Upcoming earnings from major hyperscalers are expected to set the tone for near-term sentiment as the market weighs potential changes in artificial intelligence infrastructure spending, Oppenheimer said in a note emailed Tuesday.

Several AI infrastructure-focused stocks fell sharply on Monday as investors responded to China-based startup DeepSeek's faster-performing models and reports of lower training costs, the analysts noted.

Vertiv Holdings (VRT), Modine Manufacturing (MOD), GE Vernova (GEV) and Bloom Energy (BE) witnessed major drops, while American Superconductor (AMSC), Johnson Controls International (JCI), Trane Technologies (TT), Carrier Global (CARR) and Cummins (CMI) also faced pressure, the investment firm said.

Industry orders and price softening are expected in fiscal 2025, while evolving AI capabilities could introduce both fresh uncertainties and long-term growth prospects, according to the note.

DeepSeek's recently launched V3 and R1 models suggest training costs could be significantly reduced. However, there are some doubts on whether headline figures reflect full research and development expenses, according to Oppenheimer.

If training becomes cheaper, the resulting higher AI usage could still boost data center investments, as lower costs may fuel better returns on infrastructure deployments, Oppenheimer noted.

Geopolitical considerations remain pivotal, particularly amid US export controls on high-compute chips, which could spur further policy responses aimed at reclaiming leadership in AI technology.

Investors are closely monitoring forthcoming earnings from Meta (META) and Microsoft (MSFT) for any signals on AI roadmap adjustments and whether spending plans shift to H2 or 2026, Oppenheimer said.

Price: 106.11, Change: +3.51, Percent Change: +3.42