{"id":68757,"date":"2025-08-14T10:19:05","date_gmt":"2025-08-14T10:19:05","guid":{"rendered":"https:\/\/www.newsbeep.com\/ca\/68757\/"},"modified":"2025-08-14T10:19:05","modified_gmt":"2025-08-14T10:19:05","slug":"deepseeks-next-ai-model-delayed-by-attempt-to-use-chinese-chips","status":"publish","type":"post","link":"https:\/\/www.newsbeep.com\/ca\/68757\/","title":{"rendered":"DeepSeek\u2019s next AI model delayed by attempt to use Chinese chips"},"content":{"rendered":"<p>Unlock the Editor\u2019s Digest for free<\/p>\n<p class=\"article__content-sign-up-topic-description o3-type-body-base\">Roula Khalaf, Editor of the FT, selects her favourite stories in this weekly newsletter.<\/p>\n<p>Chinese artificial intelligence company DeepSeek delayed the release of its new model after failing to train it using Huawei\u2019s chips, highlighting the limits of Beijing\u2019s push to replace US technology.<\/p>\n<p>DeepSeek was encouraged by authorities to adopt Huawei\u2019s Ascend processor rather than use Nvidia\u2019s systems after releasing its R1 model in January, according to three people familiar with the matter. <\/p>\n<p>But the Chinese start-up encountered persistent technical issues during its R2 training process using Ascend chips, prompting it to use Nvidia chips for training and Huawei\u2019s for inference, said the people.<\/p>\n<p>The issues were the main reason the model\u2019s launch was <a href=\"https:\/\/www.ft.com\/content\/fb5c11bb-1d4b-465f-8283-451a19a3d425\" data-trackable=\"link\" rel=\"nofollow noopener\" target=\"_blank\">delayed<\/a> from May, said a person with knowledge of the situation, causing it to lose ground to rivals. <\/p>\n<p>Training involves the model learning from a large dataset, while inference refers to the step of using a trained model to make predictions or generate a response, such as a chatbot query.<\/p>\n<p>DeepSeek\u2019s difficulties show how Chinese chips still lag behind their US rivals for critical tasks, highlighting the challenges facing China\u2019s drive to be technologically self-sufficient.<\/p>\n<p>The Financial Times this week reported that Beijing has demanded that Chinese tech companies <a href=\"https:\/\/www.ft.com\/content\/a29bab5e-7c2f-4487-b0a5-d594d577c73a\" data-trackable=\"link\" rel=\"nofollow noopener\" target=\"_blank\">justify their orders<\/a> of Nvidia\u2019s H20, in a move to encourage them to promote alternatives made by Huawei and Cambricon.<\/p>\n<p>Industry insiders have said the Chinese chips suffer from <a href=\"https:\/\/www.ft.com\/content\/3dab07d3-3d97-4f3b-941b-cc8a21a901d6\" data-trackable=\"link\" rel=\"nofollow noopener\" target=\"_blank\">stability issues<\/a>, slower inter-chip connectivity and inferior software compared with Nvidia\u2019s products.<\/p>\n<p>Huawei sent a team of engineers to DeepSeek\u2019s office to help the company use its AI chip to develop the R2 model, according to two people. Yet despite having the team on site, DeepSeek could not conduct a successful training run on the Ascend chip, said the people.<\/p>\n<p>DeepSeek is still working with Huawei to make the model compatible with Ascend for inference, the people said.<\/p>\n<p>Founder Liang Wenfeng has said internally he is dissatisfied with R2\u2019s progress and has been pushing to spend more time to build an advanced model that can sustain the company\u2019s lead in the AI field, they said.<\/p>\n<p>The R2 launch was also delayed because of longer-than-expected data labelling for its\u00a0updated model, another person added. Chinese media reports have suggested that the model may be released as soon as in the coming weeks. <\/p>\n<p class=\"n-content-recommended__title o3-type-body-highlight\">Recommended<\/p>\n<p><a href=\"https:\/\/www.ft.com\/content\/4f7734a9-9f47-4f23-98f8-3083cd572663\" data-trackable=\"image-link\" data-trackable-context-story-link=\"image-link\" tabindex=\"-1\" aria-hidden=\"true\" rel=\"nofollow noopener\" target=\"_blank\"><img decoding=\"async\" class=\"o-teaser__image\" src=\"https:\/\/www.newsbeep.com\/ca\/wp-content\/uploads\/2025\/08\/https:\/\/www.ft.com\/__origami\/service\/image\/v2\/images\/raw\/https%3A%2F%2Fd1e00ek4ebabms.cloudfront.net.jpeg\" alt=\"OpenAI CEO Sam Altman pictured in Tokyo in February\"\/><\/a><\/p>\n<p>\u201cModels are commodities that can be easily swapped out,\u201d said Ritwik Gupta, an AI researcher at the University of California, Berkeley. \u201cA lot of developers are using Alibaba\u2019s Qwen3, which is powerful and flexible.\u201d <\/p>\n<p>Gupta noted that Qwen3 adopted DeepSeek\u2019s core concepts, such as its training algorithm that makes the model capable of reasoning, but made them more efficient to use.<\/p>\n<p>Gupta, who tracks Huawei\u2019s AI ecosystem, said the company is facing \u201cgrowing pains\u201d in using Ascend for training, though he expects the Chinese national champion to adapt eventually. <\/p>\n<p>\u201cJust because we\u2019re not seeing leading models trained on Huawei today doesn\u2019t mean it won\u2019t happen in the future. It\u2019s a matter of time,\u201d he said.<\/p>\n<p>Nvidia, a chipmaker at the centre of a geopolitical battle between Beijing and Washington, recently agreed to give the US government a cut of its revenues in China in order to resume sales of its H20 chips to the country.<\/p>\n<p>\u201cDevelopers will play a crucial role in building the winning AI ecosystem,\u201d said Nvidia about Chinese companies using its chips.\u00a0\u201cSurrendering entire markets and developers would only hurt American economic and national security.\u201d<\/p>\n<p>DeepSeek and Huawei did not respond to a request for comment. <\/p>\n","protected":false},"excerpt":{"rendered":"Unlock the Editor\u2019s Digest for free Roula Khalaf, Editor of the FT, selects her favourite stories in this&hellip;\n","protected":false},"author":2,"featured_media":68758,"comment_status":"","ping_status":"","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[20],"tags":[62,276,277,49,48,61],"class_list":["post-68757","post","type-post","status-publish","format-standard","has-post-thumbnail","category-artificial-intelligence","tag-ai","tag-artificial-intelligence","tag-artificialintelligence","tag-ca","tag-canada","tag-technology"],"_links":{"self":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/68757","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/comments?post=68757"}],"version-history":[{"count":0,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/posts\/68757\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media\/68758"}],"wp:attachment":[{"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/media?parent=68757"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/categories?post=68757"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.newsbeep.com\/ca\/wp-json\/wp\/v2\/tags?post=68757"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}