AI Infrastructure Stocks Face Diverging Paths

Recent reports across the semiconductor and infrastructure space point to powerful but uneven effects from artificial intelligence demand on leading technology stocks. Companies supplying key chips, memory and data center capacity are seeing rapid growth, major new partnerships and, in some cases, substantial financial strain as AI investments scale.

Nvidia (NVDA), Micron (MU), Broadcom (AVGO) and Marvell sit at the center of the hardware build‑out, while Oracle (ORCL) and CoreWeave illustrate how servicing AI workloads through data centers can create large backlogs alongside rising debt. Analyst views are increasingly differentiated, with some highlighting valuation risks even as others initiate or reiterate positive ratings tied directly to AI exposure.

Micron Technology (MU) has delivered one of the strongest rallies in the semiconductor sector, with its stock rising 324% over the last year as investors focus on memory and storage as bottlenecks in AI infrastructure. High‑bandwidth memory has shifted from a "nice to have" to a "must‑have" component in AI training clusters, giving Micron significant pricing power in DRAM and NAND and supporting robust revenue growth and widening gross margins.

The company’s growth is driven by model training and inference workloads that require massive, low‑latency access to data, creating what some see as a structural AI memory supercycle. However, Micron has recently faced selling pressure after Alphabet (GOOGL) unveiled a lossless data compression breakthrough, TurboQuant, which has raised concerns that more efficient software could lower demand for raw NAND and DRAM as less data needs to be stored and transferred between GPU clusters.

Nvidia remains the dominant supplier of AI infrastructure, leading in data center GPUs, high‑speed networking and the CUDA software platform. Its vertical integration of CPUs, GPUs and networking into full‑stack systems is cited as a key advantage, helping reduce total cost of ownership for customers. The company reported fiscal 2026 revenue growth of 65% after a 78% gain the previous year, and adjusted earnings in the latest quarter rose 82%.

Despite recent sideways trading, multiple sources note that Wall Street’s median target price of $265 per share implies about 50% upside from roughly $177, while some analysts project even higher levels based on expected earnings growth above 50% annually through the fiscal year ending in January 2028. At the same time, Seaport Research’s Jay Goldberg has a sell rating and a $140 target, pointing to concerns about Nvidia’s $27 billion in cloud service commitments and roughly $40 billion of equity investments in customers like Anthropic, CoreWeave and OpenAI, which he views as potentially inflating demand.

Goldberg also flags competition from custom tensor processing units developed with Broadcom (AVGO) and Google, though other analysts emphasize that these alternatives have less mature software ecosystems and narrower use cases than Nvidia’s GPUs. Separately, Simply Wall St highlights Nvidia’s revised 2027 AI revenue projection of $1 trillion, a $20 billion acquisition of Groq assets to integrate Groq 3 LPUs into upcoming Rubin rack‑scale systems, and a share price that still sits below consensus analyst targets.

Broadcom has emerged as one of Nvidia’s most significant rivals in AI infrastructure. Its Tomahawk switches are described as the industry standard for scale‑out data center networking, and it holds about 60% market share in custom AI accelerators (XPUs), including tensor processing units designed for Alphabet (GOOGL). Broadcom also provides custom silicon solutions for Meta Platforms (META), ByteDance, OpenAI and Anthropic.

AI semiconductor sales at Broadcom grew 106% in the most recent quarter, helping total revenue rise 29% to $19.3 billion and earnings increase 28% to $2.05 per diluted share. Management expects revenue growth to accelerate to 46% in the second quarter as AI products become a larger share of sales, and Wall Street forecasts earnings growth of 66% annually through fiscal 2027.

Marvell Technology is also seeing its AI profile rise. On April 2, Erste Group initiated coverage with a Buy rating, citing a doubling of net profit over the past five quarters and return on equity reaching 19%. The firm expects continued growth supported by Marvell’s strength in high‑performance analog and optical DSP technologies and its position in the AI semiconductor ecosystem.

On March 31, Nvidia and Marvell announced a broad strategic partnership to integrate Marvell’s custom silicon, networking and optical solutions into Nvidia’s AI infrastructure via NVLink Fusion, alongside a $2 billion equity investment by Nvidia. Analysts view the deal as a strong endorsement of Marvell’s capabilities and expect hyperscaler adoption to drive sustained growth.

Oracle (ORCL)’s fiscal third‑quarter 2026 results underscored how AI data center construction is reshaping enterprise balance sheets. The company reported remaining performance obligations of $553 billion, up 325% year over year, much of it tied to AI data center projects. Management expects this demand to support revenue growth in fiscal 2027 and beyond, and adjusted earnings rose 21% year over year in the quarter.

However, Oracle’s long‑term debt has increased by nearly 50% in less than a year as it funds large‑scale data center builds, prompting the company to seek more pre‑funding from customers. Commentators warn that order cancellations could turn this debt into a significant headwind. CoreWeave, where Nvidia is both supplier and investor, presents a similar contrast between growth and leverage: its 2025 revenue exceeded $5.1 billion, up 167% year over year, and backlog reached $67 billion, yet total debt climbed to more than $21 billion against about $3.9 billion in liquidity.

CoreWeave’s stock has gained about 85% since its March 2025 debut, but the company has no price‑to‑earnings ratio due to losses and continues to rely on heavy capital expenditure, highlighting the financing challenges for smaller AI infrastructure players relative to Nvidia’s substantial free cash flow and liquidity.