In the first half of 2026, a period when AI application industries saw healthy overall momentum, Yonyou Network Technology Co.,Ltd. delivered results that ran exactly counter to the sector's direction. The company generated ¥3.727 billion in revenue, up just 4.08% year-on-year, ranking second from the bottom among ten AI application-focused listed firms in expansion pace, while net losses attributable to shareholders doubled down to ¥928 million, marking the deepest deficit in the same peer group. Its long-term archrival Kingdee International Software Group Co., Ltd. posted ¥3.625 billion in first-half revenue, growing roughly 13.6% year-on-year, and swung to a net profit of ¥54.497 million attributable to shareholders, crossing the breakeven threshold. Both players compete at a comparable revenue scale, yet only one has established sustainable profitability, while the other remains mired in persistent losses — a gap that lays bare the fundamental reality of Yonyou Network's ongoing transformation.

Losses for Yonyou Network are anything but new. Since logging its first annual deficit in 2023, the software giant has posted three consecutive loss-making years, accumulating total losses exceeding ¥5.3 billion over three and a half years, with no clear indication of a turning point. The first half of 2026 brought another ¥928 million in losses, continuing a trend that saw ¥933 million lost in 2023, a sharply widened ¥2.07 billion in 2024 — its worst annual result since listing — and ¥1.351 billion in 2025. Compounded, the cumulative shortfall is massive, and the loss trajectory has yet to show signs of substantive convergence.

Of greater concern is that the AI segment, positioned as the engine of future growth, has not yet closed the loop from contract signing to revenue recognition. In the first half, AI-related contract signings amounted to ¥906 million, but AI-related revenue booked was only ¥464 million, accounting for 12.45% of total revenue for the period. The gap between contract value and confirmed income exposes the central challenge in monetizing AI offerings. While traditional ERP deployments typically complete in 3 to 6 months, enterprise-grade AI Agent implementations involve data cleansing, governance restructuring, model tuning, permission design, and multi-round validation, pushing full delivery timelines to 6 to 12 months, with complex scenarios exceeding 18 months. Key clients in this segment include China First Heavy Industries, China Nonferrous Metal Mining, Ansteel Group, and XCMG Group — large central state-owned enterprises with highly customized, low-reusability requirements. This highly tailored, long-cycle, delivery-intensive model forces Yonyou Network to front-load significant labor and resource costs before any revenue can be recognized, while high delivery costs further compress margin potential. With no standardized, lightweight, or scalable AI product suite yet in place, AI revenue is unlikely to shoulder meaningful growth responsibilities in the near term.

The workforce reduction pattern adds another layer of financial strain. Employee numbers peaked at 24,949 at the end of 2023, then entered a sustained contraction phase. By June 30, 2026, headcount had dropped to 17,498, a cumulative reduction of 7,451 staff members over two and a half years — a nearly 30% shrink in scale. This process has accelerated over time: 3,666 people were streamlined in 2024, another 2,228 in 2025, and an additional 1,557 in just the first half of 2026. The company frames this restructuring as part of "optimizing personnel structure and controlling headcount based on AI efficiency gains, while attracting top AI talent," a description echoed in its Hong Kong listing prospectus as "structure optimization under AI and cloud transformation" — trimming traditional labor-intensive delivery roles in favor of senior AI R&D hires.

Strategically, the rationale holds some logic: conventional ERP on-site implementation and customized delivery are genuinely headcount-heavy, and AI tool sets plus platform-based products could theoretically substitute portions of delivery work. But the financials tell a different side of the story. Severance costs for the first half of 2026 alone reached approximately ¥170 million, exceeding the entire dismissal benefit total for all of 2025. The immediate cash outflows triggered by large-scale layoffs sit in stark contrast to management expenses, which have not declined correspondingly. The cost-saving effects of workforce reductions will only materialize once headcount stabilizes, and until then, each round of personnel optimization depletes liquidity through real cash expenditures. Meanwhile, the company must simultaneously support delivery and maintenance for existing ERP customers while channeling resources into AI development, subjecting both human and financial resources to dual-sided strain.

Liquidity signals are even more direct. By the end of June, cash and cash equivalents had plunged 45.16% from ¥3.956 billion at the start of the year to ¥2.170 billion. Meanwhile, short-term borrowings climbed to ¥4.202 billion, with an additional ¥850 million in non-current liabilities due within one year, bringing total short-term interest-bearing debt to approximately ¥5.052 billion — meaning cash covers less than half of near-term obligations. The current ratio also deteriorated from 0.76 at the beginning of the year to 0.64 by mid-year, underscoring mounting short-term repayment pressure. With internal cash generation insufficient and external financing becoming urgent, Yonyou Network's third filing for a Hong Kong listing appears less a strategic choice and more a race against time to defend its liquidity position.