Xunce Surges 100%: China’s Palantir Aspiration
Hong Kong-listed AI stock Xunce jumps nearly 100% in two weeks — sparking comparisons to Palantir. But how close is it, really?
Since early March, Hong Kong’s AI-themed equities have surged — and the standout performer isn’t Zhipu or MiniMax. It’s Xunce, a previously under-the-radar data infrastructure firm whose shares rocketed from HK$70 to over HK$140 in just days — peaking at HK$153 after a dramatic intraday doubling.
The catalyst? A March 6, 2025 profit warning revealing RMB 1.28 billion in revenue — up 102% YoY. The driver? Large language models (LLMs) accelerating enterprise data demand.
As investor narratives exploded — “Token IPO”, “China’s Palantir” — we dissect what Xunce actually does, where it stands versus Palantir, and why replicating that model remains structurally challenging in China.
🧩 What Does Xunce Actually Do?
At its core, Xunce industrializes data operations — solving two critical bottlenecks:
- Ingest, clean, unify: Ingesting from 1,000+ external sources with millisecond latency, ensuring consistency and full traceability.
- Transform into decisions: Going beyond raw data delivery — applying business logic, semantic understanding, and real-time fusion to output actionable insights.
🔍 Real-World Example: Asset Management
- Problem: Fragmented Excel sheets, siloed systems, manual reconciliation, hours-long delays post-market close.
- Xunce’s workflow:
- News event → automatic scraping + NLP-based sentiment/impact scoring → real-time integration with portfolio metrics → AI-driven trade recommendation (Buy/Sell/Hold) — all within milliseconds to seconds.
This shift — from data provision to decision enablement — mirrors Palantir’s mission. And indeed, Xunce is China’s #4 real-time data infrastructure vendor, holding ~3.4% market share behind Alibaba, Huawei, and Tencent.
⚖️ The $32M Gap: Why Customer Lifetime Value Matters
Financially, Xunce looks deceptively similar:
| Metric | Palantir (2025) | Xunce (2024) |
|---|---|---|
| Revenue CAGR (3Y) | ~45% | ~48% |
| Gross Margin | ~80% | 76.8% |
| R&D as % of Revenue | >100% (historically) | 71.3% |
| Net Income | Profitable (since 2023) | Adjusted net loss: RMB 82M |
But one metric reveals the chasm: Average Revenue Per Customer (ARPC).
- Palantir’s top 20 clients: ~USD 9.4M/year (~RMB 64M), or ~USD 4.7M/client.
- Xunce’s 2024 ARPC: Just RMB 2.72M — over 10× lower.
Why This Gap Is Structural
Palantir’s profitability hinges on its Forward Deployed Engineer (FDE) model:
– Engineers embed on-site, co-developing domain-specific data ontologies and decision logic.
– Initial deployment is capital-intensive — but once the “ontology layer” is built, marginal cost for new use cases drops near zero.
– Result: High upfront investment → scalable, sticky, high-margin expansion.
Xunce is moving toward heavier engagements — shifting from 47% subscription revenue (2022) to 81% transaction/project-based (2024) — and layering in AI to boost value:
- VOne: AI employee platform — natural-language queries → analysis → action.
- DOne: AI data platform — converts plain English into executable data commands.
- Pricing evolution: From per-feature → per-output (e.g., cost savings, efficiency gains).
Yet execution remains unproven at scale.
🌏 Why China May Never Breed a True Palantir
It’s not about tech — it’s about commercial DNA.
1. Pricing Philosophy
- US Market: Pays for outcomes — “Make our supply chain 15% more resilient.”
- Chinese Market: Bids on projects — “Deliver a dashboard with 20 KPIs by Q3.”
→ Drives feature bloat, low margins, minimal operational ownership.
2. Organizational Authority
- Palantir’s power lies in reengineering systems and workflows. That requires cross-departmental mandate — often granted by C-suite or national security mandates (e.g., US DoD).
- In China’s fragmented governance landscape, few enterprises grant vendors authority to reshape internal data sovereignty, permissions, or process logic.
3. Trust & Strategic Preference
- When Chinese enterprises recognize strategic data value, their first move is rarely outsourcing — it’s building in-house (e.g., Ping An Technology, ICBC Smart Finance).
- External vendors face an uphill battle convincing clients to cede control over core decision logic — especially when ROI horizons exceed 18 months.
✅ Conclusion: China can — and does — build Palantir-like technology. But Palantir’s success is rooted in a unique ecosystem: outcome-based pricing, top-down authority, and deep, long-term co-creation. Absent those, “China’s Palantir” remains a compelling narrative — not an inevitable reality.
Article originally published by “Silicon Observer Pro”, author Lin Bai.