🐯虎嗅•較早收集於 11m
英偉達日賺22億，全年淨利已超4個騰訊
💡Nvidia $120B profit proves Agentic AI/token boom—scale inference ASAP
⚡ 30-Second TL;DR
有什麼變化
全年營收2159億美元，日淨利3.28億美元；2026雲巨頭資本支出近7000億美元。
為什麼重要
驗證AI運算需求爆發至2027；代理AI推升推理經濟轉變。英偉達透過生態鞏固，對抗中國晶片挑戰者。
下一步行動
Test Blackwell for inference benchmarks as Hopper stock depletes by 2027.
誰應關注：Developers & AI Engineers
關鍵要點
- •全年營收2159億美元，日淨利3.28億美元；2026雲巨頭資本支出近7000億美元。
- •代理式AI：每任務產生數千token，高速推理貨幣化。
- •收購Groq推理技術；接近OpenAI千億基礎設施協議；中國H20銷售停滯。
- •庫存排至2027；Rubin 2026下半年量產供雲端。
🧠 深度解析
背景與延伸：來自公開資料，非原文內容。引用 4 個來源。
🔑 增強重點摘要
- •Nvidia's Q4 net income reached $42.96 billion, up 94% year-over-year, with gross margin at 75%.[2]
- •Nvidia forecasts Q1 fiscal 2027 revenue of $78 billion, exceeding Wall Street expectations of $73 billion, excluding China sales.[2]
- •Grace Blackwell with NVLink delivers an order-of-magnitude lower cost per token for inference compared to prior platforms.[3]
- •Nvidia unveiled the Rubin platform with six new chips promising up to 10x reduction in inference token cost versus Blackwell; AWS, Google Cloud, Microsoft Azure, and Oracle will deploy first.[3]
🛠️ 技術深入
🔮 前景展望AI analysis grounded in cited sources
Nvidia expects $500B revenue visibility from Blackwell and Rubin through end of 2026.
CFO Colette Kress stated visibility to $500 billion in revenue from these offerings from start of 2025 through end of 2026 calendar year.[1]
Total AI infrastructure investment could reach $3-4 trillion annually by 2029-2030.
Nvidia's CFO indicated belief in $3 trillion to $4 trillion annual AI infrastructure spend by 2029 or 2030.[1]
⏳ 時間線
2025-01
Fiscal 2025 ends with $130.5B revenue, more than double prior year.
2026-02
Fiscal 2026 Q4 revenue hits record $68.1B, up 73% YoY; full-year $215.9B.
2026-02
Unveils Rubin platform with six chips for 10x inference cost reduction.
📎 來源 (4)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
📰
AI 週報
閱讀本週精選 AI 大事摘要 →
👉相關動態
AI 策展新聞聚合。所有內容版權歸原始發布者所有。
原始來源： 虎嗅 ↗
每週 AI 簡報
每週一封，可隨時退訂。