{"id":338698,"date":"2025-06-22T20:41:00","date_gmt":"2025-06-23T03:41:00","guid":{"rendered":"https:\/\/cms-articles.softonic.io\/en\/?p=338698"},"modified":"2025-07-01T14:19:45","modified_gmt":"2025-07-01T21:19:45","slug":"huaweis-cloudmatrix-384-outperforms-nvidias-h800-in-ai-performance","status":"publish","type":"post","link":"https:\/\/cms-articles.softonic.io\/en\/huaweis-cloudmatrix-384-outperforms-nvidias-h800-in-ai-performance\/","title":{"rendered":"Huawei&#8217;s CloudMatrix 384 Outperforms Nvidia&#8217;s H800 in AI Performance"},"content":{"rendered":"\n<p class=\"wp-block-paragraph\">Huawei&#8217;s ambitious foray into artificial intelligence infrastructure, the CloudMatrix 384, is turning heads in the tech community.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\"><strong>Equipped with 384 Ascend 910C chips<\/strong>, the new system has reportedly outperformed Nvidia&#8217;s H800 chip when<a href=\"https:\/\/www.softonic.com\/articulos\/tsmc-y-nvidia-se-alian-como-respuesta-al-desafio-de-deepseek\" target=\"_blank\" rel=\"noopener\" title=\" running the DeepSeek R1 model\"> running the DeepSeek R1 model<\/a>, which consists of 671 billion parameters.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">This claim comes from a newly published technical paper in collaboration with the Chinese AI startup SiliconFlow, suggesting that Huawei&#8217;s approach may have tackled key performance metrics where Nvidia has traditionally dominated.<\/p>\n\n\n\n<figure class=\"wp-block-embed is-type-rich is-provider-twitter wp-block-embed-twitter\"><div class=\"wp-block-embed__wrapper\">\n<blockquote class=\"twitter-tweet\" data-width=\"550\" data-dnt=\"true\"><p lang=\"en\" dir=\"ltr\">A new report says that Huawei&#39;s CloudMatrix 384 outperforms Nvidia processors running DeepSeek R1, which is to be expected given the energy use involved.   <a href=\"https:\/\/t.co\/CQat96xOeg\">https:\/\/t.co\/CQat96xOeg<\/a><\/p>&mdash; Tom&#39;s Hardware (@tomshardware) <a href=\"https:\/\/twitter.com\/tomshardware\/status\/1936010229579497738?ref_src=twsrc%5Etfw\">June 20, 2025<\/a><\/blockquote><script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script>\n<\/div><\/figure>\n\n\n\n<h2 class=\"wp-block-heading\">New AI Infrastructure from Huawei Promises High Power but at a High Energy Cost<\/h2>\n\n\n\n<p class=\"wp-block-paragraph\">The CloudMatrix 384 stands as a testament to Huawei\u2019s brute-force approach to AI capabilities, especially notable given the company&#8217;s restrictions on accessing cutting-edge chip production.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">This rack-scale system features a <strong>combination of 384 dual-chiplet HiSilicon Ascend 910C neural processing units (NPUs) and 192 central processing units (CPUs)<\/strong>, spread across 16 server racks interconnected with optical connections for enhanced speed.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Despite its high computational output, reaching an impressive 300 PFLOPs of BF16 compute, the CloudMatrix has a significant drawback: energy consumption. <strong>The system reportedly consumes four times more energy than Nvidia&#8217;s GB200 NVL72<\/strong>, racking up a total of 559 kW compared to the NVL72&#8217;s 145 kW, which raises questions about operational efficiency.<\/p>\n\n\n\n<figure class=\"wp-block-embed is-type-rich is-provider-twitter wp-block-embed-twitter\"><div class=\"wp-block-embed__wrapper\">\n<blockquote class=\"twitter-tweet\" data-width=\"550\" data-dnt=\"true\"><p lang=\"en\" dir=\"ltr\">Huawei AI CloudMatrix 384<br>China\u2019s Answer to Nvidia GB200 NVL72<br>China Abundance of Power, 100% Optics, 0% Copper<br>Power Inefficiency<br>2.6x lower FLOP per Watt<br>14 Transceivers per Chip, Linear Pluggable Optics<a href=\"https:\/\/t.co\/EhhcgXmDLs\">https:\/\/t.co\/EhhcgXmDLs<\/a><\/p>&mdash; SemiAnalysis (@SemiAnalysis_) <a href=\"https:\/\/twitter.com\/SemiAnalysis_\/status\/1912366722717479002?ref_src=twsrc%5Etfw\">April 16, 2025<\/a><\/blockquote><script async src=\"https:\/\/platform.twitter.com\/widgets.js\" charset=\"utf-8\"><\/script>\n<\/div><\/figure>\n\n\n\n<p class=\"wp-block-paragraph\">Notably, Chinese firms are unable to access Nvidia-powered AI clusters, which makes the relevance of direct performance comparisons with Nvidia somewhat diminished in that market. <strong>Furthermore, the recent drop in electricity costs in China by nearly 40% over the last three years <\/strong>may make the CloudMatrix&#8217;s energy consumption less of a hindrance to prospective customers.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">Nvidia CEO Jensen Huang has maintained that his company remains a generation ahead in technology, acknowledging that while Huawei&#8217;s advancements are notable, scalability remains a challenge.<\/p>\n\n\n\n<p class=\"wp-block-paragraph\">\u201cAI is a parallel problem\u201d, he stated, emphasizing the need for robust systems to compete effectively. Still, the CloudMatrix presents a viable option for Chinese enterprises aiming for high-performance AI solutions amid geopolitical restrictions.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>Huawei&#8217;s ambitious foray into artificial intelligence infrastructure, the CloudMatrix 384, is turning heads in the tech community. Equipped with 384 Ascend 910C chips, the new system has reportedly outperformed Nvidia&#8217;s H800 chip when running the DeepSeek R1 model, which consists of 671 billion parameters. This claim comes from a newly published technical paper in collaboration &hellip; <a href=\"https:\/\/cms-articles.softonic.io\/en\/huaweis-cloudmatrix-384-outperforms-nvidias-h800-in-ai-performance\/\" class=\"more-link\">Continue reading<span class=\"screen-reader-text\"> &#8220;Huawei&#8217;s CloudMatrix 384 Outperforms Nvidia&#8217;s H800 in AI Performance&#8221;<\/span><\/a><\/p>\n","protected":false},"author":9265,"featured_media":338699,"comment_status":"closed","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":"","wpcf-pageviews":0},"categories":[1015],"tags":[2041,3885,4793],"usertag":[],"vertical":[],"content-category":[6987],"class_list":["post-338698","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-news","tag-huawei","tag-inteligencia-artificial","tag-nvidia","content-category-ia"],"aioseo_notices":[],"_links":{"self":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts\/338698","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/users\/9265"}],"replies":[{"embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/comments?post=338698"}],"version-history":[{"count":1,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts\/338698\/revisions"}],"predecessor-version":[{"id":338700,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/posts\/338698\/revisions\/338700"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/media\/338699"}],"wp:attachment":[{"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/media?parent=338698"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/categories?post=338698"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/tags?post=338698"},{"taxonomy":"usertag","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/usertag?post=338698"},{"taxonomy":"vertical","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/vertical?post=338698"},{"taxonomy":"content-category","embeddable":true,"href":"https:\/\/cms-articles.softonic.io\/en\/wp-json\/wp\/v2\/content-category?post=338698"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}