NVDA / NASDAQ
NVIDIA: The AI Infrastructure Company
追踪全球 AI 基础设施投资周期与 NVIDIA 长期价值创造。
投资逻辑
NVIDIA 是全球 AI 基础设施投资周期的核心受益者：AI Capex 转化为 GPU、Networking 与 AI Factory 平台需求，并最终映射到收入、自由现金流和股东价值。
Aben AI Capex Index™
AI Capex is the lead indicator
Microsoft + Amazon + Alphabet + Meta + Oracle + OpenAI + xAI + CoreWeave
2026Q1
本季度总 AI Capex
$111.6B
Disclosed proxy total
YoY Growth
+70.6%
Index history sample
QoQ Growth
+13.1%
Index history sample
Index uses disclosed capital expenditure / infrastructure-spending proxies where AI-only capex is not separately disclosed. Null means no comparable AI-only or capex proxy is currently disclosed in a quarterly source.
公司贡献占比
Amazon
$34.2B / 30.6%
Capital expenditures proxy; AWS AI infrastructure demand disclosed separately through chip and GPU deployment commentary.
Microsoft
$31.9B / 28.6%
Capital expenditures proxy; management linked spending to cloud and AI infrastructure. AI-only capex not separately disclosed.
Meta
$21.4B / 19.2%
Capital expenditures proxy; AI infrastructure is a disclosed driver, but AI-only capex is not separately disclosed.
Alphabet
$17.2B / 15.4%
Capital expenditures proxy; AI-only capex not separately disclosed.
Oracle
$6.9B / 6.2%
Capital expenditures proxy for cloud infrastructure buildout; AI-only capex not separately disclosed.
Investment Thesis
NVIDIA 通过 GPU、CUDA、Networking 和 AI Factory 平台，捕获全球 AI 基础设施投资中的大部分利润池。
01
AI Demand
02
AI Capex
03
GPU Demand
04
Data Center Revenue
05
Free Cash Flow
06
Shareholder Value
Mission Dashboard
Capex to cash flow
Q1 FY2027
Data Center Dashboard
Where capex becomes revenue
Q1 FY2027
Data Center Revenue
$75.2B
USD billions
Data Center Revenue CAGR
82.4%
2-year CAGR from Q1 FY2025
Data Center Revenue %
92.1%
percent of revenue
Networking Revenue
$14.8B
USD billions
Compute Revenue
$60.4B
USD billions
Data Center Revenue Trend
Acceleration Monitor
斜率变化 / second derivative proxy
How to read acceleration
斜率 = 本季度 Data Center revenue 相比上一季度新增的收入，单位为 $B / quarter。二阶 = 本次新增额相比上次新增额的变化；大于 0 表示增长加速，小于 0 表示增长放缓。
Q4 FY2025
Q1 FY2026
Q2 FY2026
Q3 FY2026
Q4 FY2026
Q1 FY2027
Product Roadmap
Platform cadence
01 / 2022-2024
Hopper
Installed base
H100/H200 generation that scaled the first major generative AI training wave.
02 / 2024-2025
Blackwell
Ramp
B200/GB200 generation for training and inference AI factories.
03 / 2025-2026
Blackwell Ultra
Current cycle
Higher-performance Blackwell platform aimed at the next step-up in AI factory throughput.
04 / 2026-2027
Rubin
Announced
Vera Rubin platform for agentic AI factories and future hyperscale systems.
05 / 2027+
Rubin Ultra
Roadmap
Next Rubin generation to monitor for performance, networking and system-level economics.
Moat Analysis
Why the profit pool concentrates
CUDA Ecosystem
CUDA turns NVIDIA GPUs into a developer platform rather than a replaceable component.
Evidence: Model frameworks, libraries and enterprise AI workflows are deeply optimized around NVIDIA software and GPUs.
Monitor: CUDA / CUDA-X adoption, framework optimization and developer tooling updates
Networking Platform
NVLink, InfiniBand and Ethernet networking make scale-out AI clusters perform as systems.
Evidence: Q1 FY2027 Data Center networking revenue reached $14.8B, up 199% year over year.
Monitor: Networking revenue growth and attach rate
AI Factory Stack
NVIDIA sells a platform for AI factories: compute, networking, software and reference architectures.
Evidence: Management shifted reporting toward Data Center sub-markets of Hyperscale and ACIE.
Monitor: Hyperscale / ACIE mix and AI factory deployment commentary
Scale Advantage
Large-scale supply, roadmap cadence and ecosystem breadth reinforce customer standardization.
Evidence: Q1 FY2027 Data Center revenue reached $75.2B, with compute revenue of $60.4B.
Monitor: Data Center revenue, backlog commentary and product-cycle transition speed
Risk Dashboard
What can break the mapping
AI Capex Slowdown
Hyperscaler 与 AI Cloud 资本开支放缓会先影响订单能见度，再影响 Data Center revenue 增速。
Monitor: Aben AI Capex Index QoQ / Hyperscaler capex guidance
China Restrictions
出口限制可能压缩特定 GPU SKU 的收入机会，并造成库存或费用扰动。
Monitor: China data center compute revenue disclosure / export-control language
AMD Competition
MI 系列 GPU 若在训练或推理工作负载中获得规模采用，可能压低 NVIDIA 定价权。
Monitor: AMD data center GPU revenue / hyperscaler deployment comments
Custom ASIC Growth
TPU、Trainium 与其他客户自研 ASIC 会分流部分内部工作负载。
Monitor: Custom silicon deployment volume / cloud provider AI workload mix
Margin Compression
组件成本、系统集成复杂度或竞争加剧可能压缩毛利率与经营杠杆。
Monitor: Gross margin / operating margin / HBM supply pricing
Customer Concentration
少数大型云与 AI 平台客户贡献高，订单节奏变化会放大季度波动。
Monitor: Top customer concentration language / hyperscaler capex cadence
Valuation Dashboard
No DCF theater
Daily market refresh through the existing portfolio workflow. Values remain market-price dependent.
PE
As of 2026-07-23
31.63x
Trailing PE ratio from Yahoo Finance. Market-price dependent; updated 2026-07-23.
Forward PE
As of 2026-07-23
16.07x
Forward PE ratio from Yahoo Finance analyst estimates. Market-price dependent; updated 2026-07-23.
EV/FCF
As of 2026-07-23
107.16x
Enterprise value divided by trailing free cash flow, calculated from Yahoo Finance fields. Market-price dependent; updated 2026-07-23.
FCF Yield
As of 2026-07-23
0.92%
Trailing free cash flow divided by market capitalization, calculated from Yahoo Finance fields. Market-price dependent; updated 2026-07-23.
Quarterly Change Log
What changed this quarter
Q1 FY2027
Revenue reached $81.6B and Data Center revenue reached $75.2B; Data Center compute was $60.4B and networking was $14.8B under the prior sub-market disclosure.
Thesis change: No manual thesis change recorded. The quarter strengthens the AI Capex Cycle monitoring framework.