{"id":439893,"date":"2026-10-13T04:27:33","date_gmt":"2026-10-12T21:27:33","guid":{"rendered":"https:\/\/www.swingfish.trade\/blog\/market-news\/goldman-sachs-flags-power-and-compute-as-two-chokepoints-in-the-ai-boom-439893\/"},"modified":"2026-10-13T04:27:33","modified_gmt":"2026-10-12T21:27:33","slug":"goldman-sachs-flags-power-and-compute-as-two-chokepoints-in-the-ai-boom","status":"publish","type":"post","link":"https:\/\/www.swingfish.trade\/blog\/market-news\/goldman-sachs-flags-power-and-compute-as-two-chokepoints-in-the-ai-boom-439893\/","title":{"rendered":"Goldman Sachs flags power and compute as two chokepoints in the AI boom"},"content":{"rendered":"<div>\n<p dir=\"ltr\">The note points investors toward the physical layer of the AI trade, with access to high-voltage power now a tighter constraint than the capital to buy chips. That favours operators that have already secured grid capacity, and it suggests power availability, more than chip supply, could set the pace of new data centre builds through 2027. In software, budgets moving toward data and cybersecurity point to a widening gap between foundational platforms and application vendors, whose pricing models could come under pressure as agents replace human users. The emphasis on core platform returns also implies investors may reward proven monetisation of compute over speculative new revenue lines.<\/p>\n<p dir=\"ltr\">&#8212;<\/p>\n<p dir=\"ltr\">\nIn Goldman\u2019s reading of the AI boom, money can buy the GPUs, but power decides who gets to switch them on.<\/p>\n<p dir=\"ltr\">Summary:<\/p>\n<ul dir=\"ltr\">\n<li>Goldman Sachs says demand for AI infrastructure keeps outpacing capacity, and the compute supply-demand gap is widening, based on its discussions with tech company management teams<\/li>\n<li>A likely slowdown in model training is being offset by rising investment in AI safety and pacing monitoring, which keeps compute demand elevated<\/li>\n<li>Leading hyperscalers are securing power contracts for 2028 and beyond because grid capacity for 2026 and 2027 is essentially fully booked<\/li>\n<li>Capex is rising on component inflation and the need for more chips, while revenue generated per unit of power has risen by around two-thirds over the past year<\/li>\n<li>Goldman sees data and security platforms among the clearest beneficiaries of AI spending, and says application software is increasingly being built for AI agents rather than people<\/li>\n<li>Meta is generating two to three times return on investment per megawatt in its core business, and Google is using a full-stack approach spanning custom silicon, cloud, security and distribution<\/li>\n<\/ul>\n<p dir=\"ltr\">\nDemand for AI infrastructure continues to run ahead of available capacity, and the bottleneck is moving from chips toward power, according to Goldman Sachs, which drew on recent discussions with the management teams of technology companies.<\/p>\n<p dir=\"ltr\">The bank said the gap between compute supply and demand is widening, driven by new infrastructure requirements and resilient enterprise spending. Goldman expects model training to slow, but it said companies are stepping up investment in AI safety and pacing monitoring, work that needs more computing power and so keeps overall compute demand high.<\/p>\n<p dir=\"ltr\">In Goldman\u2019s view, power has become the defining constraint. Grid capacity for 2026 and 2027 is essentially booked out, the bank said, and leading hyperscalers are therefore moving aggressively to secure power contracts for 2028 and beyond. Goldman argued that having the capital to buy GPUs is no longer enough on its own, and that access to high-voltage energy infrastructure now decides who can expand. Capital expenditure is climbing on component inflation and demand for more chips. Over the same period, revenue generated from each unit of power has risen by around two-thirds compared with earlier price points over the past year.<\/p>\n<p dir=\"ltr\">The bank also described a structural shift in enterprise software. It believes data and security platforms are among the clearest beneficiaries of AI spending, because AI depends on clean, organised and secure data. Chief information officers are increasingly steering budgets toward foundational data layers and cybersecurity tools, aiming to protect data integrity and contain the risks of unsanctioned \u201cshadow AI\u201d use inside organisations.<\/p>\n<p dir=\"ltr\">Application software is going \u201cheadless\u201d, meaning it is increasingly being designed for AI agents rather than human users. Goldman said this lets vendors build more powerful AI features into existing platforms and supports industry growth for now. It cautioned, however, that the change could eventually alter how software companies price and monetise their products.<\/p>\n<p dir=\"ltr\">Among the companies it highlighted, Goldman said Meta continues to generate two to three times return on investment per megawatt of power in its core business while exploring new sources of incremental revenue. Google, the bank said, is relying on a full-stack advantage that combines custom silicon, cloud, security and a vast distribution base to drive core monetisation. In that picture, sales of its TPU chips are one part of a broader portfolio strategy.<\/p>\n<p dir=\"ltr\">Goldman concluded that AI monetisation offers substantial optionality. Even so, in its view core platform monetisation remains among the higher-confidence and higher-return uses of compute, which suggests the established platforms that already earn strong returns on their power may stay the more dependable beneficiaries as capacity remains scarce.\u00a0<\/p>\n<p dir=\"ltr\">\n<p dir=\"ltr\">\n<p>                            This article was written by Eamonn Sheridan at investinglive.com.<\/p><\/div>\n","protected":false},"excerpt":{"rendered":"<p>The note points investors toward the physical layer of the AI trade, with access to high-voltage power now a tighter constraint than the capital to buy chips. That favours operators that have&hellip;<\/p>\n","protected":false},"author":216,"featured_media":0,"comment_status":"open","ping_status":"closed","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[86],"tags":[],"class_list":["post-439893","post","type-post","status-publish","format-standard","hentry","category-market-news"],"_links":{"self":[{"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/posts\/439893","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/users\/216"}],"replies":[{"embeddable":true,"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/comments?post=439893"}],"version-history":[{"count":0,"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/posts\/439893\/revisions"}],"wp:attachment":[{"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/media?parent=439893"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/categories?post=439893"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.swingfish.trade\/blog\/wp-json\/wp\/v2\/tags?post=439893"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}