Nvidia’s H200 finally reaches Chinese buyers, though maybe not mainland China
THE SO WHAT
High-end accelerators leaking into Hong Kong and Chinese tech groups after eight months of licensing friction shows export controls are a speed bump, not a wall. If you’re betting on a durable compute moat against Chinese competitors, assume a narrower time window and plan on differentiation above the chip layer.
READ THE SOURCE
MORE FROM THE WIRE
Applied AIWhat comes after the software-defined vehicle?
The shift from software-defined to AI-defined vehicles moves the value center from ECUs and OTA stacks to model quality, data advantage, and real-time context integration. Auto and mobility operators should be treating driving data pipelines and model lifecycle management as primary assets, not side projects inside the infotainment team.
Applied AIHow to address the “white-collar exodus”
If AI anxiety is driving knowledge workers out, the constraint on automation isn’t just capability but trust and narrative. Leaders rolling out AI tooling need an explicit workforce strategy this quarter—clear role definitions, reskilling paths, and communication—or they’ll lose the very operators needed to supervise higher automation levels.
Using synthetic data for AI training is 'a big mistake,' says AI pioneer Rich Sutton
Sutton’s warning on synthetic training data is a shot across the bow for teams leaning on self-generated corpora to escape data scarcity. If your roadmap assumes heavy synthetic data use, you need a parallel track validating whether performance gains are durable or just overfitting to your own model’s biases.
Applied AIZ.ai prices GLM-5.3 API access at $1.40 per million input tokens and $4.40 per million output tokens, unchanged from GLM-5.2
Holding GLM-5.3 pricing flat at $1.40 / $4.40 per million tokens despite added cyber capabilities is a clear shot in the frontier-model price war. For teams building on APIs, budget less around list price changes and more around switching and multi-vendor orchestration costs.