Urgent.News

What's breaking now, across thousands of outlets.

Wednesday, 30 September 2026 Reading Room

AI

DeepSeek partners with Huawei to develop chip programming tools, reducing reliance on Nvidia

HONG KONG: Chinese AI firm DeepSeek said on Wednesday it has partnered with Huawei Technologies to develop programming tools for Huawei's Ascend chips, the latest sign of Chinese tech firms deepening ties as they seek alternatives to the Nvidia ecosystem.

nst.com.my ·

DeepSeek partners with Huawei to develop chip programming tools, reducing reliance on Nvidia

HONG KONG - Chinese AI company DeepSeek has announced a partnership with Huawei Technologies to create programming tools for Huawei's Ascend chips, according to a statement on its official WeChat account. The collaboration aims to reduce dependence on Nvidia's ecosystem, as Chinese tech firms strengthen ties. DeepSeek is open-sourcing infrastructure for Huawei's Ascend platform, including a high-level programming language called TileLang and related compute and communication libraries.

The primary goal is to build an independent, self-controlled GPU software ecosystem. DeepSeek emphasized that TileLang was developed to provide a universal, easy-to-program language capable of reaching the hardware's full performance potential. The high-level language offers a simpler programming model than Nvidia's CUDA, which improves development efficiency and simplifies code logic.

Written by urgent.news from New Straits Times's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.

This story

This is one outlet's version. Read the fullest account.

Read the original at nst.com.my →

More in AI

UI Architecture That AI Coding Agents Can Actually Use

When we ask an AI agent to "add a confirmation modal before deleting a user," we expect a focused pull request. What we usually get is an archaeological expedition.

  • AI agents struggle with undocumented state and tracing network calls across files
  • Bounded feature packages provide a tight operational perimeter for agents
  • Defined contracts, type checks, and lint rules enforce boundaries mechanically

·dev.to

Zet: an open-source layer on Laya that knows when to ask a human

Confidence scores aren't error rates. Zet sits on top of Laya and marks every answer sure or unsure, using conformal prediction sets and a Learn-then-Test threshold calibrated per language, so the…

  • Zet is an open-source tool for Laya language model
  • Uses conformal prediction sets for error budgeting
  • Refuses tasks when complexity exceeds 18 scenarios

·dev.to

Six Signatures, Zero Laws: What the White House Super Intelligence Accord Actually Requires

If the entire safety framework for the most powerful technology on the planet were a document, how long would you expect it to be? As of Tuesday, the answer is two pages.

  • White House Accord signed by President Trump and six AI companies on September 29, 2026
  • Accord outlines voluntary commitments for safe AI development without legal enforcement
  • Main components include internal controls, watch team, external auditor, board oversight

·dev.to

From SEO To GEO: How AI Search Really Sees Your Brand

Generative Engine Optimization (GEO) is quietly becoming the missing layer in most growth and dev marketing strategies. We obsess over Google rankings and organic sessions, but barely track what…

  • GEO focuses on optimizing entities for AI-generated answers instead of keywords.
  • AI visibility issues arise when brands aren't mentioned in AI-generated responses.

·dev.to

More from Wednesday 30 September →