On August 28, DEEPEXI TECH (01384) unveiled its interim report for the 2026 fiscal year, showcasing robust expansion. The company posted total revenue of RMB 284 million for the first half, a 115% year-on-year increase. Revenue from its core AI segment—the DeepexiOS AI-grade enterprise operating system platform solution—soared 209.2% to RMB 226 million, lifting the AI business's share of total revenue from 55.3% to 79.6%.
Beyond the impressive top-line growth, profit quality improved markedly. Gross profit climbed 120.5% to RMB 160 million, with gross margin expanding by 1.5 percentage points to 56.5%. The adjusted net loss narrowed by 48.5% year-on-year, and the company achieved profitability in the second quarter alone, posting a net profit of approximately RMB 30.1 million. Operating cash flow also turned positive in the April-to-June period, signaling tangible enhancements in both earnings capability and cash flow health, underpinning a substantive upgrade in revenue quality.
Notably, the durability of this growth is validated by the order pipeline. As of the reporting period's end, cumulative unfinished orders stood at roughly RMB 289 million, providing a solid buffer for sustaining high revenue momentum into the second half.
Breaking down the customer base by downstream industry, DEEPEXI TECH drew RMB 168 million in revenue from industrial clients, accounting for 59.1% of total sales. Within this, the manufacturing vertical contributed RMB 155 million, or 54.5% of overall revenue. The industrial sector has now cemented its position as the company's largest income source. This deep-rooted commitment to manufacturing has earned DEEPEXI TECH recognition from IDC for three consecutive years in enterprise-grade AI industrial applications, with flagship projects at partners like Xingsen Technology, Han's Zhicong, and the Shanghai Merchant Ship Design and Research Institute each securing IDC-related accolades.
During the first half of 2026, the company intensified investments in computing power and R&D for its enterprise agent platform, proprietary enterprise large models, and domain-specific datasets and skills. These efforts have yielded a stable, high-caliber enterprise AI workforce infrastructure and a token productivity platform across manufacturing, retail, scientific research, and public utilities. As a result, the DeepexiOS AI-grade enterprise operating system platform is steadily gaining market traction and broader adoption.
During WAIC 2026, DEEPEXI TECH formally introduced the upgraded DeepWorks enterprise agent platform. The platform's capabilities continue to be refined, integrating mainstream domestic large models such as Qwen, GLM, DeepSeek, and Kimi. It also bolsters the enterprise knowledge foundation by unifying, organizing, and linking information scattered across documents, business systems, and data platforms, delivering accurate, traceable, and continuously updated knowledge support to agents. Additionally, DeepWorks connects to clients' proprietary data platforms through FastData, enabling users to invoke end-to-end capabilities—including data source management, asset architecture construction, metadata governance, data modeling, data collection, pipeline development, SQL execution, and task operations—via natural language commands. Leveraging an independent, isolated execution mechanism and a secure "trial—preview—confirm" loop, the platform ensures data operations remain controlled, auditable, and traceable.
Industry observers suggest that as AI integration with the real economy deepens, enterprise-grade AI is transitioning from technical validation to a phase of creating value at scale. Supported by a quarterly profit inflection, a substantial order backlog, and sustained traction among leading manufacturing clients, DEEPEXI TECH appears well-positioned to accelerate toward a stage of large-scale profitability.
Comments