AI infrastructure spending is continuing to drive demand across the data-center supply chain, with investors increasingly focused on networking, cloud services and power-management hardware. Nvidia’s recent quarter underscored the scale and durability of the build-out, as its data center revenue surged 92% year over year, signaling a sustained investment cycle for companies supporting advanced AI workloads.
Against that backdrop, three names highlighted by the market are Astera Labs for high-speed connectivity, CoreWeave for GPU-optimized AI cloud capacity, and Vertiv for power and cooling systems designed for rising chip density.
Key takeaways
- Astera Labs: Shares have gained sharply over the past year, supported by demand for fast networking hardware used to move data within AI data centers.
- Catalyst: Nvidia’s 92% year-over-year jump in data center segment revenue reinforced expectations of continued AI “factory” build-outs.
- CoreWeave: The company reported revenue more than doubling year over year, and it cites a large $100 billion backlog as visibility into future capacity delivery.
- Vertiv: Power and cooling demand is expected to accelerate as chip density rises, with management projecting revenue growth of 34% to $13.8 billion for the year.
- Implication: Investors may continue to look beyond chipmakers toward infrastructure enablers tied to the pace of data-center construction and power-efficiency requirements.
What drove the shift toward AI infrastructure
Nvidia’s latest results provided a fresh data point for investors tracking the pace of AI deployment. According to Nvidia’s chief financial officer Colette Kress, “the build-out of AI factories is accelerating,” a view reiterated on the company’s May earnings call. The chip supplier also reported a 92% year-over-year increase in data center segment revenue last quarter.
For infrastructure suppliers, the message is straightforward: as AI workloads scale, so does the need for faster networking to reduce bottlenecks, cloud platforms that can provision compute quickly, and power and cooling systems capable of handling higher density racks. The availability and efficiency of power at data centers has become an additional constraint, increasing the focus on operational design and energy management.
Astera Labs: networking capacity for AI data centers
Astera Labs supplies high-speed networking hardware intended to help move data rapidly among chips inside data centers. The company’s approach targets performance needs created by modern AI training and inference workloads, where data movement can materially affect throughput.
According to the article, Astera Labs’ shares have risen 357% over the past year. The thesis centers on the firm’s product expansion and its ability to broaden revenue streams as data centers adopt more complex rack configurations with multiple chip types.
The company expanded its portfolio over the past three years, and the article notes that Scorpio smart fabric switches introduced in 2024 accounted for 15% of total revenue in 2025. Management also highlighted that its COSMOS software platform provides diagnostic capabilities integrated into customer systems, which the article says can make switching suppliers more difficult.
On growth fundamentals, the article cites that revenue nearly doubled year over year to $308 million in the first quarter. Analysts quoted in the article also expect adjusted earnings to rise by 69% this year, with continued high double-digit growth in subsequent years.
Still, the demand outlook is tied to data-center construction and spending trends. The article argues that in the longer term, networking requirements should increase as data centers become more complex, but any slowdown in build activity could pressure expectations.
CoreWeave: AI cloud capacity and backlog visibility
CoreWeave operates a cloud services platform optimized for GPUs and designed to support generative AI workloads. The business model aims to reduce the cost and complexity for customers that need AI compute capacity without building and managing infrastructure in-house.
According to the article, CoreWeave’s revenue more than doubled year over year in the first quarter, reaching almost $2.1 billion. The company has also expanded its data center portfolio to more than 50, with the article emphasizing its ability to navigate the build process—covering power, cooling, components and software—so facilities can be made AI-ready within weeks.
A key part of the investment case is capacity visibility. The article states that CoreWeave has a $100 billion backlog and notes that financial services customers now represent 10% of that backlog, positioning the company to mitigate customer concentration risk if spending from larger customers changes.
Analysts referenced in the article expect revenue to reach nearly $40 billion by 2028. The piece also argues that if the stock trades around 10 times sales—described as still reasonable for a fast-growing AI cloud infrastructure provider—the resulting valuation could be supported by the company’s growth trajectory.
Vertiv: power and cooling as AI density rises
Vertiv is focused on power management and cooling systems, a segment that becomes increasingly critical as chip density rises in AI data centers. According to the article, Vertiv’s revenue growth has strengthened over the past few years and accelerated to 30% year over year in the first quarter.
Management is projecting further acceleration, with full-year guidance calling for revenue growth of 34% to $13.8 billion. The article also describes Vertiv as positioned for margin improvement, supported by product breadth, deep integration into customer operations and service growth.
On profitability expectations, the article says Vertiv is on track to lift its adjusted operating margin above 27% by 2030, driven by cost leverage, service expansion and execution.
It also notes that analysts expect earnings to grow at an annualized rate of 32% over the next several years. The investment framing suggests that faster earnings growth could help support a valuation premium, particularly for an infrastructure provider tied to the constraints of power availability and the operational demands of high-density AI deployments.
What to watch next
Investors monitoring this theme should focus on ongoing data-center build-out signals, whether power and cooling capacity keeps pace with AI chip deployments, and updates to guidance from both infrastructure suppliers and GPU-focused demand drivers. Upcoming earnings reports and additional commentary on AI spending plans from large platform providers—and major data-center build indicators—will likely shape expectations for networking, AI cloud capacity and power-management demand in the quarters ahead.
