Dell'Oro Group says hyperscalers focus on capex discipline in 2026
Hyperscale Artificial Intelligence (AI) infrastructure enters a mature phase as hyperscalers prioritize Capital Expenditure (CAPEX) discipline and efficient scaling of AI compute in 2026, affecting procurement, deployment timelines, and infrastructure cost planning for enterprise IT leaders.
Market Overview
After several years of regional expansion driven by resilience, redundancy, and data sovereignty, hyperscalers now emphasize scaling AI compute and supporting infrastructure efficiently while absolute investment levels remain historically high.
Key Findings
Spending on high-end accelerated servers rose sharply in 2025 and remains the core spending driver, pulling demand for GPUs, custom accelerators, High Bandwidth Memory (HBM), high-capacity SSDs, and high-speed NICs and networks used in large AI clusters.
Inference workloads are growing as hyperscalers scale AI services to millions of users, increasing requirements for availability, geographic distribution, and latency guarantees compared with centralized training clusters.
Segment or Supplier Performance
High-end GPUs will remain the largest contributor to component market revenue growth in 2026, and NVIDIA is expected to begin shipping the Vera Rubin platform in 2H26, which adds compute and networking density and optional Rubin CPX inference Graphics Processing Unit (GPU) configurations.
AMD is positioning to gain share with its MI400 rack-scale platform and reported wins at OpenAI and Oracle; suppliers of CPUs and general-purpose server components face deceleration after short-term inventory digestion.
Technology or Trend Analysis
Near-edge data centers will require investment to meet latency, reliability, and regulatory requirements for user-facing AI services, favoring smaller but highly dense accelerated clusters with high-speed networking and local storage.
Memory vendors are prioritizing production of higher-margin HBM, constraining capacity for conventional DRAM and NAND and contributing to rising memory and storage prices, while longer lead times for advanced substrates, optics, and networking add supply-chain volatility.
Forecast or Analyst Outlook
US hyperscale cloud providers continue to raise CAPEX guidance, supporting multi-year AI investment into 2026, though infrastructure investment has outpaced revenue growth and increases scrutiny of CAPEX intensity, depreciation, and long-term returns.
Changes in depreciation treatment provide levers to optimize cash flow, but underlying ROI depends on AI monetization and successful execution of deployments, raising execution risk for infrastructure planners.
The analyst frames the 2026 cycle around CAPEX discipline and execution risk, highlighting implications for procurement and deployment planning for enterprise and technical decision-makers. This Analyst Signals brief reflects a neutral, fact-based summary of the original research note.