DeepSeek and Huawei Target a Key Source of Nvidia’s A.I. Dominance
Chinese artificial intelligence startup DeepSeek and Huawei Technologies are strengthening their technical partnership to bypass American export restrictions and challenge Nvidia. Research from SemiAnalysis indicates that DeepSeek could rely primarily on Huawei technology stacks by 2027 or 2028. Meanwhile, the newly previewed DeepSeek V4 model features 1.6T parameters, a 1M context window, and pricing that undercuts closed frontier models by up to 7x. This architecture runs on Huawei Ascend chips, pairing Nvidia-tested software with Huawei hardware while full-system performance and stability remain unproven.
Listen to Live Briefing
Real-time synthesized voice briefing · Live Feeds Desk
- ✓ DeepSeek may primarily rely on Huawei's mature technology stack by 2027 or 2028, according to SemiAnalysis.
- ✓ DeepSeek V4 features 1.6T parameters and a 1M context window.
- ✓ DeepSeek V4 pricing undercuts closed frontier models by up to 7x.
What changed
The release of the DeepSeek V4 preview running on Huawei silicon demonstrates that models with 1.6T parameters and a 1M context window can migrate to domestic chips.
Live updates
-
DeepSeek and Huawei Expand Hardware Alliance
Chinese artificial intelligence startup DeepSeek and Huawei Technologies are strengthening their technical partnership to bypass American export restrictions and challenge Nvidia. Research from SemiAnalysis indicates that DeepSeek could rely primarily on Huawei technology stacks by 2027 or 2028. Meanwhile, the newly previewed DeepSeek V4 model features 1.6T parameters, a 1M context window, and pricing that undercuts closed frontier models by up to 7x. This architecture runs on Huawei Ascend chips, pairing Nvidia-tested software with Huawei hardware while full-system performance and stability remain unproven.
Why it matters
Domestic innovation and trade barriers are reshaping the artificial intelligence hardware sector in China as local firms work to close the ecosystem gap with Nvidia. Huawei has steadily narrowed Nvidia's lead in the domestic chip market by optimizing software kernels and engineering efforts. DeepSeek is integrating its systems directly with Huawei Ascend hardware, marking a significant transition in the production AI stack.
What is confirmed
- DeepSeek may primarily rely on Huawei's mature technology stack by 2027 or 2028, according to SemiAnalysis.
- DeepSeek V4 features 1.6T parameters and a 1M context window.
- DeepSeek V4 pricing undercuts closed frontier models by up to 7x.
Still unconfirmed
- Full-system performance and stability for the migration pairing Nvidia-tested model software with Huawei hardware remain unproven.
What to watch next
- Evidence regarding the full-system performance and stability of DeepSeek V4 running on Huawei hardware
- Further deployment data on DeepSeek relying primarily on Huawei hardware by 2027 or 2028
confidence 90%Sources used for this update (6)
- www.electronicsforyou.biz — Huawei’s Coordination with DeepSeek Challenging Nvidia
- weijinresearch.substack.com — DeepSeek’s Elegant Migration Toward Training on Ascend
- startupfortune.com — DeepSeek's cheap new model is making GPU scarcity worse, not ...
- www.dplooy.com — DeepSeek V4: 1.6T Params, 1M Context, Huawei Silicon
- the-ctr.net — Huawei Narrows Nvidia's Lead in China's AI Chip Market Amid ...
- ai.miraheze.org — DeepInfer - Learn AI
-
Huawei and DeepSeek Challenge Nvidia AI Chip Dominance
Chinese artificial intelligence startup DeepSeek and Huawei Technologies are deepening their partnership to narrow the technology ecosystem gap with Nvidia. According to research from SemiAnalysis, DeepSeek is integrating its systems with Huawei Ascend 950PR chips and is projected to rely primarily on Huawei hardware. The startup released Ascend-ready builds of six kernel repositories on September 30, 2026. And Huawei engineering efforts and software kernels have reached 99.7 per cent of their theoretical efficiency, demonstrating significant progress despite inferior resources.
Why it matters
This partnership represents a concerted effort by major Chinese technology firms to bypass Nvidia hardware and software ecosystems for advanced training operations. The collaboration involves more than standard supplier-customer relations, as DeepSeek actively influences the design of future processors tailored to its models. Observers note that these domestic developments aim to establish independent artificial intelligence infrastructure within China.
What is confirmed
- DeepSeek is working closely with Huawei to influence the design of future artificial intelligence chips.
- Research from SemiAnalysis indicates that closer collaboration between Huawei and DeepSeek is narrowing the chip ecosystem gap with Nvidia.
- DeepSeek released Ascend-ready builds of six kernel repositories on September 30, 2026.
Still unconfirmed
- DeepSeek is projected to rely primarily on Huawei hardware for its future operations.
What to watch next
- Further performance benchmarks of Huawei Ascend 950PR chips in large-scale training operations.
- Additional software repository releases or open-source framework adaptations by DeepSeek.
confidence 90%Sources used for this update (8)
- help.naver.com — Futures Flat As Oil Rebounds After Trump's "Energy Ceasefire" Unravels, Memory Stocks Slide | ZeroHedge
- partofstyle.com — DeepSeek and Huawei’s Chip Alliance Could Challenge Nvidia’s ...
- gmasia.ai — ‘Jensen’s nightmare’: Huawei, DeepSeek narrow AI chip ...
- www.aiaffairs.com — SemiAnalysis says Huawei and DeepSeek are closing Nvidia’s AI ...
- www.theleadsman.uk — The Huawei Line: DeepSeek Open-Sources Its Production ...
- www.kucoin.com — Huawei and DeepSeek Narrow AI Chip Ecosystem Gap with Nvidia
- aiweekly.co — DeepSeek Opens Six Ascend Kernel Repos, Co-Designs With Huawei
- www.silicon.co.uk — DeepSeek, Huawei Narrow AI Performance Gap With Nvidia
-
DeepSeek Shapes Huawei AI Chip Design to Challenge Nvidia
Chinese artificial intelligence startup DeepSeek is collaborating directly with Huawei Technologies to shape the design of future processors tailored to its models. According to reports from Investing.com and The Information, this strategic partnership aims to enable China to build frontier artificial intelligence systems without relying on Nvidia technology. The collaboration extends beyond standard industry discussions. It represents a major step in weakening the dominance of Nvidia and its software ecosystem as domestic firms seek alternatives for advanced training operations.
Why it matters
This cooperation builds on prior efforts by DeepSeek and Huawei to develop open-source programming tools for Ascend processors. DeepSeek previously disrupted the artificial intelligence sector by launching the DeepSeek-R1 model and training its V3 model at a fraction of the cost used by competitors. Meanwhile, US lawmakers continue to pursue chip restrictions in a defense bill, though OpenAI has reportedly stopped pressing for new export controls.
What is confirmed
- DeepSeek is working closely with Huawei Technologies to design artificial intelligence chips tailored to its models.
- The strategic push could help China train advanced systems without relying on Nvidia technology.
Still unconfirmed
- OpenAI has reportedly stopped pressing for new artificial intelligence-chip export controls on China.
What to watch next
- Further legislative actions regarding chip restrictions in the defense bill.
- Commercial releases or performance metrics of the jointly designed Huawei processors.
confidence 100%Sources used for this update (3)
- digg.com — OpenAI reportedly stops pressing for new AI-chip export controls on China · Digg
- www.theinformation.com — How DeepSeek Is Shaping Huawei’s AI Chips in a Growing ...
- finance.yahoo.com — DeepSeek teams up with Huawei to challenge Nvidia
-
DeepSeek and Huawei Partner to Challenge Nvidia Dominance
Chinese artificial intelligence company DeepSeek is collaborating with Huawei to develop open-source programming tools for Ascend processors, aiming to weaken Nvidia's CUDA ecosystem dominance. Based in Hangzhou, Zhejiang, DeepSeek is funded by High-Flyer and was founded in July 2023 by Liang Wenfeng. The company launched the DeepSeek-R1 model in January 2025, which provided responses comparable to OpenAI's GPT-4 and o1. DeepSeek claims it trained its V3 model for US$6 million using approximately one-tenth the computing power of Meta's Llama 3.1, upending the artificial intelligence sector with significantly lower training costs.
Why it matters
This partnership represents a calculated wager that advanced software can render alternative hardware practical ahead of physical chip parity. The collaboration directly targets Nvidia's CUDA software monopoly, which serves as a cornerstone of the company's global artificial intelligence market dominance. Analysts view the joint effort as a core component of China's broader push for compute sovereignty amid tightening international trade controls.
What is confirmed
- DeepSeek was founded in July 2023 by Liang Wenfeng, who serves as the CEO for both DeepSeek and High-Flyer.
- DeepSeek launched its eponymous chatbot alongside the DeepSeek-R1 model in January 2025.
- DeepSeek claimed it trained its V3 model for US$6 million using about one-tenth the computing power consumed by Meta's Llama 3.1.
- DeepSeek released six new tools, including an Ascend-based AI Toolbox and TileLang, on September 30, 2026.
Still unconfirmed
- DeepSeek's work with Huawei can make a second hardware platform practical before catching up on hardware itself.
What to watch next
- Adoption rates of DeepSeek's new Ascend programming tools among enterprise developers.
- Further releases of open-source software by DeepSeek and Huawei targeting Nvidia's CUDA ecosystem.
confidence 95%Sources used for this update (7)
- en.wikipedia.org — DeepSeek - Wikipedia
- openrouter.ai — DeepSeek V4 Pro 0813 - API Pricing & Benchmarks | OpenRouter
- build.nvidia.com — AI Models by DeepSeek AI | Try NVIDIA NIM APIs
- www.thexpin.com — Will DeepSeek and Huawei End Nvidia's Lead? - thexpin.com
- economictimes.indiatimes.com — DeepSeek, Huawei and Nvidia: China's AI chip strategy ...
- www.washingtonexaminer.com — China is chipping away at America's AI advantage
- article.wn.com — DeepSeek, Huawei and Nvidia: China's AI chip strategy ...
-
DeepSeek and Huawei Challenge Nvidia's AI Dominance
Chinese AI startup DeepSeek partners with Huawei to develop open-source programming tools for Ascend processors, aiming to weaken Nvidia's CUDA ecosystem dominance. DeepSeek released six new tools, including an Ascend-based AI Toolbox and TileLang, on September 30, 2026. This move is part of China's push for compute sovereignty and a potential challenge to Nvidia's market share, especially in China.
Why it matters
The partnership between DeepSeek and Huawei is seen as a strategic move to challenge Nvidia's software dominance in the AI sector. Nvidia's CUDA ecosystem has been a significant factor in its success, but China's push for technological self-sufficiency and the U.S. export controls on advanced chips have created an opportunity for local players to gain ground. Developing economies, which are more optimistic about AI, face increased risks and less ability to counter them if China becomes the dominant provider of AI infrastructure.
What is confirmed
- DeepSeek partnered with Huawei to release open-source programming tools for Ascend processors.
- The released tools include an Ascend-based AI Toolbox and TileLang, designed to simplify programming and support a 128-chip Ascend 950 supernode.
- DeepSeek secured at least RMB 80 billion in funding and approaches an initial public offering at a valuation of 500 billion yuan.
- The U.S. must set dominance of the global AI sector as a strategic economic and national security priority.
Still unconfirmed
- DeepSeek's models are open-weight, meaning the exact parameters are openly shared, but the training data is not openly licensed.
- DeepSeek is considered more closely linked to China's defense industry due to previous affiliations of its researchers.
What to watch next
- GitHub star count for DeepSeek's new repositories
- Adoption rate of TileLang and Ascend-based AI Toolbox
- U.S. response to China's AI push
confidence 80%Sources used for this update (21)
- www.thecipherbrief.com — Unique Risks of AI Adoption in Developing Economies – The Cipher Brief
- www.businesskorea.co.kr — DeepSeek, Huawei Join Forces to Challenge Nvidia’s CUDA ...
- cryptobriefing.com — DeepSeek and Huawei take aim at Nvidia’s software moat
- www.ainchina.com — DeepSeek Open-Sources Huawei Ascend AI Toolchain: The CUDA ...
- thearabianpost.com — DeepSeek broadens Huawei software challenge to Nvidia
- tech.yahoo.com — Huawei and DeepSeek release open source AI tools in a bid to ...
- relvehq.com — DeepSeek, Huawei target Nvidia software moat - relvehq.com
- www.businesstimes.com.sg — DeepSeek and Huawei take aim at Nvidia’s formidable moat
- www.thedailystar.net — DeepSeek open-sources tools for Huawei's Ascend chips
- www.technotime.net — DeepSeek and Huawei Develop Open-Source Tools to Challenge ...
- aichina.news — China’s DeepSeek open-sources tools to help Huawei chips ...
- www.thedailyupside.com — Quantum is Poised for a Commercial Leap, Investors Have Noticed - The Daily Upside
-
DeepSeek and Huawei Target Nvidia CUDA With Open-Source Tools
Chinese artificial intelligence startup DeepSeek has partnered with Huawei to develop open-source programming tools for Ascend processors, aiming to weaken the software dominance of Nvidia's CUDA ecosystem. Released on September 30, 2026, the six new tools include an Ascend-based AI Toolbox and TileLang, designed to simplify programming and support a 128-chip Ascend 950 supernode. Simultaneously, DeepSeek secured at least RMB 80 billion in funding and approaches an initial public offering at a valuation of 500 billion yuan, cementing its financial position as it challenges foreign chip architecture.
Why it matters
Nvidia derives substantial market power from its proprietary CUDA software suite, which locks developers into its hardware. By building open-source alternatives tailored for Huawei Ascend chips, domestic Chinese tech firms seek to create a self-sufficient hardware and software pipeline. This push coincides with massive domestic capital injections into domestic artificial intelligence startups.
What is confirmed
- DeepSeek is developing open-source programming tools optimized for Huawei Ascend chips to counter Nvidia's CUDA ecosystem.
- DeepSeek has locked in at least RMB 80 billion in its latest funding round.
Still unconfirmed
- DeepSeek is targeting an initial public offering early next year at a 500 billion yuan valuation.
- DeepSeek's latest funding round could reach RMB 100 billion as Tencent and CATL double down.
What to watch next
- Execution of DeepSeek and Moonshot AI initial public offerings early next year
- Final total of DeepSeek's current funding round reaching RMB 100 billion
- Adoption rates of the new Ascend-based programming tools among developers
confidence 90%Sources used for this update (3)
- en.sedaily.com — DeepSeek Teams With Huawei to Challenge Nvidia's CUDA
- en.sedaily.com — China's Two AI Leaders Near IPOs at $50B and 500B Yuan Valuations - Seoul Economic Daily
- finance.biggo.com — DeepSeek's Latest Funding Round Could Reach RMB 100 Billion as Tencent and CATL Double Down
-
DeepSeek and Huawei Launch Open-Source Tools for Ascend AI Chips
DeepSeek has partnered with Huawei to develop and open-source programming tools for Huawei Ascend AI processors to reduce dependence on Nvidia. On September 30, 2026, DeepSeek released six AI tools, including an Ascend-based AI Toolbox and TileLang, which the company claims offers a simpler programming model than Nvidia. These tools aim to optimize performance toward the hardware's upper limit and support a 128-chip Ascend 950 supernode, targeting the software advantage currently held by Nvidia's CUDA suite.
Why it matters
Nvidia maintains dominance in the AI sector largely through CUDA, the industry standard software for building neural networks on GPUs. China is pushing for a more independent AI chip ecosystem to bypass reliance on US-based technology. This partnership represents a strategic effort to create a viable software alternative for domestic hardware.
What is confirmed
- DeepSeek partnered with Huawei to develop open-source programming tools for Ascend AI processors.
- DeepSeek released six AI tools for Huawei Ascend chips on September 30, 2026.
- The collaboration includes the development of a 128-chip Ascend 950 supernode.
- Nvidia's CUDA software is the current industry standard for building neural networks on GPU chips.
Still unconfirmed
- DeepSeek's AI Toolbox performance is approaching the hardware's upper limit.
What to watch next
- Adoption rates of these open-source tools by other Chinese AI firms.
confidence 90%Sources used for this update (15)
- www.scmp.com — China technology
- kimkj.com — [AI Library] Trump, the Diplomacy of Insult – Chapter 51: Standing Before the Precipice of 2027
- The New York Times — DeepSeek and Huawei Target a Key Source of Nvidia’s A.I. Dominance
- Bloomberg.com — DeepSeek Unveils Huawei AI Chip Tools That May Replace Nvidia’s
- Reuters — DeepSeek partners with Huawei to develop chip programming tools, reducing reliance on Nvidia
- South China Morning Post — DeepSeek opens tools to help Huawei chips supplant Nvidia in AI
- 富途牛牛 — Benchmarking NVIDIA's CUDA! DeepSeek has open-sourced its Ascend‑based "AI Toolbox," with performance approaching the hardware's upper limit.
- techstrong.ai — DeepSeek, Huawei Build AI Chip Software to Challenge NVIDIA - Techstrong.ai
- www.wionews.com — DeepSeek’s new Huawei partnership could reshape China’s AI ...
- finance.yahoo.com — DeepSeek and Huawei Just Took Aim at Nvidia's Biggest ...
- www.nextmsc.com — Semiconductor Market Update: DeepSeek Huawei AI Tools
- careeraheadonline.com — DeepSeek and Huawei Challenge Nvidia's AI Dominance
Community Sentiment: How do you assess this situation?
Voice your perspective · Real-time aggregated sentiment from the Live Feeds community