事件：NVIDIA FY24Q4 實現收入221億美元，(同比+265%，環比+22%)，超此前200億美元的指引，基本和市場樂觀預期一致。
GAAP淨利潤爲122.85億美元，Non-Gaap淨利潤爲128.39億美元。FY24收入達609億美元，（同比+126%），GAAP淨利潤達297.6億美元。下季度指引收入240億美元。目前PE值爲65.35。
Key results: FY24 revenue leaped 126% on FYQ4’s 265% yoy growth spike
•FY24Q4:
Nvidia generated achieved revenue of USD22.1bn in the quarter ended January 2024, expanding 265% yoy and 22% qoq, and surpassing its USD20bn guidance; it was in line with the market’s more optimistic expectations. GAAP net profit came to USD12.29bn and non-GAAP net profit to USD12.84bn.
•FY24:
revenue for the full year amounted to USD60.9bn (up 126% yoy), with GAAP net profit at USD29.76bn.
•FY25Q1E guidance:
Nvidia guided the next quarter’s revenue at USD24bn (plus or minus 2%) with GAAP/non-GAAP gross margins at 76.3%/77%, suggesting record highs. The stock’s current PE ratio is 65.35.
數據中心業務：
FY24Q4收入達到184億美元（同比+409%，環比+27%），全年收入爲475億美元（同比+217%）。這一增長由生成式AI和大型語言模型在廣泛的行業、用例和地區的訓練和推理驅動。公司數據中心平臺的先進性能和領先性能使得許多用例獲得了高回報投資，包括AI訓練和推理、數據處理和廣泛的CUDA加速工作負載。
Business division breakdown: strong showings from data centers and gaming
Data centers:
FY24Q4 generated revenue of USD18.4bn (up 409% yoy and up 27% qoq) and hefted FY24 revenue by 217% yoy to USD47.5bn. The growth spike was driven by training and inference business from generative AI and large language models (LLM) spanning a wide range of industries, use cases and geographical regions. Nvidia’s market leadership and the advanced performance of its data center platform enable high returns on investment for many use cases, including AI training and inference, data processing and accelerated workload efficiency with its proprietary CUDA architecture.
遊戲業務：
FY24Q4收入爲29億美元（同比+56%），FY24收入爲104.5億美元（同比+15%）。新推出的GeForce RTX™ 40 SUPER系列GPU以及針對1億多RTX AI PC用戶的生成AI能力，推動了業務的增長。
Gaming:
FY24Q4 revenue came to USD2.9bn (up 56% yoy), bringing FY24 revenue to USD10.45bn (up 15% yoy), due to the newly launched GeForce RTX™ 40 Super series of GPUs with generative AI capabilities that target more than 100m RTX AI PC users.
專業可視化業務：
FY24Q4收入達到4.63億美元（同比+105%，環比+11%），全年收入達16億美元（同比+1%）。本季度的環比增長得益於RTX Ada架構GPU的豐富組合，持續增長。
Professional visualization:
FY24Q4 revenue came to USD463m (up 105% yoy and up 11% qoq), with full-year revenue at USD1.6bn (up 1% yoy); qoq growth was due to steady growth generated by the rich portfolio of RTX Ada Lovelace architecture GPUs.
汽車業務：
FY24Q4收入爲2.81億美元（同比-4%，環比+8%），全年收入爲11億美元（同比+21%），首次突破10億美元大關，這得益於汽車製造商持續採用NVIDIA DRIVE平臺。
Autonomous driving:
FY24Q4 revenue arrived at USD281m (down 4% yoy and up 8% qoq), which brought full-year revenue to USD1.1bn (up 21% yoy), breaking past the USD1bn mark for the first time. The key factor was automakers’ continuing adoption of the Nvidia Drive platform.
公司FY25Q1收入指引爲240億美元（+ or -2%），GAAP/Non-GAAP毛利率指引分別爲76.3%和77%。繼續創下歷史新高
我們的觀點：
1) 推理需求強勁！目前已經佔需求40%以上：全球各地的企業、行業和國家對AI的需求正在激增。大型雲服務商、專業GPU提供商、企業軟件公司、消費者互聯網公司、汽車、金融、醫療等各個垂直行業,都在積極擁抱AI。去年,數據中心大約40%的收入來自AI推理,而超過一半的數據中心收入來自大型雲服務提供商。
2) 推薦系統，從CPU爲主導到現在的GPU爲主導
3) 公司認爲客戶對英偉達路線圖反饋非常良好，未來英偉達新產品會有革命性能力提升
4) 下季度：關注GTC大會和B100產品發佈。我們認爲B100有望顯著提高 AI 計算性能，進一步鞏固英偉達的算力優勢
Our takeaways: inference demand spike; GPU takes over; eyes on GTC 2024
•Surging inference demand:
inference currently accounts for more than 40% of AI demand, which is surging as companies, industries and countries embrace AI. They include large cloud service providers, professional GPU providers, enterprise software providers, consumer internet companies, automotive, finance, medical and other vertical industries. AI inference business generated about 40% of data center revenue last year and large cloud service providers generated more than 50%.
•GPUs dominate:
the recommended system has shifted to GPU from CPU.
•Company prospects:
Nvidia reveals it has received positive customer feedback on its future roadmap. We believe its new products will have revolutionary capabilities.
•FY25Q1E focus:
we would monitor data points on Nvidia’s upcoming GTC 2024, its AI developers’ conference that will kick off on 18 March, along with the widely anticipated launch of its B100 GPU. We expect B100 will greatly raise AI computing performance and help Nvidia consolidate its computing performance advantage.
