Alibaba's AI Chip Push Meets a Coding-Agent Security Test

Tue, Sep 22 ~2 min read ✓ Reviewed by Get AI Decoded Editorial Team

⚠️ Not financial advice. All content is informational only. We may hold positions in securities mentioned. Always do your own research before making investment decisions. Affiliate Disclosure →

Two signals show the AI race expanding from model scale into chips and enterprise security.

🗄️ Alibaba Pushes Deeper Into the AI Stack

Decoded: Alibaba (BABA) plans to train a model with 5 trillion to 10 trillion parameters, Chief Executive Eddie Wu said Tuesday, while unveiling its Zhenwu V900 AI chip. The company said the chip delivers three times the performance of its previous design as it expands across models, silicon and data centers. Reuters, Sep. 22

Why it matters: Building more of the stack could reduce Alibaba's reliance on outside suppliers and tighten the link between its cloud services and hardware roadmap. For investors, execution and developer adoption matter more than parameter count alone, while NVIDIA (NVDA) faces another domestic alternative in China.

🔐 Z.ai Disables Features After Repository Upload Reports

Decoded: Chinese startup Z.ai disabled features in its ZCode assistant after users reported that entire local code repositories were uploaded to overseas cloud servers without consent, Reuters reported Monday. Z.ai said the problem came from a default-enabled codebase-indexing feature and that it had patched the vulnerability. Reuters, Sep. 21

Why it matters: Coding agents can touch proprietary source code, so defaults and data-routing controls are becoming procurement issues, not minor product settings. Vendors that can prove local processing, explicit consent and auditable retention policies should have an edge with enterprise buyers.

Stay decoded. See you tomorrow.

— The Get AI Decoded Team