Analysis: Ten Chinese firms compared to Palantir across four paths

The starting point of this discussion is clear: no Chinese company has fully replicated Palantir. Palantir's product form and pricing model are shaped by the budget cycles of its specific markets. A more useful question is "which part does it resemble?" Some companies come close in ontological modeling, others in their government and enterprise client structure.

Domestic comparable companies can be grouped into four paths:

• Path 1: Independent Control & Government/Enterprise Foundation
• Path 2: Operations & Growth Decision-Making
• Path 3: Data & Knowledge Intelligence
• Path 4: Industrial & Defense Decision-Making

Note: The order of paths and companies below follows the article's structure, not a ranking or similarity score. Each path first explains why these companies are compared to Palantir, then details each company using the same set of fields: comparison entry point, corresponding capabilities, client structure, commercialization validation metrics, and gaps relative to the Chinese market.

Defining Palantir's Four Comparable Elements

Palantir's approach can be broken down into four elements, which serve as the yardstick for evaluating domestic companies.

Element 1: Business Object Ontology Organizing scattered data, rules, and actions into a unified semantic layer, allowing the system to understand relationships between "customers, orders, and equipment" rather than just table structures. A common gap in China is that while data is connected and reports exist, the relationships between business objects and executable actions are not formally defined.

Element 2: Cross-Scenario Reuse The same underlying capabilities are reused across different applications and workflows, so the second scenario doesn't require building from scratch. A common gap in China is that the second scenario requires a new project and new modeling.

Element 3: Delivering Outcomes, Not Tools Charging is not based on software licensing but on solving problems together with the client, with payment tied to results. In China, most deliveries are still software licensing or project acceptance, where the client buys a feature checklist.

Element 4: Client-Side Decision System Accumulation Client-specific business semantics, decision rules, execution records, and feedback remain on the client side long-term, creating switching costs. A common gap in China is that only delivery documents remain on the client side, not sustainable rules and feedback, making replacement costs low.

Path 1: Independent Control & Government/Enterprise Foundation — Solving Access First

Clients in this path are similar to Palantir's government and defense business: qualifications and Xinchuang (domestic IT localization).

China Electronics Shenzhen Sangda A (000032.SZ)

• Comparison Entry Point: Xinchuang
• Corresponding Capabilities: Full-stack Xinchuang foundation and decision intelligence system, with cloud, data, models, and upper-layer applications on the same technology stack.
• Client Structure: Concentrated in defense and military electronics, acting as a cloud infrastructure contractor.
• Commercialization Validation Metrics: Its AI business is still in early stages; public evidence of ontological modeling and cross-scenario reuse is insufficient.

Path 2: Operations & Growth Decision-Making — From Single Business Problems to Multiple Growth Tasks

Companies in this path serve enterprise operations departments. They must answer "what results does this action bring?" but outcome metrics are harder to unify: military and public security have clear success standards, while marketing and growth are influenced by channels, pricing, and competition. Cross-scenario reuse is a key differentiator.

DeepZero Intelligence (02723.HK)

• Comparison Entry Point: Cross-scenario operations decision-making: AI moves from a single business problem to multiple growth tasks for the same client.
• Corresponding Capabilities: Enterprise decision-making AI agents covering consumer insights, new products, channels, advertising, and sales/customer service. Business forms include agent software and agent services.
• Client Structure: Primarily large clients in FMCG, retail, and automotive.
• Commercialization Validation Metrics: DeepZero is still in early stages regarding cross-industry replication breadth, product standardization, and client-side data foundation.

4Paradigm (06682.HK)

• Comparison Entry Point: Enterprise-level AI platform.
• Corresponding Capabilities: Enterprise AI platform and industry solutions, standardizing modeling, features, and deployment into a toolchain, then adding financial, retail, and manufacturing solutions.
• Client Structure: Primarily large enterprises and financial institutions. Procurement decisions involve both technical and business departments, with implementation cycles often spanning years.
• Commercialization Validation Metrics: Platform revenue share and client renewal rates. Renewal rates reflect whether clients are willing to put more scenarios on the same platform.
• Gap with the Chinese Market: The link between platform capabilities and business outcomes is longer and harder to attribute. Clients often see only model metric improvements, not revenue changes.

Path 3: Data & Knowledge Intelligence — Solving Data Understandability First

Companies in this path handle the first half of the four elements: organizing documents, leads, and graphs into machine-understandable formats. They are compared to Palantir because Palantir's ontology layer also starts with knowledge organization, but these companies are still distant from decision-making actions.

Transwarp Technology (688031.SH)

• Comparison Entry Point: Data foundation.
• Corresponding Capabilities: Big data infrastructure software, distributed databases, and data governance.
• Client Structure: Covers finance, government, and energy. Implementation focuses on replacing foreign infrastructure software and building unified data platforms, with buyers being data-heavy institutions.
• Commercialization Validation Metrics: Whether upper-layer AI applications can truly run on Transwarp's data foundation without being replaced by other engines. This ratio determines Transwarp's position stability within the tech stack.
• Gap with the Chinese Market: Transwarp sits below the application layer, with a pricing model quite different from Palantir's. Ontology and decision actions are not within its delivery scope, so client replacement costs for upper-layer applications are low.

Haizhi Technology (02706.HK)

• Comparison Entry Point: Graph-model integrated knowledge organization.
• Corresponding Capabilities: Combining knowledge graphs with large models, using graph structures to represent entity relationships and large models for Q&A and reasoning.
• Client Structure: Primarily government/enterprise and industry clients. Projects often start from specific knowledge scenarios, such as lead sorting and equipment attribution.
• Commercialization Validation Metrics: Publicly available information is limited. Key observables are scenario replication speed and the proportion of platform delivery in revenue.
• Gap with the Chinese Market: There is still a distance between knowledge engineering and decision-making closure. Whether a built graph can drive an executable action depends on the client's own process transformation, which is usually outside Haizhi's delivery scope.

Path 4: Industrial & Defense Decision-Making — Putting Models into Simulation and Production Sites

Clients in this path are close to Palantir's defense business in scenario nature: high cost of decision errors and sensitive data. These companies embed models into specific processes like simulation, scheduling, and production. Clients are concentrated, and revenue depends on a few large projects.

Geovis Technology (688568.SH)

• Comparison Entry Point: Defense and emergency decision-making.
• Corresponding Capabilities: Digital twin and simulation systems for military, emergency, and urban governance.
• Client Structure: Primarily military, emergency, and urban governance, with orders mostly from system-internal units.
• Commercialization Validation Metrics: Structural changes in specialized field revenue vs. emerging business revenue. The former represents the base, while the latter determines Geovis's ability to expand into broader industrial scenarios.
• Gap with the Chinese Market: Business is highly concentrated in the Chinese market, lacking revenue diversification from multinational government clients. Budget cycles in a single market directly impact performance. Upstream data collection and licensing also have cycles.

Huaru Technology (301159.SZ)

• Comparison Entry Point: Military simulation and decision-making.
• Corresponding Capabilities: Simulation software and decision systems, moving from simulation tools to decision systems.
• Client Structure: Primarily military clients.
• Commercialization Validation Metrics: Revenue scale and project count.
• Gap with the Chinese Market: Business scope is narrow and project-based. Cross-scenario reuse is limited by client numbers. There is no public path for moving from military simulation to decision systems in other industries.

Zhongshu Ruizhi (Unlisted)

• Comparison Entry Point: Industrial-level causal intelligence and high-reliability decision-making.
• Corresponding Capabilities: Causal models and dynamic ontology engines, emphasizing applications in power, energy, and industrial control.
• Client Structure: Primarily state-owned enterprises in energy and industry.
• Commercialization Validation Metrics: Limited public information; key observables are project delivery and client retention.
• Gap with the Chinese Market: Business is concentrated in specific industries, with limited cross-industry replication cases.

Collective Gaps Across the Four Elements in China

Looking at all ten companies together, the gap is not in model capability. Domestic large models are already usable for Chinese language understanding and industry Q&A. The real differentiators are threefold:

1. Delivery Reusability: Low — every project feels like the first time.
2. Cross-Industry Replication Capability: Weak — methods validated in one industry require rebuilding the semantic layer for an adjacent industry.
3. Client-Side Data Foundation: Thin — clients internally lack unified definitions for "customer," "order," and "equipment."

These three factors are interdependent; improving any single one alone is unlikely to be effective.

Conclusion: Use "China's Palantir" as a Framework, Not a Title

Measured against the four elements, most domestic companies cover only one or two:

• Government/Enterprise Foundation covers compliance thresholds and delivery, but lacks evidence of ontological modeling and cross-scenario reuse.
• Data & Knowledge Intelligence covers part of ontology but remains distant from decision actions and outcome-based pricing.
• Industrial & Defense Decision-Making covers outcome delivery, but replication radius is limited by client numbers.
• Operations & Growth Decision-Making shows intent for cross-scenario reuse, but product standardization and client-side accumulation are still early.

Companies with all four elements are rare in China.

"China's Palantir" is better used as an analytical framework than a label for any single company. To judge whether a company is on this path, consider two questions:

• Will the client make repeat purchases? Is the second scenario still entrusted to the same company?
• Can delivery be reused? Does the second project require redoing the first?

These two answers are more revealing than any company's own positioning statement.

(Source: Jiemian News)