IDC eleva 2,7 veces su previsión para 2029 del gasto en infraestructura reconocida por OCP

IDC raised its 2029 forecast for annual spending on OCP-recognized IT equipment to approximately $800 billion, 2.7 times the prior-year estimate of $295 billion; it projects $892 billion in 2030, including $814 billion for servers. Spending is forecast to rise from about $154 billion in 2025 at a roughly 42% CAGR, with AI-related infrastructure accounting for approximately $628 billion, or nearly 70%, of the 2030 total. The study attributes the outlook to expanding AI infrastructure investment and broader adoption of open designs.

Analysis

The investable signal is not the headline growth rate; it is the potential shift in where data-center dollars accrue. If the forecast translates into funded deployments, power delivery, liquid cooling, networking and rack infrastructure become bottlenecks alongside accelerators. That could broaden beneficiaries beyond GPU vendors to Vertiv, Eaton and Schneider Electric, while Arista Networks and Broadcom could benefit if open Ethernet designs gain share. Nvidia remains exposed to the buildout, but open, modular designs may also make it easier for customers to qualify alternative accelerators and reduce dependence on proprietary systems. Standardization is therefore a volume tailwind but could pressure differentiation and pricing for server assemblers such as Dell Technologies, HPE and Supermicro.

Treat the estimate as a directional demand signal, not a revenue forecast: IDC’s study was commissioned by OCP, and “OCP-recognized” equipment is not equivalent to OCP revenue, incremental spending, or guaranteed supplier share. The unusually large forecast revision may reflect changed assumptions or category coverage; reconcile definitions and compare with hyperscaler capex, supplier orders and backlog before underwriting it. Near term, the announcement alone is unlikely to establish a new earnings trajectory. Over 1–3 months, capex guidance and power/cooling order trends are the useful tests. Over 6–18 months, grid access, equipment lead times and utilization could constrain realized spending or delay returns. Contrarian risk: the market may extrapolate an ambitious ecosystem forecast while overlooking power availability and the possibility that AI infrastructure spend shifts toward custom silicon and lower-cost architectures.

Key Decisions for Investors

• Do not trade the forecast revision as a standalone earnings catalyst. First verify the study’s category definitions and whether the 2025–2030 series is comparable with the prior forecast.
• Watch for a conditional relative-value opportunity: favor power and cooling exposure, including Vertiv, Eaton and Schneider Electric, over undifferentiated server-assembly exposure if orders and backlog confirm capacity tightness. Avoid initiating solely on this release.
• Track hyperscaler capex guidance, liquid-cooling and power-equipment orders, and data-center interconnection timelines over the next 1–3 months. These would validate or falsify the forecast’s conversion into shipments.
• Reassess the bullish infrastructure thesis if major cloud providers reduce capex plans, supplier backlogs weaken, or grid constraints defer deployments; reassess server-vendor risk if open designs demonstrably broaden qualified accelerator supply and intensify price competition.