Morgan Stanley has dismissed concerns about lower AI spending in 2025, asserting continued strong demand for chips. Analyst Joseph Moore highlighted that the need for more inference chips is driving a wave of very strong demand, countering the idea of an AI digestion phase.

Moore pointed to recent comments from OpenAI's Sam Altman and Alphabet's Sundar Pichai, indicating AI companies still face challenges in obtaining enough GPU chips. He noted that the focus in Silicon Valley has shifted to managing growth in tokens generated, which has increased more than fivefold since the beginning of the year, stressing the ecosystem and driving investment to handle these workloads.

AI stocks have faced weakness this year, initially triggered by DeepSeek's efficient large language model debut in late January, which raised concerns about reduced GPU chip needs from Nvidia. Additional pressures came from President Donald Trump's tariffs in early April. Nvidia's shares have dropped 28% since late January, and major tech companies tied to AI have fallen about 21% from their recent peaks.

Regarding Nvidia, Moore acknowledged that supply constraints and export restrictions on H20 chips might limit revenue upside in the coming quarters. However, he anticipates significant growth in 2026 once these supply issues are resolved. Moore stated Nvidia had minimal revenue for Blackwell in October, $11 billion in January, and likely over $30 billion in the current quarter, projecting continued growth.

Moore increased his calendar year 2026 revenue and earnings per share estimates for Nvidia by 10.7% and 11.9%, respectively, reaffirming the chip giant as a "top pick." He reiterated an "Overweight" rating on Nvidia with a $160 price target, suggesting a potential 45% increase from current levels.