Jefferies says China software more resilient to AI than US peers
Investing.com -- Jefferies said China's software sector faces less disruption from artificial intelligence compared to US counterparts, though weak IT budgets remain a concern for the industry.
The investment bank said China software stocks fell 13% year-to-date, underperforming US software stocks which declined 6.4% and the CSI 300 index which rose 18%. The China software selloff was driven by earnings downgrades and multiple compression, according to Jefferies.
Since October 2025, US software stocks sold off on concerns that AI would disrupt software-as-a-service businesses. Anthropic's release of Claude Cowork on January 12, 2026 intensified the decline through March 31, with stocks falling about 23% before recovering 21% as disruption fears eased.
Jefferies said China software companies are more resilient to AI disruption due to their project-based business models and module-based pricing structures. Many Chinese software firms serve large state-owned enterprises requiring heavy systems integration and customization. Data security concerns and asset-based preferences have limited public cloud SaaS adoption among these enterprises.
The bank said Chinese SaaS companies rarely use seat-based pricing, instead charging by module. This pricing structure makes China software less vulnerable to AI replacing human operators compared to US peers using per-seat models.
IT budget pressure for large enterprises in a weak macroeconomic environment remains the primary challenge for China software companies, Jefferies said.
The firm named Kingdee as its top pick in China software. Jefferies said Kingdee's ERP system is deeply embedded in client workflows and costly to replace. The company has launched an AI-native ERP suite and Lingee AIOS, guiding 1.0 billion yuan in AI revenue for 2026.
Jefferies said industrial software, financial IT, and ERP are most resilient to AI disruption given deep workflow integration, proprietary data, and strict vendor requirements.
