Palantir: the software company that tries to give institutions a nervous system

Palantir is one of the most powerful, unusual, and controversial software companies operating today. It builds systems that combine enormous quantities of scattered data, organize that data into a working model of reality, and help people make operational decisions from it.

Its name comes from Tolkien’s palantíri, the seeing stones that allowed rulers to observe distant events. The metaphor is almost suspiciously perfect: Palantir promises institutional vision, but Tolkien’s stones also warned that seeing more information does not necessarily mean understanding the whole truth.

The simplest explanation

Imagine an organization has:

financial records in one system
vehicle locations in another
maintenance histories in spreadsheets
satellite images somewhere else
personnel records behind separate security permissions
forecasts generated by several AI models
hundreds of people trying to coordinate decisions

Palantir connects those systems without necessarily replacing them. It creates a common digital representation of the organization, applies permissions and security rules, and gives people tools for analysis, planning, workflow management, simulation, and AI-assisted action.

It is not merely a dashboard. The ambition is much larger: to become the operating layer between an institution’s data and its real-world decisions.

Palantir began in 2003 by developing software for the United States intelligence community and counterterrorism work. It subsequently moved into commercial enterprises facing similar problems involving fragmented data, complex permissions, and high-stakes decisions. Alex Karp remains CEO, Stephen Cohen is president, and Peter Thiel is chairman.

At the end of 2025, Palantir reported 954 customers and approximately $4.5 billion in annual revenue. Government customers provided 54% of revenue and commercial customers 46%; 74% of revenue came from the United States. Those figures reveal the hybrid creature beneath the hood: half defense-and-government contractor, half enterprise software company.

What Palantir actually sells

1. Gotham: intelligence and operational command

Gotham was Palantir’s original flagship platform. It is primarily associated with defense, intelligence, national security, law enforcement, disaster response, and other government operations.

Gotham can combine information such as:

intelligence reports
maps and satellite imagery
sensor feeds
communications data
personnel and equipment locations
known relationships between people or organizations
previous incidents and operational plans

Users can search relationships, track events, examine patterns, coordinate missions, and maintain a shared operational picture. Palantir describes Gotham as extending from command centers to the “tactical edge,” meaning deployed personnel, vehicles, ships, aircraft, and field locations.

A crude analogy would be that Gotham gives an intelligence or military organization something resembling a live, permission-controlled map of everything it knows.

That is extraordinarily useful. It is also where many of the ethical thunderstorms gather.

2. Foundry: the operating system for an enterprise

Foundry is the commercial and institutional data platform. Manufacturers, hospitals, energy companies, banks, pharmaceutical researchers, utilities, and supply-chain operators use it to connect data and turn it into working applications.

Suppose an automobile manufacturer wants to understand why a production line is slowing down. Foundry could connect:

component inventories
supplier deliveries
machine sensors
staffing schedules
quality-control failures
customer orders
production forecasts

Instead of five departments arguing from five spreadsheets, they can work from a shared model and test possible actions.

Foundry includes data integration, pipelines, analytics, application development, workflow tools, APIs, software-development kits, access controls, and audit capabilities. Palantir says every one of its commercial customers now uses Foundry.

3. AIP: artificial intelligence connected to real operations

AIP, the Artificial Intelligence Platform, connects large language models and other AI systems to an organization’s data and workflows.

This is different from giving employees a corporate chatbot.

AIP can allow an organization to build:

AI agents
automated workflows
document-analysis systems
forecasting applications
maintenance assistants
logistics planners
decision-support tools
natural-language interfaces to internal data

The important part is governance. AIP is designed to place AI inside existing permissions, audit trails, security classifications, and human-approval processes. An AI agent may be allowed to recommend a change, for example, but prohibited from executing it until an authorized human approves it.

Palantir does not insist on a single AI model. AIP can connect to commercial, open-source, or privately hosted models. Palantir says customer prompts and responses sent to approved third-party model services are not retained or used to train those models.

4. Apollo: software deployment everywhere

Apollo manages and updates software across unusually complicated environments:

public clouds
private clouds
customer-owned data centers
classified networks
ships
vehicles
factories
remote field equipment
disconnected or intermittently connected systems

It continuously distributes new features, configurations, security patches, and software versions while accounting for the requirements of each environment. Palantir originally built Apollo to deploy its own platforms, then began selling those deployment capabilities to customers for their own software.

Apollo is the least glamorous product, but it may be the plumbing holding the cathedral upright.

The secret ingredient: the Ontology

Palantir’s most important concept is its Ontology.

Here, “ontology” means a digital model of the things an organization cares about and the relationships among them.

A hospital ontology might contain:

Patient
Bed
Doctor
Procedure
Medication
Appointment
Operating room

A military ontology might contain:

Unit
Vehicle
Sensor
Target
Supply route
Mission
Weather event

A factory ontology might contain:

Product
Part
Supplier
Machine
Production order
Defect
Shipment

Palantir then connects the raw data to those objects, attaches business logic and permissions, and defines possible actions. “Shipment 417 is delayed” stops being an isolated database entry and becomes something connected to affected customers, production lines, alternate suppliers, financial consequences, and available responses.

The Ontology is therefore less like a filing cabinet and more like a digital nervous system. It connects perception, memory, reasoning, and action.

This is why Palantir insists it is not merely a data-analytics company.

What services does Palantir offer?

Palantir earns money from more than licensing boxes of software.

It offers hosted Palantir Cloud subscriptions, installations inside a customer’s own cloud or hardware environment, continuous operations and maintenance, critical updates, technical support, professional services, training, interface configuration, and ongoing data and Ontology modeling.

Its particularly distinctive service is Forward Deployed Engineering. Palantir sends engineers and deployment specialists directly into customer environments. They may work beside soldiers, factory managers, physicians, logistics planners, or intelligence analysts to understand the problem and configure the platform around actual operations.

This creates an unusually close relationship between vendor and customer. Palantir is selling software, but it often behaves like a combination of:

enterprise software provider
systems integrator
operational consultancy
embedded engineering unit

Palantir also runs short AIP boot camps and pilots in which customer data and real workflows are used to produce working prototypes within days.

That high-touch approach helps explain why the software becomes deeply embedded. Palantir is not dropped onto an organization from a parachute and left blinking in the parking lot. Its engineers help weave it into the institution’s circulatory system.

Palantir’s core values

Palantir has two overlapping sets of values: its stated ethical principles and its broader institutional ideology.

Its formal ethical principles

Palantir identifies four major principles:

Privacy by design. Privacy controls should be built into the architecture rather than added later as decorative legal parsley.

Humans must remain responsible. Decisions affecting freedom, opportunity, or basic rights should not be left entirely to automated systems.

Accountability and oversight. Systems should record how information was accessed and used, enforce permissions, and allow supervisors or auditors to reconstruct decisions.

Technology is not the answer to every problem. Some activities should not be automated, and Palantir says there are situations in which it will decline work.

Its software therefore emphasizes granular permissions, purpose-based access, data lineage, audit logs, human-review checkpoints, and controls over what AI models can see or do.

Its cultural values

The company also promotes a distinct internal culture:

Mission over consumer convenience. Palantir portrays itself as an antidote to Silicon Valley companies primarily building advertising systems, social apps, and consumer distractions.

Outcomes over procedure. Employees are judged by whether something works in the field, not merely whether they followed an internal process.

Embedded problem-solving. Engineers are expected to get close to real operational conditions rather than designing everything from a distant office.

Independence and intellectual contrarianism. Its careers materials describe employees as free thinkers, catalysts, and independents who are expected to challenge assumptions and take ownership.

Support for Western power and institutions. Palantir explicitly says that it was founded to support Western democracies and open societies and currently describes its employees as delivering outcomes for “the West’s most important institutions.” This is not politically neutral corporate oatmeal. The company regards technological superiority, national security, and the defense of Western-aligned institutions as part of its mission.

My reading is that Palantir’s deepest value is institutional effectiveness: important organizations should be able to understand reality, decide quickly, and act coherently. Privacy, human judgment, and accountability are presented as engineering constraints on that power, rather than reasons to avoid building powerful systems.

Which industries rely on Palantir?

“No entire industry could function without Palantir” would be too strong. Public information does show, however, that particular agencies, companies, and operational networks have made Palantir central to important workflows.

Defense and intelligence

This is Palantir’s deepest area of institutional reliance.

Its platforms support intelligence analysis, military readiness, logistics, financial management, targeting support, battlefield awareness, and software deployment. In 2025, the U.S. Army announced an enterprise agreement consolidating 75 Palantir-related contracts and allowing Army and other Defense Department organizations to purchase Palantir products under a ceiling of up to $10 billion over the agreement period.

That does not mean Palantir runs the entire military. It does mean the company has moved far beyond being an experimental analytics vendor.

Healthcare and public health

Hospitals and health systems use Palantir for capacity planning, waiting-list management, supply distribution, clinical and administrative coordination, research data, and healthcare-finance workflows.

NHS England’s Federated Data Platform is built on Foundry and is supplied by a Palantir-led consortium. NHS organizations use separate platform instances to combine operational information held in different systems.

Manufacturing, automotive and shipbuilding

Manufacturers use Palantir to connect production planning, equipment health, inventory, quality control, suppliers, staffing, and customer demand.

Palantir has increasingly packaged these capabilities as industrial operating systems for automotive production, aerospace, advanced manufacturing, and the U.S. maritime industrial base. Its own filings identify automotive manufacturing workers, supply-chain managers, shipbuilding, and industrial partnerships as significant areas of use.

Aviation

Airbus developed its Skywise aviation-data ecosystem with Palantir technology. Foundry has been used to combine aircraft, maintenance, engineering, and operational information for Airbus and participating airlines. Airbus described Skywise as connecting airlines, manufacturers, and suppliers around a shared aviation-data platform.

Energy and utilities

Oil and gas companies, renewable-energy producers, utilities, and water organizations use Palantir for equipment monitoring, production optimization, maintenance, demand forecasting, emissions tracking, infrastructure management, and commodity logistics. Palantir’s customer descriptions specifically include utility analysts and oil-and-gas operators.

Finance and insurance

Financial institutions use Foundry for fraud detection, anti-money-laundering investigations, customer analysis, regulatory reporting, risk modeling, compliance, and operational coordination.

Because permissions can be attached to individual data fields and purposes, Palantir is particularly attractive where employees must analyze sensitive information without being given universal access to it.

Pharmaceuticals and life sciences

Researchers use Palantir to combine trial data, laboratory information, patient records, manufacturing data, regulatory materials, and supply chains. The company’s filings name pharmaceutical research as a significant use case, and newer healthcare partnerships connect Foundry and AIP to enormous clinical and claims datasets.

Telecommunications, retail, construction, disaster response, humanitarian assistance, government administration, insurance, and environmental services also appear throughout Palantir’s documented deployments and product offerings.

Why customers become dependent on it

Palantir usually enters through one difficult problem: a supply shortage, military-readiness question, hospital backlog, manufacturing bottleneck, fraud investigation, or maintenance crisis.

Once the platform works, the organization adds more:

More data sources
More departments
More applications
More AI agents
More suppliers and partners
More decisions routed through the Ontology

Eventually, Foundry or Gotham may contain not only the data, but also the institution’s workflows, terminology, permissions, business logic, operational history, and custom applications.

Palantir openly targets large, technically demanding projects with complicated installations and high barriers to competition. It says its products become “sticky” because users can continually build and test new operational tools within the platform.

This produces real value, but also switching risk. Replacing Palantir may mean rebuilding not one dashboard but an entire living model of the organization.

Why Palantir is controversial

Palantir says it is not a data broker and does not sell customer data. Customers generally control their own separate software instances, choose the datasets, assign permissions, and determine permitted uses. Palantir describes itself as a data processor acting under customer instructions.

But that does not end the ethical argument.

A platform can dramatically increase a government’s ability to identify, connect, rank, monitor, investigate, or act against people even when the software vendor does not own the underlying data. The question is not merely who possesses the database, but what new power appears when previously separated databases become operationally unified.

Civil-liberties organizations have criticized Palantir’s involvement in immigration enforcement, policing, military operations, and large public-health systems. Amnesty International and the ACLU have argued that some government deployments create risks involving mass surveillance, political expression, migration enforcement, discrimination, and due process. Palantir disputes characterizations of itself as a surveillance company and points to its access controls, auditability, privacy engineering, human-rights policy, and customer-specific data environments.

Both statements can contain truth:

Palantir builds unusually sophisticated privacy and governance controls.
Its software can still make state, military, and corporate power much more effective.

A seat belt does not determine where the vehicle is driven.

The bottom line

Palantir builds decision infrastructure.

Its central promise is:

Give an institution a coherent model of its world, connect that model to its people and AI systems, and enable the institution to act rapidly without losing security, permissions, or accountability.

At its best, that can mean hospitals finding beds, factories preventing failures, aid organizations directing supplies, and analysts discovering connections hidden inside bureaucratic rubble.

At its most troubling, it can mean creating extraordinarily efficient machinery for surveillance, targeting, exclusion, or coercion.

That duality is not an accidental side effect of Palantir. It is the central drama of the company. Palantir manufactures institutional sight, and sight becomes power the moment someone decides where to point the stone.

Item added to your cart

What is Palantir