Cloud growth, margins and AI spending—who is executing and who is stalling

A fresh round of Big Tech results sketched a clearer playbook for the final stretch of 2025. Cloud platforms that married enterprise AI demand to disciplined capital spending extended their lead; ad-funded platforms rose on better ad yields and user growth, but faced volatility from brand safety and regulatory headwinds; device makers with premium tiers offset softer midrange demand through upselling and services. The through-line: investors rewarded credible paths to free cash flow, punished vague AI narratives, and demanded transparency on data-center capex and electricity contracts. Companies breaking out inference versus training workloads, power procurement, and per-unit compute economics saw calmer post-earnings trade. Those leaning on “AI-adjacent” talk without cost controls met sharp drawdowns.

Winners, laggards and the road to 2026

Winners paired AI revenue with durable gross margins—think usage-based cloud tied to committed enterprise spend, ad platforms with improved targeting, and subscription bundles that cut churn. Laggards overbuilt capacity or chased loss-making hardware to “show” AI. Guidance mattered: when executives quantified data-center megawatts, network spend, and depreciation schedules, the street modeled earnings with less fear. Capital returns helped—buybacks signaled confidence where net cash swelled. The market also re-rated firms with credible on-device AI plans that reduce inference costs, and those localizing models for privacy-sensitive clients. Heading into 2026, watch: power constraints in key regions, GPU supply normalization, and whether AI features prompt real upgrade cycles rather than one-off curiosity. The lesson is simple: prove unit economics, show power strategy, and deliver sticky use cases—or expect multiple compression.