Компания IDC повысила прогноз расходов на инфраструктуру, признанную OCP, на 2029 год в 2,7 раза до 800 млрд долларов

IDC projects annual spending on OCP-recognized IT equipment will reach $800 billion in 2029, 2.7 times its prior forecast for that year, and $892 billion in 2030. The 2030 forecast includes about $628 billion in AI-related infrastructure spending (around 70% of the total) and $814 billion for servers; spending is projected to grow from about $154 billion in 2025 at a roughly 42% CAGR. The estimates come from an IDC study commissioned by the OCP Foundation and reflect forecasts, not reported spending.

Analysis

The investable signal is less the headline spending curve than the shift in where AI infrastructure economics accrue. Open designs can widen the supplier pool and make standard server/ rack hardware more substitutable, pressuring differentiation and potentially margins for assemblers. Scarce complements—accelerated compute, high-speed networking, power delivery, and liquid cooling—are likelier to retain pricing power if deployments convert into orders. That favors exposure to the bottleneck layer (for example, Nvidia and AMD in compute; Broadcom and Arista Networks in networking; Vertiv, Eaton, and Schneider Electric in power/cooling), while treating generic server exposure more cautiously.

The key qualification: IDC’s study was commissioned by OCP and measures spending on equipment recognized as OCP, not total incremental AI capex, supplier revenue, or realized profits. The sharp forecast revision may reflect expanded scope, classification, or assumptions as well as demand. Do not capitalize the forecast as an earnings estimate. Near term, sentiment may lift the AI-infrastructure complex; over 1–3 months, validate through hyperscaler capex guidance, supplier orders/backlogs, and power/cooling lead times. Over 6–18 months, grid access, power availability, and deployment returns may constrain the buildout, while open standards can increase competition and reduce hardware rents. Contrarian risk: the consensus may overvalue unit growth and underweight commoditization and the cost of electricity. The thesis weakens if hyperscalers trim capex or supplier backlog/order growth rolls over; it strengthens if power and cooling remain binding constraints alongside sustained orders.

Key Decisions for Investors

• Do not trade the IDC forecast revision in isolation: it is a commissioned, category-specific spending estimate, not a company-level revenue or earnings forecast. Verify forecast scope and methodology before updating long-term estimates.
• Prefer a relative-value expression over a broad AI-infrastructure chase: favor diversified power/cooling and networking beneficiaries such as Vertiv, Eaton, Schneider Electric, Broadcom, or Arista Networks against less differentiated server-assembly exposure. Size only after confirming order/backlog trends and valuation support; no single-name entry is justified by this release alone.
• Over the next 1–3 months, monitor hyperscaler capex guidance, AI-networking orders, liquid-cooling deployments, and utility/grid connection timelines. Sustained orders with constrained power availability support the bottleneck thesis; falling bookings or capex reductions falsify it.
• Avoid treating OCP adoption as an automatic moat: open standards may expand the addressable market while simultaneously intensifying price competition among hardware suppliers. Reassess on evidence of realized margins, not adoption claims.