Yield Theory
Subscribe

GOOGL · AI infrastructure profile

Google AI Capex in 2026: $175B–$185B Guide

Track Alphabet and Google's $175 billion to $185 billion 2026 capex plan, server and data-center mix, Cloud allocation, depreciation, and investor signals.

Calendar 2026 guidance

$175B–$185B

Companywide capital expenditure. It is not a disclosed pure-AI total.

Reviewed
Evidence
2 external references · Method
2026 guidance
$175B–$185B
Calendar 2026
2025 capex baseline
$91.4B
About 60% servers and 40% data centers and networking
Sources
2
Primary company materials
Reviewed
2026-08-10
Refresh after material guidance changes

Alphabet 2026 capex guidance

Alphabet expects $175 billion to $185 billion of capital expenditure in 2026, nearly doubling the $91.4 billion invested in 2025 at the midpoint. Management says the mix should remain similar to 2025: approximately 60% servers and 40% data centers and networking equipment. The spending supports Google Cloud, DeepMind, Search, advertising systems and other businesses.

What the number includes

Alphabet defines capital investment primarily around technical infrastructure and office facilities. The timing of supplier payments can move reported quarterly capex, and the company does not isolate one audited AI-capex line. Management expects just over half of 2026 machine-learning compute to be allocated to Google Cloud, but that allocation is a compute measure rather than a direct capex split.

  • Servers and machine-learning compute for Google Cloud, DeepMind and first-party services.
  • Long-duration data centers and networking equipment needed to deliver that compute.
  • Infrastructure supporting Search, YouTube, advertising systems and selected Other Bets alongside external Cloud demand.

How investors should read the spending

Alphabet's disclosed 60/40 infrastructure mix creates a useful bridge between near-term server deployment and long-lived construction. Servers may translate into capacity and depreciation sooner, while buildings and networks create a longer placement-in-service schedule. Investors should compare Cloud growth and backlog with the accelerating depreciation and data-center operating costs management has already flagged.

What to monitor next

  • Quarterly capex cadence relative to the $175 billion to $185 billion full-year range.
  • The server versus data-center and networking mix as infrastructure ramps.
  • Google Cloud growth, backlog and evidence that new capacity is relieving supply constraints.
  • Depreciation and data-center operating-cost growth across Alphabet's income statement.

Comparison limits

  • Alphabet does not publish a clean AI-only capital-expenditure measure.
  • The disclosed compute allocation is not the same as a dollar capex allocation.
  • Payment timing and the placement-in-service schedule can make quarterly capex and depreciation move differently.

Primary sources

Guidance can change each quarter. Yield Theory preserves the company's accounting definition and does not relabel the entire amount as AI spending. Read the research methodology for our sourcing and correction standards.

Compare hyperscaler capex

Microsoft AI capexAbout $190BAmazon AI capexAbout $200BMeta AI capex$130B–$145B

Continue the analysis

2026 hyperscaler AI capex trackerCompare the four-company spending envelopeAI capex depreciation return modelModel the revenue and return required by a capex cycleThe AI depreciation clockTrace when infrastructure spending reaches earningsAlphabet stock researchPut the infrastructure plan in company context

The capex number is the start of the thesis.

Members get the source-linked investment case, catalysts, risks, and evidence that would change the view.

Unlock every issue →

$15/mo or $150/yr · cancel future renewals anytime · sources included