英伟达预计下季度营收780亿！股价盘后涨1%，AI热潮还没停

Core Insights - Nvidia reported a net profit of $120 billion over the past 12 months, marking a nearly 30-fold increase from $4.4 billion three years ago, a rare achievement in modern tech history [1] - The company holds approximately 90% of the high-end AI chip market, which is crucial for tech giants building data centers [1] - Despite the significant investments in AI, Wall Street is becoming cautious about the sustainability of the AI boom, leading to a stabilization in Nvidia's stock price [1] Financial Performance - In Q4 of fiscal year 2026, Nvidia's revenue grew by 73% year-over-year to $68.1 billion, with profits nearly doubling to approximately $43 billion, or $1.76 per share [3] - The data center business generated $62.3 billion in revenue, exceeding expectations and accounting for over 91% of total sales [3] - Nvidia's total revenue for the year reached $215.9 billion, surpassing previous forecasts, with earnings per share at $4.77 [5] Business Segmentation - The data center business is segmented into computing, graphics chips, CPUs, and networking, with computing revenue up 58% and networking revenue soaring 263% to $11 billion [5] - The increase in networking revenue indicates a trend towards customers purchasing complete cluster systems, suggesting large-scale deployments [5] Customer Concentration - Nvidia's sales concentration has increased, with a few key customers accounting for 36% of sales, up from 34% the previous fiscal year [6] AI Ecosystem and Future Outlook - CEO Jensen Huang emphasized Nvidia's role in the AI ecosystem, stating that the company's platform supports a wide range of AI projects and models, enhancing its competitive edge [7][9] - Nvidia is in advanced negotiations with OpenAI for a potential $1 trillion investment in AI infrastructure, which could be a significant development in the AI landscape [10] - Huang noted that the current year is expected to see a shift from model competition to application explosion, indicating a transition in AI usage from training to large-scale inference [10][11]