ABLE DIGITAL (Shanghai Able Digital Science&Tech Co., Ltd.) reported unaudited first-half 2026 revenue of RMB 327.94 million, up 19.1% year on year, driven by continued demand for its AI-driven education technology portfolio. Gross profit advanced 28.0% to RMB 165.36 million, lifting the gross margin by 3.5 ppts to 50.4%. Net loss deepened 17.6% to RMB 116.41 million as the company accelerated investment in research, product commercialisation and sales capacity.
Nearly 90% of revenue originated from the AI Knowledge Asset Development segment, which expanded 17.0% to RMB 294.15 million on stronger sales of knowledge graphs and digital resources. The newer AI Agents & Virtual-Physical Integrated Scenarios line grew 40.6% to RMB 33.79 million, reflecting early traction for the “Polymas” agent suite and immersive space solutions. This mix shift, coupled with efficiency gains, underpinned the margin improvement.
Operating costs rose in line with expansion plans. Cost of sales increased 11.2% to RMB 162.58 million, while distribution and selling expenses climbed 19.9% to RMB 137.87 million amid sales-force growth. Research & development expenditure jumped 26.1% to RMB 102.49 million, accounting for 31.3% of revenue, as the firm pushed its full-modality large model “Polymas” to Version V5 and expanded agent capabilities. General and administrative expenses fell 29.4% to RMB 28.58 million after one-off listing costs in 2025.
Free cash flow remained negative due to seasonally back-ended customer collections and higher working-capital needs. Cash and equivalents plus short-term investments declined to RMB 410.50 million from RMB 618.10 million at end-2025, while bank borrowings rose to RMB 288.29 million, lifting the gearing ratio to 41.6%. Management expects receivable inflows in the second half to alleviate liquidity pressure.
Strategically, ABLE DIGITAL advanced its core technology stack: Polymas V5 now integrates a self-developed Vision Language Model and Virtual Object Model, supporting 800 billion monthly token usage across Chinese universities. The company was named co-developer in multiple Ministry of Education “intelligent agent” and “101 Program” projects, signalling deeper participation in national AI standards for higher education.
A multidisciplinary product suite has been rolled out with C9 League universities, and strategic cooperation agreements were signed with Alibaba Cloud, ByteDance’s Volcano Engine and Baidu AI Cloud to blend general-purpose LLMs with ABLE DIGITAL’s vertical data and delivery network. Internationally, the firm launched AI education initiatives in Kyrgyzstan and Indonesia under Belt-and-Road frameworks, while extending its knowledge-graph and multimodal AI solutions into healthcare, research and industrial sectors.
Total order backlog reached RMB 505.38 million, up 34.7% year on year, and the customer base expanded to 1,266 institutions. High-value customers—those generating over RMB 1 million annually—grew 26.7% and now account for 33.3% of revenue, up from 29.0% a year earlier.
Looking ahead, management plans sustained capital expenditure on computing clusters and agent engineering, deeper collaboration with cloud-AI majors, and wider international deployments. The board declared no interim dividend. Post-period, the company approved a plan to convert 4.71 million domestic shares into H-shares under an impending “full circulation” scheme, pending regulatory filing.