IDC relève de 2,7 fois ses prévisions pour 2029 concernant les dépenses d'infrastructure reconnues par l'OCP, soit 800 milliards de dollars.

IDC projects annual spending on OCP-recognized IT equipment will reach $800 billion in 2029—about 2.7 times its prior forecast—and $892 billion in 2030, up from about $154 billion in 2025 (approximately 42% CAGR). In 2030, about $628 billion is forecast to be AI-related and $814 billion to be servers, with OCP-recognized equipment representing roughly 51% of spending in the infrastructure categories measured. The commissioned study points to expanding AI infrastructure investment and wider adoption of open designs; it reports forecasts, not realized spending.

Analysis

The investable signal is not the headline growth rate; it is a possible shift in where data-center economics accrue. If open rack, power and cooling designs become more interchangeable, buyers may gain procurement leverage and reduce dependence on vertically integrated systems. That could benefit component and infrastructure suppliers while pressuring vendors whose differentiation relies on proprietary platforms. Potential exposure includes liquid-cooling and power-system providers such as Vertiv and Schneider Electric, and networking suppliers such as Arista Networks; the forecast alone does not establish incremental orders or margins for any of them.

Treat the forecast revision cautiously: it was commissioned by OCP, measures a defined set of OCP-recognized equipment, and the article offers no methodology bridge explaining the 2.7x change versus last year’s estimate. Recognition is not proof that deployments will use OCP designs exclusively. Near term, the announcement is more likely a sentiment input than an earnings catalyst. Over 1–3 months, verify the underlying IDC report, hyperscaler capex plans, supplier order commentary and whether power/cooling lead times constrain installations. Over 6–18 months, the key question is whether open designs lower system costs enough to expand deployments—or commoditize hardware and shift value toward accelerators, software and power availability.

Contrarian risk: the market may extrapolate infrastructure spending while underweighting utilization, financing and electricity constraints. A slowdown in AI returns or grid delays could defer non-GPU equipment too, despite a structurally larger addressable market.

Key Decisions for Investors

• Do not trade the revised forecast as confirmed revenue. First obtain the IDC methodology and prior-year scope reconciliation; treat a material scope change as a reason to discount the headline growth signal.
• Put Vertiv, Schneider Electric and Arista Networks on an order-and-guidance watchlist, not an automatic buy list. Look for evidence of AI-related backlog conversion and profitable mix in upcoming disclosures.
• Conditional 1–3 month expression: if supplier order commentary and hyperscaler capex plans confirm continued buildout, consider a diversified power/cooling exposure rather than concentrating on a single forecast beneficiary. Reassess if capex guidance is cut or equipment lead times ease alongside weaker orders.
• Falsification triggers: hyperscalers defer data-center projects, supplier backlog or guidance weakens, or reported deployments fail to show adoption beyond OCP-aligned buyers. Separately monitor grid connection and power availability, which can cap realized spend even if budgets remain intact.