IDC prognozuje 2,7-krotny wzrost wydatków na infrastrukturę zgodną ze standardami OCP w 2029 r. do 800 mld USD

IDC podniósł prognozę rocznych globalnych wydatków na sprzęt IT zgodny ze standardami OCP w 2029 r. do około 800 mld USD, czyli 2,7-krotności wcześniejszej prognozy 295 mld USD. Wydatki mają osiągnąć 892 mld USD w 2030 r., w tym około 814 mld USD na serwery; około 628 mld USD, czyli 70% całości, ma być związane z infrastrukturą AI. Prognoza zakłada wzrost z około 154 mld USD w 2025 r. i CAGR na poziomie około 42% do 2030 r.; badanie zostało zlecone przez Fundację OCP.

Analysis

The investable signal is not that “OCP” captures a forecast share of spending; it is that AI data-center buildouts are broadening demand from accelerators into power delivery, thermal management and networking. A standards-compliant design does not guarantee that OCP members, or any particular supplier, win the purchase orders. Nor does a foundation-commissioned market forecast establish incremental revenue or returns: the sharp forecast revision makes methodology, category definitions and assumptions worth checking before treating it as a demand revision.

Over the next few days, expect sentiment support for AI infrastructure, but little basis for repricing individual companies from this release alone. Over 1–3 months, watch hyperscaler capex guidance and supplier orders/backlogs for confirmation. Over 6–18 months, power availability, grid interconnection and cooling deployment could determine which projects proceed and shift value toward enabling equipment rather than server volume. The contrary case: forecasts extrapolate planned spend, while project delays, utilization shortfalls or improving AI economics reduce realized orders; open specifications may also intensify price competition rather than protect margins.

Key Decisions for Investors

• Do not chase the forecast headline. Treat it as a sector-level demand signal, not company-specific earnings evidence; verify IDC’s scope and methodology against actual orders and capex disclosures.
• Watch a relative-value entry: long a basket of power/cooling suppliers such as Vertiv, Eaton and Schneider Electric versus a basket of server OEMs such as Dell, HPE and Super Micro, only after results show sustained order/backlog conversion. The thesis is that facility bottlenecks may make enabling equipment more valuable than undifferentiated server volume; this is not established by the forecast itself.
• Falsify that relative trade if power/cooling orders or backlog weaken, supplier guidance fails to convert demand into revenue, or server-OEM guidance materially outperforms on margins. Also reassess if hyperscalers cut AI capex or disclose lower infrastructure utilization.
• Before sizing any position, check valuations, customer concentration, backlog quality and whether reported growth is AI-related; these data are absent here. If confirmation is lacking, no trade is preferable to buying the headline.