Last Updated -

April 19, 2026

MiningLamp

Company Profile and Market Insights

Explore the business model, global strategy, and market performance including insights into its position in China.

Key facts

Founded 2010 • HKEX: 2718 • FY 2025 results (year ended Dec 31, 2025)

RMB 1.43b

Revenue (FY 2025)

RMB (6.41)b

Net loss (FY 2025)

RMB 42.0m

Adjusted net profit (FY 2025, non-HKFRS)

55.4%

Gross margin (FY 2025)

RMB 1.54b

Total cash position (Dec 31, 2025)

RMB 100.2m

Agentic Services revenue (FY 2025)

About

MiningLamp traces its roots to 2006, when founder Minghui Wu launched Miaozhen Systems, and the current Mininglamp Technology Group was formed in 2019. Headquartered in Beijing, the company positions itself as a Chinese provider of enterprise-level data intelligence application software. Its stated mission is “By linking data, we build trusted intelligence to enable enterprises to operate efficiently and accelerate innovation,” and it listed on the Main Board of the Hong Kong Stock Exchange on November 3, 2025 under stock code 2718.

‍

At its core, MiningLamp sells software and AI applications that help enterprises turn complex data into business decisions across marketing, operations, and store execution. The company says its technology stack rests on data intelligence, enterprise knowledge graphs, and data privacy technology, while its product portfolio includes Miaozhen marketing intelligence tools, Jinshuju data tools, operational intelligence software, and the DeepMiner trusted business intelligence agent. MiningLamp also says it serves more than 2,000 leading enterprises globally and holds over 2,000 technical patent certifications plus more than 400 software copyright registrations.

‍

MiningLamp’s profile is more AI-centered today than it was a few years ago. On its homepage, the company says 2025 revenue reached RMB 1.426 billion, gross profit reached RMB 790 million, and adjusted net profit reached RMB 42 million, marking a return to adjusted profitability. It is also pushing deeper into Agentic AI through DeepMiner, which it describes as its proprietary large-model product line for trusted enterprise productivity, and in April 2026 it open-sourced the GUI-VLA model Mano-P 1.0 while showcasing its Agentic AI capabilities at the World Internet Conference Asia-Pacific Summit in Hong Kong. That shows MiningLamp moving from a data intelligence software vendor toward a broader enterprise AI platform.

Business Model and Market Position

MiningLamp runs a hybrid enterprise software and AI services model rather than a pure SaaS or pure consulting model. In 2025, revenue rose to RMB 1.426 billion and gross margin improved to 55.4%. Data Intelligence services remained the core business at 88.4% of revenue, split between marketing intelligence at 50.4% and operational intelligence at 38.0%. Agentic Services contributed 7.0% in their first full year, while the legacy industry solutions business fell to 4.6% of revenue as MiningLamp continued phasing that line out. That mix shows a company shifting away from lower-quality project revenue toward repeatable software, data products, and AI-led delivery.

‍

Data Intelligence is the core engine
MiningLamp’s main business still sits inside Data Intelligence. The company’s own architecture places this unit on top of DeepMiner and uses it to sell independent marketing measurement, cross-platform consumer insight, strategy optimization, offline sales process digitization, and smart store operations. Its core product families include Miaozhen Systems, Jinshuju, private-domain tools based on the Tencent ecosystem, conversational intelligence, and the smart store operating system. MiningLamp says this business has served more than 2,000 brand customers and more than 240,000 enterprise clients, which gives it a broad enterprise base across online marketing and offline operating scenarios.
‍
Agentic Services are the new growth layer
The biggest change in 2025 was the move into Agentic Services. MiningLamp built DeepMiner as its enterprise-grade trusted AI Agent platform and started R&D in early 2025, then launched V1 in September 2025 and iterated rapidly into V2 within the same year. Management describes this business as a shift from process-based output to outcome-driven delivery, with DeepMiner compressing workflows that previously took weeks into hours or even real-time responses. In revenue terms, Agentic Services reached RMB 100.2 million in 2025, including RMB 95.4 million from agentic marketing services. For investors, that matters because MiningLamp is trying to replace labor-linear service delivery with a more scalable AI-enabled model.
‍
The monetization model blends subscriptions, software tools, and performance-linked delivery
MiningLamp’s revenue model is not built on one pricing structure. In Jinshuju, the company still talks about a core subscription system and has started testing a subscription plus transaction model through Xiaojin Merchants. In its Tencent-ecosystem private-domain tools, management said average revenue per user stayed stable in the core subscription base. In Agentic Services, the company said delivery is increasingly centered on customer KPIs and is moving away from traditional manpower-based billing. This mix gives MiningLamp a broader commercial structure than a single-license software vendor, though it also means the company still carries service intensity in part of the model.
‍
Market position is strongest in China’s marketing and operational intelligence niches
MiningLamp’s position looks strongest in selected enterprise AI and data-intelligence segments inside China rather than in broad horizontal cloud software. The company says Miaozhen has consistently ranked first in China’s digital marketing technology and data service market. In Tencent-ecosystem private-domain tools, MiningLamp says it holds a top-three industry position, with over 240,000 enterprise clients, 790,000 enterprise users, and a renewal rate above 60%. At the group level, substantially all revenue in 2025 came from Chinese mainland customers, which underlines that MiningLamp’s market position is still primarily domestic.

‍

MiningLamp’s market position today rests on three things: a large China-focused enterprise data base, long-built workflow knowledge in marketing and operations, and a faster-than-before push into agent-based AI. The key point is that MiningLamp is no longer only a data intelligence vendor. It is trying to become a China-focused enterprise AI platform that links software, data, and outcome-driven services inside one operating stack.

Performance in China

China is effectively the whole operating story for MiningLamp today. The company states in its IPO risk disclosures that substantially all of its revenue is sourced from China, which means domestic enterprise spending, AI adoption, and client budgets drive the business almost entirely. In 2025, group revenue reached RMB 1.426 billion, with Data Intelligence still contributing 88.4% of sales and Agentic Services adding 7.0% in their first full year.

‍

Key strategic drivers in China include:

‍

Leadership in core domestic niches
MiningLamp’s strongest position in China is not broad horizontal cloud software. It is in specific enterprise intelligence categories. In its prospectus, the company cited Frost & Sullivan data showing it was the largest data intelligence application software provider in China by 2024 revenue. In the 2025 results, it also said Miaozhen consistently ranked first in China’s digital marketing technology and data service market.
‍
A large local enterprise base with repeat usage
MiningLamp said its businesses served more than 2,000 brand customers and more than 240,000 enterprise clients. In its Tencent-ecosystem private-domain tools business, it said it held a top-three industry position, with more than 240,000 enterprise clients, 790,000 enterprise users, and a renewal rate above 60%. At the group level, 2025 KA client renewal reached 96.0%, and more than 30% of new KA clients came through Agentic Services. That points to solid customer stickiness inside China even as the product mix shifts toward AI-led delivery.
‍
The China growth focus is shifting from data tools to enterprise AI agents
The biggest domestic change in 2025 was the launch of DeepMiner, MiningLamp’s enterprise-grade trusted AI Agent platform for business data analysis and decision-making. Agentic Services generated RMB 100.2 million in 2025, including RMB 95.4 million from agentic marketing services. In practical terms, MiningLamp is trying to move Chinese clients from software-assisted workflows toward outcome-driven AI services tied to marketing, operations, and store execution.

‍

Competition in China is rising fast. MiningLamp itself said in the 2025 results that AI industry competition is intensifying and client expectations for outcome delivery are rising. Its IPO risk factors also describe the Chinese data intelligence and operational intelligence markets as highly competitive, with rapid technological change and constant new product launches. For investors, the key point is that MiningLamp still has real domestic depth in marketing intelligence and operational software, though its next step in China depends on whether DeepMiner and Agentic Services scale faster than competition in enterprise AI.

Growth and Future Prospects

MiningLamp enters 2026 with a cleaner operating profile than it had a year earlier. In 2025, revenue rose 3.2% to RMB 1.426 billion, gross profit rose 10.8% to RMB 789.6 million, gross margin improved to 55.4%, adjusted operating profit reached RMB 25.0 million, and adjusted net profit turned positive at RMB 42.0 million. The balance sheet also strengthened after the Hong Kong listing, with cash, time deposits, and restricted cash of RMB 1.541 billion at year-end against RMB 256.2 million of interest-bearing bank borrowings and lease liabilities.

‍

Key growth drivers include:

‍

Agentic Services are moving from pilot to scale.
MiningLamp’s clearest growth lever is the expansion of Agentic Services, the business line it formally launched in 2025. That segment generated RMB 100.2 million in its first full year, centered mainly on pay-for-performance agentic marketing services. Management’s own 2026 plan is explicit: scale proven Agentic Services models across a broader range of scenarios and customer groups, deepen content marketing and content entertainment capabilities, and push further into digital white-collar work. For investors, the key shift is that MiningLamp is trying to move from selling tools and labor-heavy project work toward AI-led outcome delivery.
‍
DeepMiner is becoming the core platform behind commercial expansion.
MiningLamp said it invested further in DeepMiner in 2025 and upgraded the platform to Version 2, with stronger task planning, data ingestion, and tool-use capabilities. The company also said DeepMiner helped lift delivery efficiency within marketing intelligence by up to 400% and cut task resolution time in operational intelligence by more than 30%. That matters because stronger workflow automation improves both client ROI and MiningLamp’s own unit economics.
‍
The technology stack is widening beyond one enterprise agent.
MiningLamp’s 2026 technical momentum is not limited to DeepMiner. In April 2026, the company open-sourced Mano-P 1.0 under Apache 2.0 and said the model ranked first on the OSWorld proprietary-model benchmark with a 58.2% task success rate. Around the same time, MiningLamp showcased its Agentic AI work at the 2026 World Internet Conference Asia-Pacific Summit in Hong Kong. These moves raise the company’s visibility in enterprise AI and support its pitch that trusted agents, local deployment, and task execution matter as much as generic chat interfaces.
‍
Data Intelligence still provides the base for future monetization.
MiningLamp is not starting from zero. Data Intelligence still produced 88.4% of 2025 revenue, and the company said its 2025 KA client renewal rate reached 96.0%, with more than 30% of new KA clients acquired through Agentic Services. That gives MiningLamp an installed customer base to upsell from data tools into agentic delivery. It also gives the company a way to feed more workflow data back into model training and product iteration.
‍
Edge deployment and overseas expansion are the next optional layers.
MiningLamp’s 2026 plan includes upgrading Lingting into an edge-side AI device for real-time multimodal data ingestion and working with hardware vendors on compact on-premise models. Management also said it wants to use domestic enterprises’ overseas services and digital content as anchors for a wider overseas push. Those two directions matter because they extend the story beyond software subscriptions and marketing analytics into data-sovereign deployment and cross-border enterprise services.

‍

Challenges ahead include:

‍

Competition in enterprise AI is intensifying fast.
Management said directly in the 2025 results that AI industry competition is intensifying and client expectations for outcome delivery are rising. The report also notes that basic AI tools are becoming more widespread, which lowers the barrier for generic capabilities and puts more weight on proprietary data, know-how, and execution. That raises the pressure on MiningLamp to prove that its vertical data assets and workflow knowledge translate into durable pricing power.
‍
Scaling Agentic Services brings real operating complexity.
MiningLamp also said it recognizes the operational complexity that comes with scaling the Agentic Services model and is increasing investment in technology and delivery capabilities during this period. That shows up in the numbers. Research and development expense rose to RMB 360.6 million in 2025, and selling and marketing expense rose 37.5% to RMB 175.1 million as the company expanded its sales team and brand efforts. The growth case is clear, though the cost discipline test is still in front of the company.
‍
Working-capital quality still needs watching.
MiningLamp’s revenue mix is improving, though receivables remain a point to monitor. Net impairment losses on financial assets and contract assets rose 43.3% in 2025 to RMB 34.9 million, mainly due to aging receivables from industry solutions and higher trade receivables. Trade and bills receivables also rose to RMB 657.7 million at year-end. For investors, that means the shift away from legacy projects and toward more scalable AI-led services needs to show up not only in margins, but also in cash conversion.

‍

MiningLamp’s future now depends less on being a data-intelligence vendor and more on proving that Agentic AI becomes a repeatable business system. The upside rests on scaling DeepMiner, converting its large domestic client base into agentic customers, and extending trusted AI into edge and overseas scenarios. The core debate is whether the company turns early technical momentum and first-year commercialization into a durable operating model with stronger margins and better cash quality.

This Company Profile was written by Dominik Diemer

Dominik Diemer blends an investor mindset with execution discipline.

He is a SAFe Program Consultant (SPC) and Lean Portfolio Management (LPM) practitioner at DMG MORI Digital, working as a SAFe Release Train Engineer and internal consultant in the Lean-Agile Center of Excellence (LACE).

His focus is prioritization, flow, and dependency management that turns strategy into outcomes. With experience across Bertelsmann and the Founders Foundation, he bridges corporate and startup thinking.

He also invests privately in private equity deals, sharpening his view on business models, value drivers, and go-to-market.

StockCounterParts reflects that lens.

Dominik Diemer

Founder of StockCounterparts.com