DeepSeek and Huawei Reportedly Team Up on AI Chips to Rival Nvidia

By Priya Raman

DeepSeek and Huawei are working together on new AI chip designs, according to a report by The Information carried by TradingView. The reported goal is direct: challenge Nvidia's position in AI accelerators. The source article supplies no chip specifications, no timeline and no production volumes. Every claim below rests on that single report.

What is actually claimed?

The report makes one core assertion. Two Chinese technology firms, DeepSeek and Huawei, are collaborating on new designs for AI chips. The stated target of the collaboration is Nvidia, the dominant supplier of graphics processing units (GPUs, the chips that train and run large AI models) for data centers worldwide.

What the report does not say matters just as much:

• No named chip products or architecture details.
• No fabrication node, foundry partner or packaging approach.
• No dates for tape-out (the step where a chip design is finalized for manufacturing) or volume shipment.
• No financial terms, investment amounts or revenue projections.

The absence of specifics is routine for early-stage chip projects. It also means the claim cannot be independently stress-tested yet.

Why would these two companies pair up?

The logic, as the report frames it, is complementary capability. DeepSeek builds large language models and has demonstrated an ability to train them under severe compute constraints. Huawei designs its own AI accelerators, such as its Ascend line, and sits at the center of China's domestic semiconductor push. A design collaboration would let a leading model developer tune silicon to its own workloads.

That is inference, not reporting. The Information's article, as surfaced through the TradingView syndication, does not break down the division of labor between the two firms.

How large is the gap they are trying to close?

Nvidia holds the overwhelming share of the market for AI training chips, and its hardware anchors most frontier-model training runs outside China. US export controls have restricted Nvidia's most advanced accelerators from Chinese buyers since 2022, with the rules tightened repeatedly since. Those restrictions created the opening that Huawei's Ascent line and other domestic designs now occupy.

A DeepSeek-Huawei pairing would extend that substitution effort from hardware alone into co-designed hardware plus software. Chips tuned to one customer's models can squeeze out efficiency that general-purpose GPUs leave on the table, roughly the way a bespoke engine outruns an off-the-shelf one in a single race.

What should readers watch next?

Confirmation will come, or fail to come, through verifiable markers rather than statements of intent. Indicators to track:

• Huawei disclosures on next-generation Ascent parts and any DeepSeek reference in official materials.
• Procurement filings or supply-chain signals showing DeepSeek shifting training workloads off Nvidia hardware.
• US Commerce Department responses, since deeper Chinese AI-chip integration invites further export-control action.

The caveat

One outlet has reported this. No company has confirmed the collaboration publicly, and neither DeepSeek nor Huawei has released design documents, partnership terms or schedules. Chip development from design to volume production typically spans years, not quarters, even for established teams. Treat the Nvidia challenge as a stated ambition, not a shipping product.