Wall Street analysts upgraded Dell and ASML while naming NVIDIA the top AI pick for 2026, betting on companies selling infrastructure rather than applications. The moves reflect growing confidence that enterprise AI spending will concentrate on cloud platforms and hardware rather than software applications.
Microsoft Azure, Google Vertex AI, AWS Bedrock, and NVIDIA DGX Cloud are launching products to lock in enterprise customers migrating AI workloads from on-premise systems. Each platform offers different advantages: Azure integrates with Microsoft 365 deployments, Vertex AI provides Google's TPU access, AWS Bedrock enables multi-model switching, and NVIDIA DGX Cloud delivers H100 GPU clusters.
Snowflake announced Cortex AI, Notebooks, Feature Store, and Agent Evaluations at BUILD London to capture spending on AI development tools. The releases target data teams building custom models who currently piece together tools from multiple vendors. Cortex AI provides LLM access within Snowflake's data warehouse, eliminating data movement costs that can reach 20-30% of AI project budgets.
The infrastructure layer attracts investor capital because switching costs increase once enterprises standardize on platforms. Companies training models on Azure's infrastructure face retraining costs if they migrate to AWS, while AWS customers lose access to proprietary services like Bedrock's model evaluation tools. Lock-in dynamics explain why infrastructure stocks trade at higher multiples than application software.
Dell's upgrade reflects demand for on-premise AI servers from enterprises in regulated industries unable to use public cloud. Banks and healthcare companies need local GPU clusters for data residency compliance, creating a market segment Dell dominates through PowerEdge servers optimized for NVIDIA chips.
ASML produces lithography machines required to manufacture advanced chips. The company holds a monopoly on extreme ultraviolet (EUV) machines needed for 5nm and smaller process nodes. TSMC, Samsung, and Intel depend on ASML equipment to produce AI accelerators, making ASML a direct play on chip demand regardless of which hardware vendor wins market share.
Enterprise migration patterns favor established cloud providers over AI-native startups. Companies prefer platforms offering compute, storage, and AI tools from one vendor rather than integrating point solutions. This dynamic explains why Microsoft, Google, and AWS gain share while smaller AI infrastructure companies struggle to scale.