AI Capex Turns Contracted: Big Tech Has $3.15 Trillion in Commitments

AI Infrastructure • Data Analysis

AI Capex Turns Contracted: Big Tech Has $3.15 Trillion in Commitments

The AI buildout is increasingly moving from planned spending toward contracted capacity, with Google, Meta, Microsoft, Amazon and Oracle showing roughly $3.15 trillion in combined reported commitments.

5 Hyperscalers $3.15T Commitments AI Infrastructure Multi-Year Buildout
AI infrastructure and Big Tech AI capex commitments
THE BIG PICTURE

The AI infrastructure boom is increasingly moving from planned spending toward contracted capacity.

AI Capex Commitment Dashboard
Based on the supplied five-company dataset
01 $3.15T Combined reported commitments across the five-company dataset
02 $946B Largest reported Total Commitments figure: Alphabet
03 +116% Largest Q/Q increase in Total Commitments: Alphabet
04 $1.16T Combined future leases shown in the supplied table
01
THE THESIS

AI Capex Is Becoming a Contracted Infrastructure Cycle

The next phase of the AI infrastructure boom may be less about what hyperscalers plan to spend and more about what they have already committed to secure.

Across Amazon, Alphabet, Meta, Microsoft and Oracle, the supplied dataset shows $3.151 trillion in combined Total Commitments.

The figure is large enough to change how the AI infrastructure cycle should be viewed: not simply as a sequence of quarterly capital-spending decisions, but as a buildout that increasingly involves purchases, current leases, future leases and other contractual arrangements.

DATA INTEGRITY

$3.15 trillion is not $3.15 trillion of AI capex already spent. It is the combined Total Commitments figure from the five-company dataset.

That distinction matters because the table combines several different categories. Some commitments relate directly to infrastructure, while others can cover broader procurement, leasing, financing or related arrangements.

02
THE DATASET

The Five Hyperscalers Have $3.15 Trillion in Reported Commitments

The table below recreates the supplied company-report dataset. It preserves the reported categories, Q/Q changes and company totals. All figures are shown in billions of U.S. dollars.

Hyperscaler AI infrastructure commitments table
Hyperscaler Purchases & Lease Commitments USD $bn
USD $bn AMZN GOOGL META MSFT ORCL
Purchases $130 $811 $349 $194 $32
Q/Q Change 25% 144% 47% 76%
Current Leases $133 $21 $35 $114 $53
Q/Q Change 5% 12% 32% 5% 41%
Future Leases $137 $85 $347 $329 $260
Q/Q Change 29% 13% 90% 67% 0%
Other Financings $11 $8 $0 $35 $3
Q/Q Change 9% -16% 8% 50%
Pending AI Stakes $41 $22 $0 $0 $0
Q/Q Change -25% 0%
Total Commitments $453 $946 $731 $672 $349
Q/Q Change 13% 116% 59% 50% 16%

Source: Company Reports

Note: META Current Lease Commitments & ORCL Purchase Commitments reflect values as of FY25 and FY26, respectively.
Methodology: The $3.15 trillion figure is the sum of the five reported Total Commitments figures shown above. Because the dataset combines different commitment categories and reporting periods, the combined figure should not be interpreted as $3.15 trillion of AI capex already spent.
03
COMPANY BREAKDOWN

Who Is Carrying the Largest Commitment Load?

Alphabet / Google

#1
$946B +116% Q/Q

Alphabet has the largest reported Total Commitments figure in the dataset. Its $811 billion in purchases also represents the largest purchase figure among the five companies.

Meta

#2
$731B +59% Q/Q

Meta's commitment stack is particularly notable for its $347 billion in future leases, which are up 90% quarter over quarter in the supplied dataset.

Microsoft

#3
$672B +50% Q/Q

Microsoft's future leases reach $329 billion in the supplied table, making future infrastructure capacity a major component of its commitment profile.

Amazon

#4
$453B +13% Q/Q

Amazon's commitment mix includes purchases, current leases, future leases, other financing and pending AI stakes.

Oracle

#5
$349B +16% Q/Q

Oracle's $260 billion in future leases stand out as the largest component of its commitment mix.

04
RANKING

Alphabet Leads the Five-Company Commitment Race

Total Commitments USD billions · supplied dataset
Alphabet
$946B
Meta
$731B
Microsoft
$672B
Amazon
$453B
Oracle
$349B

Alphabet leads the five-company comparison, followed by Meta and Microsoft. But the ranking alone does not tell the entire story. The composition of each company's commitment stack is equally important.

05
FORWARD CAPACITY

Future Leases Put $1.16 Trillion of Capacity in Focus

One of the clearest signals in the supplied table is the scale of future leases.

Across the five companies, the supplied figures add up to approximately $1.158 trillion.

AI infrastructure capacity and future lease commitments
$347B Meta future leases
$329B Microsoft future leases
$260B Oracle future leases
Future Lease Commitments USD billions
Meta
$347B
Microsoft
$329B
Oracle
$260B
Amazon
$137B
Google
$85B

Future leases matter because they represent infrastructure capacity arranged for future periods rather than capacity that is already fully operating today.

That makes them an important indicator of the expected duration and scale of the infrastructure buildout, but not a measure of cash already spent.

06
MOMENTUM

Alphabet Has the Fastest Q/Q Increase in the Dataset

Total Commitment Q/Q Change Percentage change
Google
+116%
Meta
+59%
Microsoft
+50%
Oracle
+16%
Amazon
+13%

The Q/Q figures show that the commitment cycle is not moving at the same pace across all five companies. Alphabet's reported increase is by far the largest in the supplied dataset, while Meta and Microsoft also show significant acceleration.

07
COMPANY CONTEXT

What the Individual Commitment Mixes Tell Us

Alphabet: The Largest Purchase Commitment

Alphabet's $946 billion Total Commitments lead the dataset. Its $811 billion in purchases also represents the largest purchase figure among the five companies.

Alphabet's June 2026 filing separately reported $85.2 billion of future lease payments for leases that had not yet commenced, primarily related to data centers.

Meta: Future Capacity Stands Out

Meta ranks second at $731 billion, with total commitments up 59% Q/Q. Its $347 billion in future leases, up 90% Q/Q in the supplied table, make future capacity a major part of its commitment profile.

Microsoft: Large Forward Lease Exposure

Microsoft reports $672 billion in Total Commitments, up 50% Q/Q. Future leases reach $329 billion in the supplied table, up 67% Q/Q.

Amazon: A More Distributed Commitment Mix

Amazon's $453 billion Total Commitments include purchases, current leases, future leases, other financings and pending AI stakes.

Oracle: Future Leases Dominate the Mix

Oracle reports $349 billion in Total Commitments, up 16% Q/Q. Its $260 billion in future leases represent the largest component of its commitment mix.

08
INFRASTRUCTURE ARCHITECTURE

What Does “Contracted AI Buildout” Actually Mean?

The phrase does not mean that every dollar in the $3.15 trillion figure is an irrevocable AI-only expense. Instead, it describes a broader shift in how infrastructure capacity is being secured.

01
AI Demand More models, agents and workloads
02
Compute More accelerators and servers
03
Infrastructure Data centers, networking and power
04
Commitments Purchases, leases and financing
05
Capacity Multi-year infrastructure pipeline

AI infrastructure is unusually time-intensive to build. Data centers need land, power, cooling, networking and equipment, while the hardware itself can require long procurement cycles.

Securing capacity ahead of demand can therefore reduce the risk of being unable to deploy compute when customers or internal AI workloads need it.

This is why the supply chain increasingly resembles a multi-year infrastructure buildout rather than a simple quarter-by-quarter spending cycle.

09
IMPORTANT CONTEXT

$3.15 Trillion Does Not Mean $3.15 Trillion of AI Capex

READ THIS BEFORE COMPARING THE NUMBER WITH ANNUAL CAPEX
Committed ≠ Spent ≠ Current Liability ≠ AI-only Capex
$3.15T = combined reported commitments in this dataset

The $3.15 trillion number combines purchases, current leases, future leases, other financings and pending AI stakes. These categories do not represent one identical accounting measure.

Some future obligations may relate to infrastructure that has not yet started operating. Others may cover broader technical infrastructure or procurement rather than AI specifically.

This distinction is particularly important for readers comparing the commitment number with annual capital expenditure. A future lease is not the same thing as cash already paid for a completed data center.

The right interpretation is therefore not “Big Tech has already spent $3.15 trillion on AI.” The stronger interpretation is that the five-company dataset shows an enormous pool of future-facing infrastructure and contractual commitments surrounding the AI buildout.

10
THE ECOSYSTEM

What This Means for the AI Infrastructure Supply Chain

The AI infrastructure race is bigger than GPUs. Compute capacity requires an entire physical and energy ecosystem.

01 Compute Accelerators, GPUs, servers and related hardware.
02 Data Centers Facilities capable of housing increasingly dense compute clusters.
03 Power Electricity generation, transmission and long-term supply.
04 Networking High-speed interconnects and data-center networking infrastructure.
05 Cooling Thermal management for increasingly dense computing environments.
06 Construction Buildings, electrical systems and supporting physical infrastructure.

The key point is not that every supplier in these categories will automatically benefit. Rather, the scale of hyperscaler commitments shows why AI infrastructure is becoming an ecosystem-wide investment cycle.

11
LONG-DURATION BUILDOUT

The AI Buildout Is Becoming a Multi-Year Infrastructure Cycle

The infrastructure story is increasingly stretching beyond a single year's spending plan. The combination of purchases, current leases and future leases points toward capacity being arranged over longer periods.

Multi-year AI infrastructure buildout across compute power data centers networking and cooling
01 Large Purchase Commitments Alphabet alone shows $811 billion in purchases in the supplied table.
02 Large Future Lease Commitments Meta, Microsoft and Oracle each show more than $250 billion in future leases.
03 Rapid Q/Q Growth Alphabet, Meta and Microsoft show 116%, 59% and 50% increases respectively.

Taken together, these signals suggest that AI infrastructure is increasingly being planned and secured over multiple years.

12
RISK FRAMEWORK

The Commitment Boom Also Creates New Risks

Large commitments can make an infrastructure cycle more durable, but they can also create risks if AI demand, economics or technology change faster than physical infrastructure can be adjusted.

Demand Risk If AI workloads grow more slowly than expected, some capacity could take longer to reach high utilization.
Financing Risk Higher financing costs or weaker operating cash flow can make a large future commitment stack more difficult to manage.
Power & Construction Risk Power availability, grid connections, construction schedules and cooling infrastructure can become bottlenecks.
Technology Risk AI hardware and model architectures evolve quickly, creating a potential mismatch between today's infrastructure assumptions and tomorrow's technology.
13
INVESTOR WATCHLIST

What to Watch Next

The next phase of the AI infrastructure cycle will be easier to judge by watching the commitment pipeline alongside actual spending, utilization and revenue.

01
Future leases Are hyperscalers continuing to secure capacity years ahead?
02
Purchase commitments Are procurement commitments accelerating or stabilizing?
03
AI revenue Is infrastructure investment translating into meaningful monetization?
04
Utilization Are newly deployed facilities being used efficiently?
05
Financing How are companies funding the infrastructure cycle?
06
Power Can electricity supply and grid infrastructure keep pace with data-center demand?
14
QUICK ANSWERS

Frequently Asked Questions

AI capex generally refers to capital investment in infrastructure used to support AI workloads, including data centers, servers, networking and related computing infrastructure. Not every commitment in this article is AI-only capex.
The five reported Total Commitments figures in the supplied dataset add up to $3.151 trillion, rounded to approximately $3.15 trillion.
Alphabet leads the supplied table with $946 billion in Total Commitments.
No. The figure combines purchases, leases, financing and related commitments. It should not be interpreted as $3.15 trillion of AI capex already spent.
Future leases indicate infrastructure capacity arranged for future periods. They provide a useful signal about the scale and duration of the infrastructure pipeline, although they are not the same as cash already spent or currently operating capacity.
INDIA AI TRENDS — BOTTOM LINE

The AI Buildout Is Moving From Plans Toward Contracts

The five-company dataset shows approximately $3.15 trillion of combined reported commitments across purchases, leases, financing and related AI commitments.

The most important takeaway is not that $3.15 trillion has already been spent. It is that hyperscalers are increasingly securing future infrastructure capacity through contracts and long-duration arrangements.

That makes the AI buildout look increasingly like a multi-year contracted infrastructure cycle—one where compute, data centers, power, networking and financing all have to scale together.

The next question for the market is whether AI revenue, utilization and economics can grow fast enough to justify the infrastructure commitments being made today.

0 Comments

Post a Comment

Post a Comment (0)

Previous Post Next Post