DeepSeek, Huawei Co-Design AI Chips; Ascend 950DT Pre-Training Starts Late 2026

DeepSeek is working directly with Huawei to determine the specifications of future artificial intelligence chips, according to reports from The Information. This collaboration marks a significant shift in how the Chinese AI developer approaches hardware dependencies. Sources familiar with the matter indicated that the two companies are coordinating closely on these engineering efforts. The partnership involves joint work on the design of future AI chips.

In plain English

This story is about a software company deciding what computer parts it needs to build smarter AI systems. DeepSeek is working closely with Huawei to shape the design of future AI chips. This allows DeepSeek to ensure the hardware matches its specific software needs. It also creates a tighter coupling between the AI developer and the chip manufacturer. For system administrators, this signals a move toward specialized hardware ecosystems.

The background

The Information reported that DeepSeek plans to start using Huawei’s Ascend 950DT chips for pre-training large language models. Pre-training is the intensive process of teaching an AI model using vast amounts of data. This phase requires massive computing power and specialized hardware. The report states that DeepSeek aims to begin this specific phase using the Ascend 950DT as soon as the end of 2026. This timeline suggests a phased integration of Huawei’s hardware into DeepSeek’s core operations.

What changes now

The collaboration means DeepSeek is investing in the long-term development of Huawei’s silicon. By working closely with Huawei to shape the design of future AI chips, DeepSeek ensures the chips will support its unique algorithms. This strategy differs from simply purchasing off-the-shelf graphics processing units. It creates a tighter coupling between the AI developer and the chip manufacturer. System administrators should monitor this trend as it may influence hardware procurement and software compatibility standards.

Key terms explained

When companies co-design hardware and software, they optimize performance for specific workloads. Teams usually assess whether custom silicon offers better efficiency than general-purpose processors. Check whether your organization relies on hardware that may become obsolete if vendors shift focus. Understand that pre-training requires sustained high-performance computing, while inference uses less power. Evaluate how hardware dependencies affect your supply chain resilience and long-term maintenance costs.

Background: Supply chain attacks

Supply chain attacks compromise your systems indirectly. An adversary infiltrates a trusted vendor or software update. When you install that legitimate-looking update, the malware executes inside your network. This method bypasses perimeter defenses because the code carries a valid digital signature from a reputable source.

Read the full guide: What Is Supply Chain Attacks: How Indirect Compromise Works

What to do and how to stay safe: DeepSeek

• Review your current AI infrastructure to identify dependencies on specific chip manufacturers.
• Monitor vendor announcements regarding hardware compatibility with major AI frameworks.
• Plan for potential shifts in hardware supply chains by diversifying procurement sources.
• Once the vendor provides an update, test new hardware for performance and stability.

Step-by-step guide: Stop AI Chatbots Leaking Sensitive Business Data

General security guidance from the Vector Update newsroom. It is not confirmed advice from the organisations named in this story.

Frequently asked questions

Who is collaborating on new AI chip designs?

DeepSeek and Huawei are collaborating on the design of future AI chips, according to The Information.

When does DeepSeek plan to use Huawei chips?

DeepSeek plans to start using Huawei's Ascend 950DT chips for pre-training by the end of 2026.

What specific chip is DeepSeek adopting?

DeepSeek intends to use the Ascend 950DT processor for its pre-training workloads.