Intelligent Content Block Recognition
Transform scattered text into semantic knowledge blocks that LLMs can understand — goes far beyond traditional line-by-line text segmentation.
A multimodal data parsing engine — built for the LLM era
Parse documents, tables, and charts into structured semantic data. AI-Ready outputs for LLM training, inference, and AI Agent applications.
EasyDoc — Intelligent Parsing
行业研报.pdf
5/ 35
70%
中原证券 CENTRAL CHINA SECURITIES锂电池
1. 业绩与行情回顾
1.1. 业绩回顾
1.1.1. 板块营收与净利润
根据万得数据统计：2017 年以来，锂电池板块营收总体持续增长，板块增速可分为三个阶段：一是 2017 年-2020 年增速总体回落；二是 2021 年-2022 年的高增长阶段；三是 2023 年以来，板块营收增速持续回落，并首次负增长。2023 年板块营收 2.25 万亿元，为近年来的最高值，较 2022 年同比增长 6.65%。
2024 年前三季度，锂电池板块营收 1.57 万亿元，同比下降 5.21%，板块营收自 2017 年以来首次出现负增长，营收增长中位数为 -6.21%，其中 46 只个股营收同比增长，营收增长个股占比 43.81%，显示板块细分领域标的营收增速出现显著分化，营收同比增长个股自 2023 年以来改善。板块营收同比负增长主要系产业链细分领域产能持续释放，行业需求放缓等多因素叠加导致产业链价格总体持续承压。考虑 2024 年全年新能源汽车出货增速将回落，以及上游原材料价格走势，总体预计 2024 年全年锂电池板块营收小幅下降。
图 1：锂电池板块营收及增速
资料来源：中原证券研究所，Wind
注：板块财务数据为万得锂电池板块（最新为 105 家企业）
图 2：锂电池板块净利润及增速
资料来源：中原证券研究所，Wind
2017 年以来，板块净利润增速总体也可以分为三个阶段：一是 2017-2019 年板块净利润增速持续回落，其中 2019 年板块净利润出现少有的亏损；二是 2020-2022 年净利润恢复为正且增速持续提升，与板块营收呈现高度正相关；三是 2023 年以来，板块净利润增速持续回落。2023 年板块净利润为 1414.1 亿元，同比下降 33.61%，净利润增速显著低于板块营收增速，主要系产业链价格下行时，行业盈利能力下降，同时行业存在一定的减值。
2024 年前三季度，锂电池板块净利润 870.14 亿元，同比下降 26.56%，跌幅持续收窄，但净利润增长中位数为 -37.19%，显示个股净利润差异化显著，其中 38 只个股净利润同比增长，净利润同比增长个股占比 36.19%，净利润增长中位数和净利润增长个股占比大幅低于营收增长中位数和营收增长个股占比。净利润增速大幅回落最主要因素为产业链细分领域产能持续释放叠加需求增速放缓，从而导致产业链价格短期持续大幅回调，以及行业竞争显著加剧。
本报告版权属于中原证券股份有限公司 www.ccnew.com 请阅读最后一页各项声明第 5 页 / 共 35 页
Trusted by leading enterprises
Complexity simplified. Structure revealed.
Hierarchical Structure Analysis
Auto-detect document hierarchy and build a structure tree — giving LLMs a structured context they can actually navigate.
Deep Chart & Image Understanding
Extract deep semantic meaning from complex tables and images. Supercharges any multimodal AI application.
Unlock data intelligence. Unleash AI potential.
Structure-Aware RAG
Hierarchical document structure awareness, intelligent context linking, multi-level knowledge retrieval augmentation
Document Search & Recommendation
Semantic-precise search, cross-modal content linking, context-aware retrieval augmentation
Document Knowledge Graph Construction
Automatic entity-relationship extraction, multi-source knowledge alignment, knowledge reasoning & completion
Document Information Extraction
Auto-extract chart data, cross-page information linking, intelligent key-information summarization
Document Comparison
Multi-document diff detection, semantic-level content comparison, change tracking on key information
Document Machine Translation
Layout-aware translation, intelligent multi-language formatting preservation, chart content translation
30+ document formats. 100+ languages.
PDF
Word
PPT
Excel
TXT
JPG/PNG
Video
Audio
Built for real industries
Financial Services
Contract parsing, financial report extraction, compliance document processing, risk review materials
Healthcare & Life Sciences
Medical literature parsing, clinical data extraction, structured drug labeling
Industry Research
Research report parsing, multi-source information synthesis, cross-document opinion comparison
Education
Academic paper & literature parsing, formula recognition & extraction, textbook content structuring
Manufacturing
Extract key info from production guides and technical drawings, process documentation parsing
Tech & Equipment Management
Technical manual parsing, equipment document management, maintenance record extraction
Why choose EasyDoc
Expert Team Backing
Founded by veterans from Fortune 500 and top-tier internet companies. Years of deep expertise in document parsing, computer vision, and NLP — delivering data intelligence to nearly a hundred enterprises. From design to deployment, we provide hands-on technical support every step of the way.
Enterprise-Grade Deployment Flexibility
Modular API design — call what you need, combine however you want. Built for high-concurrency, elastically scales with your business volume. SaaS out of the box, or deploy on-prem in your own environment.
Technology Leadership
Self-developed, AI-native document parsing algorithms and vision-semantic models. Industry-leading on complex layout recognition, table structure reconstruction, and image-text relationship understanding. Backed by continuous R&D that tracks the latest LLM breakthroughs.
Unified Multimodal Data Processing
Breaks through siloed data types — parse documents, tables, images, charts, formulas, and more through a unified pipeline into structured output. High-quality data foundation for downstream retrieval, training, inference, and AI Agent applications.
Security & Full Control
Supports multiple domestic and international LLM backends — no vendor lock-in. On-prem deployment keeps data within your perimeter. Fully compliant with local data security regulations, ensuring enterprise data sovereignty.