AI Infrastructure Stocks Gain Analyst Favor on Shift to Inference Workloads

Several specialized artificial intelligence infrastructure providers received favorable analyst ratings this week, as Wall Street increasingly focuses on the market's shift from model training to the widespread deployment of AI applications, known as inference.

Analysts argue that as more enterprises adopt open-source models and AI agents, the bulk of AI-related spending will move toward running these models in production environments. This trend is creating opportunities for specialized cloud providers that offer integrated platforms, positioning them as viable alternatives to traditional hyperscale competitors.

Key Upgrades and Initiations

Investment banks Truist and Baird highlighted three companies poised to capitalize on this next phase of the AI infrastructure cycle:

• CoreWeave (CRWV): Truist upgraded the company to Buy from Hold, calling it the leading "neocloud" provider. The firm noted CoreWeave has about 1 gigawatt of active power online and a significant revenue backlog. Truist set a price target of $126, slightly down from $131, but maintained its revenue forecasts.
• DigitalOcean (DOCN): Baird initiated coverage with an Outperform rating and a $165 price target. The brokerage described DigitalOcean as a platform evolving to serve digital-native enterprises with a full-stack AI offering, benefiting from a flexible capacity model.
• Nebius (NBIS): Baird also initiated coverage on Nebius with an Outperform rating and a $250 price target. The firm views Nebius as one of the strongest beneficiaries of the inference trend, pointing to its full-stack cloud platform, in-house hardware design, and specialized software like its Token Factory.

Ad

Market Implications

The analyst commentary suggests the investment landscape for AI is broadening beyond the initial focus on model training. As the market matures, demand is growing for full-stack solutions that combine raw computing power with the software and services needed for efficient inference.

According to Baird, companies like DigitalOcean and Nebius are well-positioned to capture demand from enterprises, particularly smaller businesses that may be underserved by larger hyperscalers. Truist acknowledged potential risks for CoreWeave, including new cloud initiatives from major tech firms like Meta and policy constraints on data center development, but believes overall demand for AI compute remains robust.