<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:wfw="http://wellformedweb.org/CommentAPI/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:slash="http://purl.org/rss/1.0/modules/slash/"
	>

<channel>
	<title>NPO technology Archives - Gizmochina</title>
	<atom:link href="https://www.gizmochina.com/tag/npo-technology/feed/" rel="self" type="application/rss+xml" />
	<link>https://www.gizmochina.com/tag/npo-technology/</link>
	<description>Latest Tech News, Product Reviews and Deals</description>
	<lastBuildDate>Fri, 18 Sep 2026 08:46:02 +0000</lastBuildDate>
	<language>en-US</language>
	<sy:updatePeriod>
	hourly	</sy:updatePeriod>
	<sy:updateFrequency>
	1	</sy:updateFrequency>
	<generator>https://wordpress.org/?v=5.9.9</generator>
	<item>
		<title>Huawei Unveils World’s First NPO-Based AI SuperPoD Built for Massive AI Models</title>
		<link>https://www.gizmochina.com/2026/09/18/huawei-unveils-worlds-first-npo-based-ai-superpod-built-for-massive-ai-models/</link>
		
		<dc:creator><![CDATA[Joane]]></dc:creator>
		<pubDate>Fri, 18 Sep 2026 08:45:06 +0000</pubDate>
				<category><![CDATA[AI News]]></category>
		<category><![CDATA[Huawei]]></category>
		<category><![CDATA[News]]></category>
		<category><![CDATA[10 trillion parameter AI]]></category>
		<category><![CDATA[AI Computing]]></category>
		<category><![CDATA[AI supercomputer]]></category>
		<category><![CDATA[Ascend 960]]></category>
		<category><![CDATA[Hi-ONE]]></category>
		<category><![CDATA[Huawei AI]]></category>
		<category><![CDATA[Huawei Atlas 960E]]></category>
		<category><![CDATA[Huawei SuperPoD]]></category>
		<category><![CDATA[NPO technology]]></category>
		<guid isPermaLink="false">https://www.gizmochina.com/?p=754895</guid>

					<description><![CDATA[<img width="300" height="171" src="https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-300x171.png?x96852" class="webfeedsFeaturedVisual wp-post-image" alt="Huawei Atlas 960E SuperPoD" style="display: block; margin: auto; margin-bottom: 5px;max-width: 100%;" link_thumbnail="" srcset="https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-300x171.png 300w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-1024x583.png 1024w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-768x437.png 768w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-696x396.png 696w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-1068x608.png 1068w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-738x420.png 738w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117.png 1135w" sizes="(max-width: 300px) 100vw, 300px" /><p>Huawei has unveiled the Atlas 960E SuperPoD, a new AI computing system designed to handle increasingly large artificial intelligence models. The company announced the system on September 17 at HUAWEI CONNECT 2026 in Shanghai. Huawei says the Atlas 960E is the industry’s first SuperPoD based on Near-Packaged Optics (NPO). The system supports training and inference [&#8230;]</p>
<p>The post <a rel="nofollow" href="https://www.gizmochina.com/2026/09/18/huawei-unveils-worlds-first-npo-based-ai-superpod-built-for-massive-ai-models/">Huawei Unveils World’s First NPO-Based AI SuperPoD Built for Massive AI Models</a> appeared first on <a rel="nofollow" href="https://www.gizmochina.com">Gizmochina</a>.</p>
]]></description>
										<content:encoded><![CDATA[<img width="300" height="171" src="https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-300x171.png?x96852" class="webfeedsFeaturedVisual wp-post-image" alt="Huawei Atlas 960E SuperPoD" loading="lazy" style="display: block; margin: auto; margin-bottom: 5px;max-width: 100%;" link_thumbnail="" srcset="https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-300x171.png 300w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-1024x583.png 1024w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-768x437.png 768w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-696x396.png 696w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-1068x608.png 1068w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-738x420.png 738w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117.png 1135w" sizes="(max-width: 300px) 100vw, 300px" />
<p>Huawei has unveiled the <strong>Atlas 960E SuperPoD</strong>, a new AI computing system designed to handle increasingly large artificial intelligence models. The company announced the system on September 17 at HUAWEI CONNECT 2026 in Shanghai.</p>



<p>Huawei says the Atlas 960E is the industry’s first SuperPoD based on Near-Packaged Optics (NPO). The system supports training and inference for AI models approaching 10 trillion parameters.</p>



<div class="wp-block-image"><figure class="aligncenter size-large"><img loading="lazy" width="1024" height="583" src="https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-1024x583.png?x96852" alt="" class="wp-image-754901" srcset="https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-1024x583.png 1024w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-300x171.png 300w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-768x437.png 768w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-696x396.png 696w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-1068x608.png 1068w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117-738x420.png 738w, https://www.gizmochina.com/wp-content/uploads/2026/09/image-117.png 1135w" sizes="(max-width: 1024px) 100vw, 1024px" /></figure></div>



<h3><strong>Built for Large-Scale AI</strong></h3>



<p>The Atlas 960E combines Huawei’s UnifiedBus technology, also known as Lingqu, with the company’s Hi-ONE optical system. The architecture uses unified memory addressing, allowing computing resources to work more like a single system rather than separate machines.</p>



<p>A single SuperPoD can support up to 4,096 Ascend AI processors and deliver up to 8 EFLOPS of FP8 performance and 16 EFLOPS of FP4 performance. Huawei says the system uses a fully liquid-cooled design and an orthogonal architecture.</p>



<h3><strong>Hi-ONE Cuts Optical Hardware</strong></h3>



<p>A key part of the system is Huawei’s Hi-ONE NPO optical engine. Huawei says it offers 7.2 terabits per second of transmission capacity and includes an integrated light source.</p>



<p>The Atlas 960E uses about 5,500 Hi-ONE units, replacing roughly 48,000 traditional 800G optical modules. Huawei claims the design can reduce power consumption by more than 550 kilowatts, double fault-free operating time and provide 99.8% system availability.</p>



<h3><strong>Scaling to 1 Million AI Processors</strong></h3>



<p>Huawei is also extending the SuperPoD approach to general-purpose computing. Its upgraded TaiShan 950 SuperPoD supports up to 4,096 nodes and a 256 TB unified memory pool, while the OceanStor M900 provides petabyte-scale key-value caching for AI workloads.</p>



<p>Huawei also plans to connect multiple Atlas 960E SuperPoDs into larger AI clusters using UnifiedBus and RoCE networking. The company says its architecture can connect up to 512,000 AI processors, while a multi-rail topology can scale an Ascend SuperCluster to 1 million processors.</p>



<p>The Atlas 960E is scheduled for Q3 2027. Huawei is also preparing the Ascend 960 DT for Q1 2027, the Ascend 960 PR for Q3 2027, followed by the Ascend 970 in 2028 and Ascend 980 in 2029.</p>



<h3><strong>Huawei Atlas 960E SuperPoD: Key Specifications</strong></h3>



<figure class="wp-block-table"><table><thead><tr><th>Feature</th><th>Huawei Atlas 960E SuperPoD</th></tr></thead><tbody><tr><td>Launch</td><td>Q3 2027</td></tr><tr><td>AI model target</td><td>Up to 10 trillion parameters</td></tr><tr><td>NPO optical engine</td><td>Hi-ONE</td></tr><tr><td>Transmission capacity</td><td>7.2 Tbit/s</td></tr><tr><td>Maximum Ascend cards</td><td>4,096</td></tr><tr><td>FP8 performance</td><td>8 EFLOPS</td></tr><tr><td>FP4 performance</td><td>16 EFLOPS</td></tr><tr><td>Optical modules</td><td>5,500 Hi-ONE</td></tr><tr><td>Replaced 800G modules</td><td>Around 48,000</td></tr><tr><td>Claimed power reduction</td><td>Over 550 kW</td></tr><tr><td>System availability</td><td>99.8%</td></tr><tr><td>Interconnect</td><td>UnifiedBus / Lingqu</td></tr><tr><td>Cooling</td><td>Fully liquid-cooled</td></tr><tr><td>Maximum SuperCluster scale</td><td>Up to 1 million Ascend processors</td></tr></tbody></table><figcaption><strong>Disclaimer: Specs are based on available data. Actual performance may vary. Verify details from official sources before buying.</strong><br></figcaption></figure>



<p><strong><span style="text-decoration: underline;"><em>Read More:</em></span></strong></p>



<ul><li><a href="https://www.gizmochina.com/2026/09/18/tecno-pova-8-pro-5g-tonino-lamborghini-limited-edition/">TECNO POVA 8 Pro 5G Tonino Lamborghini Limited Edition Blends Italian Design With Smart Technology</a></li><li><a href="https://www.gizmochina.com/2026/09/15/lenovo-legion-c700-price-specs-availability/">Lenovo Legion C700 Now Available at Giztop: Specs and Price</a></li></ul>



<p>(<a href="https://www.huawei.com/en/news/2026/9/hc-ascend960-supernode" target="_blank" rel="noreferrer noopener">via</a>)</p>
<p>The post <a rel="nofollow" href="https://www.gizmochina.com/2026/09/18/huawei-unveils-worlds-first-npo-based-ai-superpod-built-for-massive-ai-models/">Huawei Unveils World’s First NPO-Based AI SuperPoD Built for Massive AI Models</a> appeared first on <a rel="nofollow" href="https://www.gizmochina.com">Gizmochina</a>.</p>
]]></content:encoded>
					
		
		
			</item>
	</channel>
</rss>

<!--
Performance optimized by W3 Total Cache. Learn more: https://www.boldgrid.com/w3-total-cache/

Object Caching 34/41 objects using Redis
Page Caching using Disk: Enhanced 
Content Delivery Network Full Site Delivery via cloudflare
Database Caching 15/24 queries in 0.008 seconds using Redis
Fragment Caching 2/3 fragments using Redis

Served from: www.gizmochina.com @ 2026-09-18 09:48:42 by W3 Total Cache
-->