A frequently heard statement in the tech sector is that the speed of development is crucial and the time it takes to launch a product or service is the key factor in determining success. You likely know the well-known saying, “move quickly and break things.” This ideology, initially based in software development, has gradually influenced the significant infrastructure deployments that underpin the AI transformation. The emergent AI tasks of the future have induced remarkable surges in rack power densities, escalating from a controllable 10-15kW to an astounding 200kW+ per rack. As GPU roadmaps move from biennial updates to annual releases, hyperscalers are installing infrastructure with shortened timelines. Cooling technologies such as direct-to-chip liquid cooling, which were previously considered specialized, have become essential standards. Within this context, streamlining development cycles can be the deciding factor between seizing a market trend or completely missing it. This situation presents an essential dilemma for hardware designers, especially those working on cooling systems. Rapid advancement and breaking of things in software development is a feasible approach when updates can be disseminated instantly and bugs can be rectified post-deployment. Conversely, hardware operates under distinct constraints. Once a cold plate design goes into production, making design modifications becomes expensive, conducting qualification tests takes up precious time, and reliability problems can negatively affect system performance, availability, and customer trust. – Getty Images. Concurrently, the thermal requirements of contemporary AI servers are leaving a diminishing tolerance for mistakes.<\/p>\n
\u00a0<\/p>","protected":false},"excerpt":{"rendered":"
A frequently heard statement in the tech sector is that the speed of development is crucial and the time it takes to launch a…<\/p>\n","protected":false},"author":0,"featured_media":2161,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[11],"tags":[],"class_list":["post-2160","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-data-center"],"_links":{"self":[{"href":"https:\/\/trustedainews.com\/index.php?rest_route=\/wp\/v2\/posts\/2160","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/trustedainews.com\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/trustedainews.com\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/trustedainews.com\/index.php?rest_route=\/wp\/v2\/users\/2"}],"replies":[{"embeddable":true,"href":"https:\/\/trustedainews.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=2160"}],"version-history":[{"count":0,"href":"https:\/\/trustedainews.com\/index.php?rest_route=\/wp\/v2\/posts\/2160\/revisions"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/trustedainews.com\/index.php?rest_route=\/wp\/v2\/media\/2161"}],"wp:attachment":[{"href":"https:\/\/trustedainews.com\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=2160"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/trustedainews.com\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=2160"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/trustedainews.com\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=2160"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}