A spinout from JD.com's (京东) in-house chip team, led by core member Jeffrey Xu, has raised hundreds...

A spinout from JD.com's (京东) in-house chip team, led by core member Jeffrey Xu, has raised hundreds of millions of RMB in angel funding for AI inference chips and full-stack systems. The startup was founded in May 2026 following the restructuring of JD's chip unit, Shanghai Gamma Technology (伽脉科技), with some team members moving to joint venture Zijing Infinite (紫京无限).

Why it matters: This event fits the recurring pattern of large Chinese internet companies incubating and then spinning out internal chip teams, similar to Baidu's Kunlun and Alibaba's T-Head trajectories. It updates the player map for China's AI chip segment, which is seeing a wave of inference-optimized startups emerge as the market shifts from training-centric demand to inference-heavy workloads driven by agent and reasoning-model deployment. The rapid angel raise—reportedly topping RMB 100M (~$14M)—signals strong investor appetite for inference silicon, even as the broader chip startup environment faces capital compression.

Grounded expert take: The spinout reflects structural forces in China's AI semiconductor ecosystem: large tech firms experiment with internal chip projects, then spin them out to reduce balance-sheet risk while retaining strategic alignment. The focus on inference hardware is well-timed as DeepSeek V4 and other open-weight models accelerate the need for cost-efficient inference silicon, and as Chinese AI chip startups increasingly target performance parity with NVIDIA's H100 in inference scenarios. However, the startup faces a fragmented market where customers remain risk-averse and often default to big-name configurations, as highlighted in the same article's reporting on enterprise AI hardware procurement.

#JDChip #AIChipStartup #InferenceSilicon #ChinaSemiconductors #AISpinout #AngelFunding