C
ChaoBro

Five Tech Giants AI CapEx $805 Billion in 2026, $1.1 Trillion in 2027: WTS Raises Forecast

Five Tech Giants AI CapEx $805 Billion in 2026, $1.1 Trillion in 2027: WTS Raises Forecast

Conclusion: AI Infrastructure Investment Enters “Trillion-Dollar” Competition

WTS (Wall Street Technology Services) latest report again raised AI capital expenditure forecasts for five hyperscaler tech companies (Amazon, Alphabet, Meta, Microsoft, Apple):

  • 2026: $805 billion
  • 2027: $1.1 trillion

To understand the scale of this number: AI capital expenditure in 2026 alone roughly equals the combined capital expenditure of all non-tech companies in the S&P 500 in 2025. AI has moved from a tech industry investment theme to the investment focus of the entire economy.

Number Comparison: Exponential Growth in AI Investment

YearFive Giants AI CapExYoY GrowthKey Driver
2024~$200 billionBaselineInitial AI infrastructure
2025~$410 billion+105%Training cluster expansion
2026E~$805 billion+96%Inference + training dual drive
2027E~$1.1 trillion+37%Scaled inference deployment

Key trends:

  1. Growth rate still astonishing: Nearly doubling from 2025 to 2026, but 2027 growth slows to 37%
  2. Inflection point signal: 37% growth rate, while still high, suggests investment peak may be approaching
  3. Inference share rising: Shift from “buying GPUs for training” to “building inference infrastructure”

Company Breakdown: Who Is Leading?

Based on Q1 earnings and forward guidance estimates:

Company2026E CapExQ1 ActualCore InvestmentStrategic Logic
Meta~$225 billion~$30 billionLlama open-source ecosystem + computeOpen-source AI standard setting
Microsoft~$190 billion~$38 billionAzure AI + OpenAICloud AI platform
Google~$160 billion~$32 billionTPU + data centersOwn chips + own cloud
Amazon~$150 billion~$30 billionAWS AI servicesEcosystem + investment dual line
Apple~$80 billion~$10 billionEdge AI + chipsDevice differentiation

Meta highest budget reflects its comprehensive bet on open-source AI—not just model training, but also building infrastructure for the entire Llama ecosystem.

Morgan Stanley “Shock” Warning

Aligned with WTS capital expenditure raise, Morgan Stanley recently warned its clients: AI lab executives told investors to prepare for capability breakthroughs that will “shock” them.

Combined with capital expenditure data, the logical chain of this warning is:

  1. $805 billion investment → Massive compute expansion
  2. Compute expansion → Larger-scale training and inference
  3. Larger scale → Capability leap
  4. Capability leap → Fundamental business model change

Morgan Stanley wording is notable—not “gradual improvement,” but “shock.” This means they expect not 10-20% performance improvements, but capability breakthroughs that could change industry dynamics.

Three Stages of Investment Logic

Based on capital expenditure data, AI investment can be divided into three stages:

Stage 1: Compute Construction Phase (2023-2025) ✅ Basically Complete

  • GPU procurement, data center construction
  • Core driver: Training large models
  • Main winners: NVIDIA, TSMC, cloud vendors

Stage 2: Inference Deployment Phase (2025-2027) 🔄 In Progress

  • Inference optimization, edge deployment
  • Core driver: AI application scaling
  • Main winners: Inference optimization companies, AI application platforms

Stage 3: Application Explosion Phase (2026-2028) 🚀 About to Begin

  • Agent economy, AI-native applications
  • Core driver: AI creating direct revenue
  • Main winners: AI-native application companies

We are in the middle of Stage 2. This means:

  • Compute investment shifting from “buying GPUs” to “optimizing inference efficiency”
  • Training costs starting to decline, inference costs becoming core competitive point
  • AI application layer about to experience real explosion

Risks and Warnings

  1. Investment return uncertainty: $805 billion investment needs to generate corresponding revenue growth; AI direct revenue still far below investment
  2. Valuation bubble risk: Some AI-related company valuations have priced in overly optimistic growth expectations
  3. Regulatory uncertainty: EU AI Act and other regulations may increase compliance costs
  4. Technical route risk: If next-generation model architecture disrupts current paradigm, some infrastructure investment may depreciate

Action Recommendations

  • Infrastructure investors: Focus on inference optimization, edge computing, AI safety—next-stage beneficiary directions
  • Entrepreneurs: While giants invest in infrastructure, focus on application layer innovation
  • Enterprise decision-makers: Evaluate build vs. buy AI infrastructure cost-benefit
  • Individual investors: Watch AI sector valuation risk, focus on companies with actual revenue

$805 billion is not the endpoint, but the starting point of AI becoming economic infrastructure. But smart investors have already shifted from “who invests most” to “who invests most effectively.”