The current AI industry chain earnings season is basically wrapped up, and the market’s narrative focus is undergoing a critical shift.

The hardware side remains strong. The performances delivered by NVIDIA and Marvell continue to confirm vigorous demand in the compute-power and high-speed networking connectivity sectors. Marvell’s revenue surged 37% year over year, and it also issued next-quarter guidance that exceeded market expectations—showing that the logic behind strong hardware demand remains solid.

The biggest highlight of this earnings season, however, comes from clear commercialization signals taking shape on the software side. CrowdStrike’s quarterly revenue rose 26%, net new annual recurring revenue jumped 51% to $333 million, and the company also raised its full-year outlook. Salesforce and Okta likewise gained market attention thanks to results and guidance that came in better than expected. AI is no longer just a procurement boom for hardware chips; it is increasingly translating into monetization across enterprise applications and software services.

As this happens, a split is emerging as well. After Synopsys released its earnings, its share price fell under pressure—an unmistakable signal that the capital markets will no longer give blanket, indiscriminate premium valuations to every AI-themed stock. Simply being associated with “AI” is no longer enough to win over investors.

The focus of market discussion has shifted completely. It has moved from the original question of “whether there is actually AI demand” to the next core issue: whether enterprises can turn their earlier, high AI investments into tangible outcomes—new orders, stable recurring revenue, and free cash flow.
#Ai需求从硬件扩散至软件 $OPENAI