Record Data Sheet

Amount
$380B vs $390B
Vertical
Alphabet vs Amazon
Period
Fiscal year 2026

A comparison of capital expenditure, data-center projects, energy infrastructure, construction partnerships and the economics of the AI boom

The AI Infrastructure Spending Race

Google and Amazon are expanding the physical infrastructure required to train and run artificial intelligence systems. The investment extends beyond data-center buildings to servers, networking equipment, accelerators, land, electricity supply, cooling systems and the wider cloud infrastructure required to serve enterprise customers.

In June 2026, Alphabet forecast capital expenditure of $180 billion–$190 billion for the year. The company said the overwhelming majority would go towards technical infrastructure. Amazon had announced an approximately $200 billion company-wide capital expenditure plan for 2026 in February, citing opportunities in AI, chips, robotics and other businesses.

Together, the forecasts imply a combined annual investment program of approximately $380 billion–$390 billion. That is a useful indicator of the scale of the contest, but it is not a precise measure of spending on AI data centers alone. Alphabet’s forecast includes technical infrastructure and other property and equipment, while Amazon’s consolidated capital expenditure also supports its fulfilment network and other operations.

The latest comparable half-year reporting period available for this article ended on June 30, 2026. Alphabet reported $80.6 billion in purchases of property and equipment during the first six months of the year. Amazon reported $96.3 billion in cash capital expenditure over the same period. Amazon’s measure includes certain items such as leasehold improvements, vendor incentives and proceeds from asset sales, so the figures are not perfectly interchangeable.

The figures nevertheless establish the scale of investment already being deployed. The key question is no longer simply how much the companies have announced. It is how quickly they can turn capital into operational computing capacity, secure the energy required to run it and generate returns from their expanding cloud businesses.

Sources: Alphabet investor presentation, June 2026; Alphabet Q2 2026 Form 10-Q; Amazon Q2 2026 Form 10-Q; Amazon Q4 2025 results.

1. The Financial Comparison: Alphabet vs Amazon

The following table separates announced full-year forecasts from reported expenditure.

MeasureAlphabet / GoogleAmazon
Announced 2026 capex forecast$180bn–$190bnApproximately $200bn
Date of cited forecastJune 2026February 2026
First-half 2026 reported measure$80.6bn purchases of property and equipment$96.3bn cash capital expenditure
First-half 2025 comparison$39.6bn purchases of property and equipment$55.6bn cash capital expenditure
Main stated investment driversTechnical infrastructure, including servers, networking, data-center land and buildingsTechnology infrastructure, primarily supporting AWS, plus fulfilment-network capacity
Full-year 2026 forecast subsequently restated in Q2 filing?The June forecast remains the cited guidanceThe Q2 filing did not provide a replacement precise full-year figure

Sources: Alphabet Q2 2026 Form 10-Q; Amazon Q2 2026 Form 10-Q.

What the first-half figures show

Alphabet’s purchases of property and equipment rose from $39.6 billion in the first half of 2025 to $80.6 billion in the first half of 2026. The company said these investments primarily reflected technical infrastructure. Its filing explains that data-center projects are generally multi-year developments: land and buildings are acquired, construction proceeds in stages, and servers and networking equipment are installed as the facilities progress.

Amazon’s cash capital expenditure rose from $55.6 billion to $96.3 billion over the same six-month comparison. Amazon said the spending primarily reflected technology infrastructure, most of which supports AWS, and additional capacity for its fulfilment network. The company also said both categories were expected to increase during 2026.

The comparison has an important limitation: Alphabet’s figure is purchases of property and equipment, whereas Amazon’s is its reported cash capital expenditure measure. The article preserves the companies’ reported definitions rather than implying that the accounting measures are identical.

Future commitments are not current-period expenditure

Alphabet disclosed $85.2 billion in future lease payments for leases primarily related to data centers that had not yet commenced as of June 30, 2026. The leases were expected to commence between 2026 and 2031. These are future contractual commitments, not additional current-period capital expenditure.

Alphabet also reported material purchase commitments and other contractual obligations of $811.0 billion at June 30, 2026, including commitments relating to technical infrastructure, inventory, content licences and energy arrangements. This figure covers a much broader set of contractual obligations and must not be added to the annual capex forecast.

The distinction is essential. A capital expenditure figure measures investment recognized or paid under the relevant accounting definition; a lease commitment or purchase obligation describes a contractual exposure that may be fulfilled over several future periods.

Source: Alphabet Q2 2026 Form 10-Q.

2. Google’s Investment Strategy: Compute, Cloud and Energy

Alphabet’s investment program supports Google Search, YouTube, Google Cloud, Gemini and other products. The company’s infrastructure includes data-center land and buildings, servers, networking equipment and specialised computing systems.

In its June 2026 investor presentation, Alphabet said the overwhelming majority of its $180 billion–$190 billion capital expenditure forecast would be directed towards technical infrastructure. It also said it expected capital expenditure in 2027 to increase significantly compared with 2026.

This is a multi-year expansion program rather than a single construction project. Demand for computing capacity, the time required to build facilities and the need to secure power all influence when announced investment becomes operational capacity.

Google Cloud’s commercial position

Alphabet reported Google Cloud revenue of $24.768 billion in Q2 2026, up 82% year over year. Segment operating income was $8.814 billion. In its earnings commentary, the company highlighted accelerating demand for AI infrastructure and AI solutions.

These figures help explain the economic rationale for expanding capacity. Google Cloud sells computing and related services to businesses that need infrastructure to train models, run applications and process data. The company must invest before all the capacity is generating revenue, while customers may contract for services that are delivered over time.

Revenue and operating income do not prove the return on every new data center. They do, however, provide measurable indicators of the business that the investment is intended to support.

Source: Alphabet’s official Q2 2026 results.

How Alphabet is financing its investment

Alphabet’s Q2 2026 filing reported $49.6 billion in net proceeds from a combination of common stock and mandatory convertible preferred stock issued in June. The company said the proceeds were for general corporate purposes, including capital expenditure to scale AI infrastructure and global computing.

Alphabet also reported $20.3 billion in net proceeds from senior unsecured notes issued during the second quarter for general corporate purposes.

These amounts are sources of corporate financing, not additional capex. They should not be added to the investment forecast as though they represented separate construction spending.

Source: Alphabet Q2 2026 Form 10-Q.

3. Amazon’s Investment Strategy: AWS and the Wider Business

Amazon’s 2026 investment program serves two major needs: expanding AWS infrastructure and supporting the company’s wider retail, logistics and operational network.

In its Q4 2025 results, Amazon announced that it expected approximately $200 billion in capital expenditure during 2026, citing opportunities in AI, chips, robotics and other businesses. That was a company-wide forecast, not an AI-only budget.

Its Q2 2026 filing subsequently reported $96.3 billion in cash capital expenditure during the first half of the year, compared with $55.6 billion in the corresponding period of 2025. Amazon attributed the spending primarily to technology infrastructure, most of which supports AWS, and to additional capacity for its fulfilment network.

Amazon’s second-quarter results also reported AWS sales of $42.2 billion, an increase of 37% year over year, and AWS operating income of $16.6 billion.

The financial results show why Amazon is expanding capacity: AWS is a large, growing business that requires substantial infrastructure. However, the reported figures do not identify exactly how much of the $200 billion forecast is allocated to AI buildings, conventional cloud servers, chips, robotics or logistics facilities.

Sources: Amazon Q4 2025 results; Amazon Q2 2026 results.

Amazon’s $1 billion community commitment

On October 2, 2026, Amazon announced Built Together, a framework for investing in communities where the company operates data centers across the United States. Amazon said it would add more than $1 billion over five years to its existing community activities.

The program includes education, job training, energy affordability, water and energy preservation, and priorities selected with local communities. It is a separate category of announced community investment, not a construction-only budget.

The announcement is relevant to the economics of data-center expansion because large facilities can affect local employment, infrastructure and resource use. But the $1 billion commitment should not be added mechanically to Amazon’s $200 billion capex forecast: the two amounts have different purposes and accounting definitions.

Source: Amazon’s official Built Together announcement and program information.

4. The Projects and Companies Behind the Infrastructure

A data-center project is rarely delivered by the technology company alone. It may involve land development, building contractors, electrical systems, energy generation, grid connections, battery storage, cooling, network equipment and servers.

Public disclosures do not identify every contractor or supplier for every facility. This section therefore names only companies whose roles are confirmed in official announcements. Where a contract value has not been disclosed in the cited source, no amount is assigned.

Google: Meitner Energy Center, Texas

On June 4, 2026, Google and Intersect announced the Meitner Energy Center in Gray and Roberts Counties, Texas. The project combines a data center with more than one gigawatt of wind, solar and battery-storage systems. Google said most of the site’s power would come from clean energy from the start of operations, with on-site gas providing a minority share of demand to support reliability.

The project also includes advanced air cooling designed to limit water use. Google announced plans for an 800-acre Caprock Workforce Hub in nearby Wheeler County, intended to accommodate up to 3,500 construction workers.

Confirmed organizations and roles

  • Google: data-center developer and operator within its wider computing network.
  • Intersect: the original project and energy-infrastructure partner. Alphabet announced an agreement to acquire Intersect in December 2025, and Intersect’s own historical releases state that the acquisition took place in 2026. It should therefore not be described as an independent external partner after the acquisition.
  • Gray and Roberts Counties: local jurisdictions involved in the project’s development and community context.

The official announcement does not provide a complete public breakdown of the project’s total construction cost or a full list of building contractors, electrical contractors and equipment suppliers. The project’s capacity figures are not financial expenditure and cannot be converted into a project cost without further published data.

Sources: Google’s Meitner Energy Center announcement; Intersect’s official announcement; Intersect historical releases.

Google: $50 million for communities in Texas

On October 14, 2026, Intersect announced a $50 million community investment associated with Google and the Meitner and Roman Energy Center communities across Deaf Smith, Gray and Roberts Counties in Texas. The announcement said the funding would establish endowed funds for long-term community development and civic programs, with additional grants and support for local events and organizations. The Amarillo Area Foundation will oversee the funds in partnership with the companies.

This is a disclosed community investment, not the construction cost of either energy center. The cited announcement does not provide a total project capex figure for Meitner or Roman.

Source: Intersect’s official $50 million community investment announcement.

Google: Pine Island, Minnesota — Xcel Energy and Form Energy

On February 24, 2026, Google announced plans for a data center in Pine Island, Minnesota, in collaboration with the city and Xcel Energy.

Google said it would pay all costs associated with its electric service under the agreement. The energy plan includes 1,400 megawatts of additional wind capacity, 200 megawatts of solar and 300 megawatts of iron-air battery storage from Form Energy. Google also committed $50 million to Xcel Energy’s Capacity*Connect program, which is designed to deploy smaller batteries across the utility’s system.

Confirmed organizations and roles

  • Google: data-center customer and participant in the electricity arrangement.
  • Xcel Energy: electricity provider and partner in the grid and energy program.
  • Form Energy: supplier associated with the planned iron-air battery storage.

The 1,400 MW, 200 MW and 300 MW figures describe capacity, not money spent. The $50 million commitment relates to the Capacity*Connect program and is not the total cost of the data center or its complete power system.

Source: Google’s official Pine Island announcement.

Amazon: Contractors and suppliers

Amazon’s public disclosures describe its data-center construction activity, infrastructure needs and community programs. However, the official sources cited in this article do not establish a complete contractor-by-contractor list for its current data-center projects, with the value of each contract.

It would be misleading to name a construction company as a contractor for a specific Amazon facility without an official project announcement or a disclosure from the contractor itself confirming the relationship. The same applies to electrical contractors, cooling specialists, engineering firms and server suppliers.

The absence of a published contractor ledger is itself a limitation for anyone trying to calculate how much of Amazon’s total capex goes to each part of the construction supply chain.

5. What the Consulting Firms Say About the Spending Race

Consulting firms can provide useful context for the wider industry. Their forecasts must remain separate from company-reported expenditure, and published research should not be treated as evidence that a consultancy was hired by Google or Amazon.

Deloitte: Global AI data-center capex

Deloitte’s 2026 technology predictions estimate that global AI data-center capital expenditure could reach $400 billion–$450 billion in 2026. Its analysis estimates that more than half of that total—approximately $250 billion–$300 billion—would go to chips, with the remaining expenditure covering land, construction, power, permitting and related infrastructure.

Deloitte’s estimate is a market-wide forecast, not an audited figure for Google and Amazon. It also includes spending by the wider industry, so it should not be added to the two companies’ capital expenditure forecasts.

The analysis highlights an important distinction: the physical buildings are only part of the investment. Accelerators and other chips can account for a substantial share of the infrastructure budget, while power availability and deployment constraints affect how quickly new capacity can be used.

Source: Deloitte Insights — More compute for AI, not less.

McKinsey: More than $700 billion in hyperscaler capex

McKinsey’s June 25, 2026 analysis says Amazon, Google, Meta and Microsoft are collectively committing more than $700 billion in capital expenditure during 2026, with a substantial majority directed towards AI infrastructure.

That is a four-company industry estimate, not a Google-versus-Amazon total. It is also a broader corporate capex measure rather than a project-by-project construction ledger. The estimate should therefore be cited as market context, not used to infer the exact amount spent by either company on data-center buildings.

McKinsey also describes the investment as extending across data-center construction, accelerator procurement and networking. This supports a broader reading of the contest: the infrastructure race encompasses a supply chain of chips, equipment and energy, not only real estate.

Source: McKinsey — Frontiers of compute: The technologies to reduce AI inference costs.

What the consulting estimates do—and do not—prove

Deloitte’s $400 billion–$450 billion estimate and McKinsey’s more-than-$700 billion estimate are not directly comparable totals. They use different market definitions: Deloitte focuses on AI data-center capex and includes a breakdown of chips versus other infrastructure, while McKinsey discusses the wider capex commitments of four hyperscalers.

Neither estimate should be combined with Alphabet’s and Amazon’s company forecasts. Doing so would double-count spending already included in the companies’ figures.

6. The Economics of the Infrastructure Race

The companies’ disclosed figures point to three related questions.

Can electricity supply keep pace?

A data center requires reliable electricity, and the demand can be large enough to influence generation and grid investment. Google’s Pine Island and Meitner announcements show how the company is addressing this through new generation, storage, utility arrangements and co-location.

The announced capacity figures describe planned physical resources. They do not disclose the total cost of the associated power systems, and the full amount paid by Google under all related agreements is not publicly itemised in the sources cited here.

How much is spent on buildings versus equipment?

Alphabet’s filings describe technical infrastructure as including servers, networking equipment, data-center land and building construction. Amazon’s filings identify technology infrastructure as a primary driver of cash capex, but do not give a complete dollar breakdown between buildings, servers, chips, network equipment and fulfilment facilities.

That means the headline capex forecasts cannot be treated as construction budgets. A rigorous project-level comparison would require disclosures showing the cost of land, buildings, electrical systems, cooling, servers, networking and other equipment separately.

Will the investment generate adequate returns?

Alphabet’s Google Cloud revenue and operating income, and Amazon’s AWS revenue and operating income, provide relevant indicators of commercial performance. But neither company’s reported results establish the return on an individual data center or the future return on the entire capital program.

Investors should track actual expenditure, operational capacity, cloud revenue, operating margins, depreciation and the timing of projects entering service. The useful comparison is not simply which company spends more, but how the investment supports the growth and economics of each business.

7. What Should Be Monitored Next?

The most useful future disclosures will be:

  • Updated full-year capital expenditure guidance from both companies.
  • Actual spending reported in subsequent quarterly filings.
  • Project-specific construction costs, where officially disclosed.
  • Electricity-generation and grid-upgrade commitments associated with named sites.
  • Details of construction contractors, engineering partners and equipment suppliers.
  • Evidence that new facilities are operational and generating revenue.
  • Further disclosures on energy use, water use and community investment.

As of October 24, 2026, the latest company quarterly financial results used in this article are for the quarter ended June 30, 2026. No later quarterly figures have been assumed.

Conclusion: The Race Is About More Than the Headline Budget

Alphabet’s announced 2026 capital expenditure forecast of $180 billion–$190 billion and Amazon’s earlier forecast of approximately $200 billion demonstrate the scale of the AI infrastructure build-out. Their first-half financial filings confirm that investment is already substantial, while project announcements show that data-center development increasingly involves energy generation, storage, utilities and local communities.

The comparison is not a clean contest between two construction budgets. The companies report different measures, operate different businesses and have not publicly itemised all project costs or supplier payments. The combined $380 billion–$390 billion forecast range is therefore a useful indicator of the scale of their announced corporate investment plans, not a precise total for AI construction or final spending.

The next stage of the race will depend on whether each company can convert capital into reliable computing capacity, secure the electricity required to operate it and generate returns from the cloud and AI services that use the infrastructure.

For investors, the question is how effectively capital expenditure supports future revenue and profitability. For local communities, it is how the costs and benefits of new facilities are distributed. For the technology sector, it is whether power, equipment and construction supply chains can keep pace with the demand for AI computing.

Frequently Asked Questions

How much are Google and Amazon investing in 2026?

Alphabet forecast $180 billion–$190 billion in company-wide capital expenditure in June 2026. Amazon announced an approximately $200 billion company-wide capex forecast in February 2026. The combined range is $380 billion–$390 billion, but it combines forecasts announced at different dates and is not a construction-only total.

How much did Google spend on capital expenditure in the first half of 2026?

Alphabet reported $80.6 billion in purchases of property and equipment during the six months ended June 30, 2026. Its filing says these purchases primarily reflected technical infrastructure.

How much did Amazon spend on capital expenditure in the first half of 2026?

Amazon reported $96.3 billion in cash capital expenditure for the six months ended June 30, 2026. The company said the spending primarily reflected technology infrastructure, most of which supports AWS, and additional capacity for its fulfilment network.

Which companies are involved in Google’s data-center projects?

Google’s publicly announced projects include Intersect in the Meitner Energy Center in Texas, and Xcel Energy and Form Energy in the Pine Island, Minnesota energy arrangements. Alphabet acquired Intersect in 2026, so it should not be described as an independent external partner following the acquisition. The public announcements do not provide a complete contractor list for every project.

How much is Amazon investing in data-center communities?

Amazon announced more than $1 billion in additional community investment over five years in October 2026 through its Built Together program. The commitment covers community priorities including education, job training, energy affordability and water preservation. It is distinct from the company’s wider capital expenditure forecast.

What do Deloitte and McKinsey estimate for AI infrastructure spending?

Deloitte estimates global AI data-center capex of $400 billion–$450 billion in 2026, with more than half directed towards chips. McKinsey says Amazon, Google, Meta and Microsoft together are committing more than $700 billion in 2026 capex. These estimates have different scopes and should not be added together or treated as company-reported expenditure.

Official Sources

Company financial disclosures

  1. Alphabet investor presentation, June 2026
  2. Alphabet Q2 2026 Form 10-Q
  3. Alphabet Q2 2026 results
  4. Amazon Q4 2025 results
  5. Amazon Q2 2026 results
  6. Amazon Q2 2026 Form 10-Q
  7. Amazon Built Together program

Project and energy disclosures

  1. Google — Meitner Energy Center, Texas
  2. Intersect — Meitner project announcement
  3. Intersect — historical releases and acquisition status
  4. Intersect — $50 million Texas community investment
  5. Google — Pine Island, Minnesota

Consulting-firm research

  1. Deloitte — More compute for AI, not less
  2. McKinsey — Frontiers of compute: The technologies to reduce AI inference costs