Investors are pivoting beyond chipmakers to AI infrastructure leaders like Arista Networks, Vertiv, and CoreWeave, which provide the essential networking, power, and cloud capacity for hyperscale data centers.
Vertiv is capitalizing on the physical infrastructure needs of AI data centers, specifically through its liquid-cooling and power management systems. The company raised its full-year 2026 EPS guidance to $6.65-$6.75, reflecting strong demand as data centers transition to higher-voltage, liquid-cooled architectures.
Vertiv provides the 'picks and shovels' including switchgear, UPS systems, battery storage, and liquid-cooling loops necessary to operate hyperscale AI sites.
CoreWeave is experiencing explosive growth with 112% revenue increases and a $104 billion backlog, positioning it as a primary provider of GPU cloud capacity. However, the company faces significant financial pressure, characterized by heavy quarterly CapEx of $6.4 billion and deeply negative free cash flow, making it a high-beta, high-leverage investment.
CoreWeave rents NVIDIA GPU clusters by the hour or via multi-year contracts to AI labs and hyperscalers that require high-performance compute capacity without managing the hardware themselves.
Arista Networks, Vertiv, and CoreWeave are driving the physical AI buildout through networking, power, and cloud capacity. While Arista delivers record margins and Vertiv scales its cooling and electrical systems, CoreWeave faces high-growth volatility with significant capital expenditure requirements.
These companies provide the physical infrastructure—networking fabrics, power/cooling systems, and GPU cloud capacity—that enables AI models to function at scale beyond the chips themselves.