<?xml version="1.0" encoding="utf-8" standalone="yes"?><rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom"><channel><title>Custom-Silicon on Best of AI</title><link>https://bestofai.io/tags/custom-silicon/</link><description>Recent content in Custom-Silicon on Best of AI</description><generator>Hugo</generator><language>en-US</language><lastBuildDate>Sat, 03 Oct 2026 00:00:00 +0000</lastBuildDate><atom:link href="https://bestofai.io/tags/custom-silicon/index.xml" rel="self" type="application/rss+xml"/><item><title>Alibaba Zhenwu V900</title><link>https://bestofai.io/hardware/alibaba-zhenwu-v900/</link><pubDate>Sat, 03 Oct 2026 00:00:00 +0000</pubDate><guid>https://bestofai.io/hardware/alibaba-zhenwu-v900/</guid><description>&lt;p&gt;Alibaba&amp;rsquo;s chip design unit T-Head unveiled the Zhenwu V900 at the company&amp;rsquo;s Apsara Conference on September 22, 2026. The chip carries 216GB of on-chip memory and connects through T-Head&amp;rsquo;s own ICN Switch interconnect fabric, which Alibaba says can scale a single cluster to as many as 500,000 chips. Alibaba claims roughly three times the compute of its predecessor, the M890, with native support for FP8 and FP4 precision. Mass production is planned for the first quarter of 2027, so the V900 is an announced design rather than a shipping product.&lt;/p&gt;</description></item></channel></rss>