Jensen Huang urges “give employees computing power”; Chinese tech giants reportedly put compute into KPIs

What happened

It has been reported that NVIDIA (英伟达) CEO Jensen Huang urged companies to “give employees computing power” as a way to accelerate AI innovation. The comment — blunt and practical — has resonated in China’s tech community. Reportedly, several domestic internet giants have responded, saying they have “learned” and are now including computing-power allocation in employee KPIs.

Why it matters

Compute is the bottleneck for modern AI. Short on GPUs? You slow down model training, research cycles and product iteration. That matters in China, where companies are racing to match Western advances while navigating U.S. export controls that restrict shipments of top-tier accelerators. The idea of treating compute as an explicit performance metric is both cultural and operational: it forces managers to budget GPUs and cloud hours, and it signals that access to hardware counts as part of a researcher’s productivity.

The broader context

China’s big platforms have been moving on multiple fronts — buying more servers, building in-house accelerators and tweaking internal processes — to mitigate supply constraints and keep R&D humming. Reportedly including computing quotas in KPIs is part of that adaptive response. But will measurement fix allocation and fairness problems? Tying scarce resources to performance reviews raises questions about hoarding, gaming and whether junior researchers will get meaningful access.

What to watch

Expect more public talk about compute democratisation and internal governance over the next quarters. Geopolitics will remain a wild card: U.S. trade policy and export curbs shape what equipment Chinese firms can buy, and that in turn determines how well Huang’s simple prescription — give people compute — can be implemented in practice.