationship.

The commercial value comes from:

less manual work
+
fewer errors
+
faster process

Information organization

Software can turn large information sets into usable systems.

Examples:

database;
CRM;
ERP;
analytics;
knowledge management.

The value is not storage alone.

It is making information useful.

Decision support

Software can:

forecast;
simulate;
recommend;
rank;
optimize.

This moves the product closer to business outcomes.

Security and control

Security software helps organizations manage:

identity;
devices;
networks;
threats;
permissions.

Here the product is partly risk reduction.

The customer pays to avoid expected loss, downtime or regulatory failure.

Developer capability

Some software is sold to people building other software.

Examples:

developer tools;
databases;
APIs;
deployment platforms.

The product increases the productivity or capability of another software team.

AI capability

Modern AI software can generate:

text;
code;
predictions;
decisions;
agents;
automations.

The business challenge is turning model capability into a reliable workflow the customer can trust.

3.The Value Chain

Software has a less visible physical supply chain than manufacturing, but it still has distinct economic roles.

3.1 Core software developer

The vendor creates:

code;
interface;
algorithms;
models;
security;
APIs.

Research and development is often one of the largest operating costs.

The product has to improve continuously because competitors can copy features and customer needs change.

3.2 Infrastructure provider

Cloud software runs on:

compute;
storage;
networking;
accelerators;
data centers.

A SaaS company may not own these physical assets.

It still pays for them.

As AI workloads grow, infrastructure cost can become more important.

Microsoft's 2026 filing explicitly notes large AI-infrastructure investment as cloud demand increased. [S21]

3.3 Application vendor

Application software turns infrastructure into a specific customer product.

Examples:

CRM;
HR;
workflow;
design;
finance;
operations.

The vendor owns the application logic and customer experience.

3.4 Platform vendor

A platform allows customers or partners to build more software on top.

Examples can include:

developer SDKs;
low-code tools;
data platforms;
integration frameworks.

Platform economics differ from a single-purpose application because customer value can increase as more applications are built around the platform.

3.5 Implementation partner

Complex software may require:

configuration;
data migration;
integration;
training;
process redesign.

That work can be delivered by:

the software vendor;
consulting firms;
system integrators;
customer teams.

A software company has to decide how much implementation to internalize.

3.6 Channel / reseller

Some vendors sell directly.

Others use:

distributors;
cloud marketplaces;
resellers;
system integrators;
OEMs.

Microsoft's filing describes both direct and partner-assisted software distribution. [S21]

Distribution changes the cost of customer acquisition and how much control the vendor has over the relationship.

3.7 Customer success / support

Recurring software depends on renewal.

That means the vendor has to ensure:

customer bought software
≠
customer received value

Customer success tries to close that gap.

4.From Code to Recurring Revenue

1. CUSTOMER PROBLEM

↓

2. PRODUCT DESIGN

↓

3. SOFTWARE DEVELOPMENT

↓

4. GO-TO-MARKETsales · self-service · partner · marketplace

↓

5. CONTRACT / PURCHASE

↓

6. IMPLEMENTATION

↓

7. ACTIVE USE

↓

8. VALUE REALIZATION

↓

9. RENEWAL

↓

10. EXPANSIONmore seats · usage · modules · workflows

The biggest mistake is to assume step 5 ends the commercial process.

In subscription software:

saleis the start of
renewal risk

A customer that does not use the product will eventually leave.

5.Who Pays Whom?

Different software models create different money flows.

Seat-based SaaS

CUSTOMER
↓ subscription per user / tier
SOFTWARE VENDOR
↓ pays
cloud + support + R&D + sales

The key variables are:

number of paid users;
price per user;
renewal;
upsell.

Platform subscription

ENTERPRISE
↓ annual platform contract
SOFTWARE VENDOR

The contract may allow broad use across an organization.

Growth comes from:

more use cases;
more departments;
more modules.

Palantir's software contract model is closer to this than a simple consumer-style monthly subscription.

Consumption model

CUSTOMER
↓ pays based on usage
SOFTWARE / CLOUD PROVIDER

Usage can be:

compute;
API calls;
storage;
transactions;
model tokens.

Microsoft recognizes some cloud services based on consumption. [S21]

License + support

CUSTOMER
↓ license payment
VENDOR

CUSTOMER
↓ recurring maintenance / support
VENDOR

Oracle remains a clear example of software-license and support economics alongside cloud subscriptions. [S24]

Subscription + professional services

CUSTOMER
↓ subscription
VENDOR

CUSTOMER
↓ implementation / service
VENDOR or PARTNER

Salesforce and ServiceNow both report professional-services revenue in addition to subscription revenue. [S22][S23]

The strategic issue is whether services accelerate software adoption or become a low-margin dependency.

Marketplace / ecosystem revenue

A platform can enable third-party applications.

Possible economics include:

transaction fees;
revenue share;
partner influence;
greater platform stickiness.

The financial contribution varies widely by vendor.

6.Business Models by Role

Software model	Primary unit	Revenue logic	Main scaling constraint
Seat-based SaaS	User / seat	Subscription	Customer acquisition + churn
Enterprise platform	Organization / contract	Multi-year subscription	Implementation + expansion
Consumption software	Usage	Pay per use	Infrastructure cost + volatility
License + support	License + maintenance	Upfront / term + recurring support	Upgrade cycle
Developer platform	Developer / workload	Subscription / usage	Ecosystem adoption
AI application	Seat / usage / workflow	Subscription + consumption	Model cost + trust
Security software	Endpoint / user / workload	Subscription	Threat evolution + integration
Deep operational software	Platform / deployment	Subscription + maintenance + service	Implementation intensity

7.Where Economic Power Comes From

Switching cost

Software becomes sticky when it contains:

data;
workflows;
configurations;
user habits;
custom applications.

Switching cost is strongest when replacing the product means redesigning business processes.

Workflow ownership

A vendor that owns an important daily workflow can become difficult to remove.

Examples:

CRM
ticketing
finance close
production planning
identity

The closer software sits to actual work, the stronger the relationship can become.

Data gravity

When years of customer data and logic accumulate inside a platform, migration becomes more difficult.

Open standards and APIs can reduce lock-in, but data gravity remains economically meaningful.

Ecosystem

Partners and developers can expand:

integrations;
applications;
implementation capacity.

A large ecosystem makes the core platform more useful.

Distribution

Self-service software can scale cheaply.

Enterprise software may require:

direct sales;
solution engineering;
partners.

A powerful distribution engine can become as important as product quality.

Brand and trust

Mission-critical customers care about:

uptime;
security;
privacy;
regulatory compliance.

Trust lowers the perceived risk of adopting important software.

Installed base

An existing customer is easier to expand than a completely new account if the first product creates value.

Software vendors often grow through:

land
↓
expand

Proprietary operating model

A company can differentiate through a unique architecture rather than one feature.

Palantir's Ontology is one example.

ServiceNow's workflow platform is another.

Microsoft's bundled productivity / cloud ecosystem is another.

8.Industry Bottlenecks

Customer acquisition cost

Enterprise sales can take months.

A vendor may spend heavily before any contract is signed.

Implementation

Complex software can fail not because the code is weak but because implementation is poor.

The customer must:

migrate data;
integrate systems;
redesign workflow;
train users.

Cloud / compute cost

Software is not physically free to deliver.

AI increases dependence on:

GPUs;
compute;
storage;
networking.

Security

A breach can destroy trust.

Security has to keep pace with:

new threats;
new integrations;
AI agents.

Talent

Software products require:

engineers;
product managers;
sales;
implementation;
support.

High-skill labor is often a key constraint.

Platform complexity

As products expand, vendors can create:

overlapping modules;
confusing pricing;
technical debt.

Large software suites have to manage complexity without slowing innovation.

Customer budget

Enterprise software is often discretionary capital or operating spend.

Economic weakness can slow:

new deals;
expansion;
renewals.

9.Industry Cycles

Software is more recurring than many physical industries but still cyclical.

Enterprise spending cycle

economic confidence improves
↓
IT budgets expand
↓
new software projects start
↓
vendors hire / invest
↓
budget pressure rises
↓
projects consolidate
↓
sales cycles lengthen

Product cycle

A new category can create a large adoption wave.

Examples:

cloud;
cybersecurity;
generative AI.

Eventually the market matures and growth depends more on replacement and expansion.

Platform transition cycle

Vendors periodically shift business models:

perpetual license
→ subscription

on-prem
→ cloud

human workflow
→ AI-assisted workflow

Transitions can temporarily pressure margins or revenue recognition.

Infrastructure cycle

AI software demand can depend on availability and cost of compute.

This makes the software industry more tied to physical infrastructure than older SaaS models suggested.

10.Major Commercial Segments

These segments describe business mechanics, not GICS subcategories.

Productivity and collaboration

Software used every day by broad employee populations.

Economics often center on:

seat count
×
subscription price

Enterprise workflow / application software

Products own specific business processes:

sales;
service;
HR;
IT operations;
finance.

The value comes from workflow integration and switching cost.

Data and operational platforms

These products integrate and operationalize data across the enterprise.

They can monetize at:

platform level;
workload level;
enterprise contract level.

Palantir fits here.

Systems / infrastructure software

Software manages:

operating systems;
security;
databases;
deployment;
infrastructure.

It is often highly technical and deeply embedded.

Developer platforms

These products sell to software builders.

Their success depends on ecosystem and developer adoption.

AI software

AI increasingly appears as a layer across all segments rather than one isolated category.

Business models include:

seat add-on;
usage;
agent transaction;
bundled platform.

The GICS providers are reviewing AI classification in 2026, but no consultation proposal is treated as effective in this document. [S25]

11.Competition by Layer

Application vs. application

Companies compete for one workflow or department.

Suite vs. best-of-breed

A customer can buy:

one broad platform

or:

several specialized tools

The suite wins on integration.

Best-of-breed can win on depth.

Cloud vs. on-prem / hybrid

Some customers prioritize simplicity.

Others require:

regulation;
sovereignty;
local infrastructure.

Software that supports more deployment modes can reach broader markets.

Subscription vs. consumption

Customers increasingly compare pricing models as well as features.

Predictable seat pricing and variable usage pricing shift risk differently.

Software vendor vs. internal build

A large enterprise can always ask:

Should we build this ourselves?

Commercial software must deliver faster or better economics than internal development.

Software vs. consulting / systems integration

Complex platforms can overlap with consulting.

The dividing line is whether the vendor is primarily monetizing reusable software or human labor.

12.Representative Companies

Business model	Representative companies	What they illustrate
Productivity / cloud suite	Microsoft	Subscription + consumption + licensing
CRM / enterprise apps	Salesforce	Subscription + support
Workflow platform	ServiceNow	High recurring subscription revenue
Database / cloud / software	Oracle	License + support + cloud applications / infrastructure
Operational data / AI platform	Palantir	Enterprise contract + deep implementation
Creative software	Adobe	Subscription application economics
Security software	CrowdStrike, Palo Alto Networks	Recurring security platform economics

These are examples, not rankings.

13.Companies That Occupy Multiple Roles

Microsoft

application software
+ systems software
+ cloud infrastructure
+ developer platform
+ AI

The same company sells seat subscriptions and consumption-based cloud services. [S21]

Oracle

database
+ enterprise application
+ cloud infrastructure
+ license / support

Oracle shows that old and new software business models can coexist. [S24]

Salesforce

CRM
+ workflow
+ data
+ integration
+ AI

Salesforce remains predominantly subscription and support revenue. [S22]

ServiceNow

workflow platform
+ application platform
+ AI

In 2025, 97% of ServiceNow revenue came from subscription revenue. [S23]

Palantir

data operations
+ application platform
+ AI
+ deployment software
+ mission software

Palantir is one example of software moving closer to real-world operations.

It is not the template for the entire Industry.

→ /companies/palantir/

14.Common Misconceptions

"Software has no marginal cost."

No.

Cloud, AI compute, support and customer success all cost money.

"Subscription revenue means all SaaS businesses are the same."

No.

A $20 self-service seat and a multi-year enterprise platform contract have different economics.

"Professional services are always bad for software margins."

Not necessarily.

Services can accelerate adoption.

The question is whether they remain necessary in direct proportion to revenue.

"Cloud means software must run in one public cloud."

No.

Enterprise software can be:

hosted;
self-hosted;
hybrid;
on-prem.

"AI software is a separate industry from all other software."

Not necessarily.

AI is increasingly embedded in existing software categories.

"High switching cost means customers never leave."

No.

Customers will switch if value deteriorates enough.

15.What Is Changing

AI is becoming an application layer rather than a standalone feature

The first enterprise AI wave focused on chat.

The next shift is toward:

agent
↓
workflow
↓
action

This increases the value of:

context;
governance;
APIs;
process integration.

Pricing is becoming more hybrid

Traditional SaaS:

seat

AI / cloud:

seat
+
usage
+
compute

Vendors have to balance predictability with infrastructure cost.

Infrastructure matters more again

The SaaS era abstracted hardware away.

AI brings attention back to:

GPUs;
data centers;
power;
networking.

Software gross margins can be affected by infrastructure intensity.

Enterprise platforms are absorbing adjacent categories

CRM, workflow, data and AI vendors increasingly overlap.

Competition shifts from:

feature vs. feature

to:

platform vs. platform

Customers demand more deployment control

Regulated customers increasingly care about:

data residency;
sovereignty;
model choice;
hybrid deployment.

This creates opportunity for software that can run across environments.

16.Evidence

GICS uses 451030 Software as the canonical Industry. [S19]
Microsoft illustrates subscription, consumption and on-premises license models. [S21]
Salesforce generated $39.388B of subscription and support revenue in fiscal 2026. [S22]
ServiceNow generated 97% of 2025 revenue from subscriptions. [S23]
Oracle combines software license / support with cloud application and infrastructure revenue. [S24]
Palantir sells cloud and on-premises software with ongoing maintenance and professional services. [S1]

17.Explore Companies

Enterprise Platforms

Microsoft
Salesforce
ServiceNow
Oracle
Palantir
Broadcom

Security Software

CrowdStrike
Palo Alto Networks

Creative / Specialized Applications

Adobe
Intuit

Where Selected Companies Fit

Palantir

Primary commercial role Operational data and AI software platform

Revenue model Multi-period software access / licenses + maintenance + professional services

Deployment model Cloud, customer cloud, on-premises, edge and disconnected environments

Why it matters Shows how software can sit deeply inside physical and institutional operations rather than only support office workflows.

Microsoft

Primary roles Productivity, systems software, developer platform and cloud

Revenue models Subscription, consumption and licensing

Why it matters Shows the breadth of possible software monetization inside one company.

Salesforce

Primary role Enterprise application / CRM platform

Revenue model Subscription and support

Why it matters Shows classic cloud subscription economics at global enterprise scale.

ServiceNow

Primary role Enterprise workflow platform

Revenue model Predominantly subscription

Why it matters Shows how workflow ownership can produce recurring platform economics.

Oracle

Primary roles Database, enterprise applications and cloud infrastructure

Revenue model Software license + support + cloud

Why it matters Shows how legacy license economics can coexist with cloud transformation.

Industry Source Registry

S19 — S&P Dow Jones Indices GICS Methodology — April 2026

Use for canonical Industry identity 451030 Software.