Which big tech company has spent the most in AI?
Every funding round, every investor, and who is really making money from AI infrastructure, in one document. Have a look ↓
Summary
Amazon is currently the best answer to “Which big tech company has spent the most in AI?” It has the largest 2026 spending plan, the largest cash bill in the latest reported quarter and the largest total in the latest completed year.
On October 29, 2026, Amazon confirmed it still expects about $220 billion of cash capital expenditure in 2026, a figure it had raised from about $200 billion in July. Alphabet kept its $195 billion to $205 billion range a day earlier. A $20 billion bump in one revision shows how fast plans move; we track how the plans moved this year in our AI infrastructure report.
Amazon also paid out the most cash in July–September 2026, the latest quarter all four companies have reported: $77.3 billion of property and equipment purchases, more than double a year earlier. Alphabet followed at $51.4 billion.
The caveat is what sits inside the number. Amazon’s budget still covers warehouses, logistics and satellites, even if it says most of it goes to AWS. Alphabet says the vast majority of its spending supports AI, so on a strict AI-only basis the two are close and no public filing settles it.
Microsoft is the hardest to place. It doubled its cash spending to $39.4 billion in the September quarter, but its October results give no full calendar-2026 figure that compares with the others.
Meta is making the biggest bet relative to its size. In October it narrowed its 2026 plan to $135 billion to $145 billion, about 70% of its 2025 revenue, and its free cash flow turned negative, at minus $2.1 billion, in the latest quarter.
The three companies with a full-year plan, Amazon, Alphabet and Meta, total about $560 billion at their midpoints. Adding Microsoft’s recent cash pace puts the four near $700 billion, about double what they spent in their latest completed years.
Strategic stakes in OpenAI, Anthropic and Scale AI do not decide the ranking. Anthropic’s October S-1 shows at least $110 billion of future cloud purchases from Amazon and $114.5 billion from Google, but those are a customer’s bills, not the providers’ capital spending.
Returns are split. AWS earns the most cloud profit, Google Cloud doubled its revenue, Microsoft’s Azure grew 48% and Meta earns its return through advertising.
The clean conclusion: Amazon leads on every broad measure, while Alphabet has the most clearly AI-heavy budget. Amazon and Alphabet both say spending rises again in 2027, without a number yet.
Written by the New Market Pitch teamWe track new markets so founders and investors can move faster. About us
Which big tech company has spent the most on AI?
Amazon is the best current answer. It leads on its 2026 plan, on cash spent in the latest reported quarter and on spending in the latest completed year. Alphabet remains the strongest alternative because more of its budget is clearly tied to AI.
What does “spent the most on AI” actually mean?
The fairest measure is money committed to AI-driven infrastructure, because company-wide capital spending includes too many unrelated assets.
One number gives a misleading answer here. Amazon builds warehouses, delivery networks, satellites and data centers. Alphabet’s servers support Search, YouTube, Google Cloud and Gemini at the same time. Microsoft mixes ordinary Azure workloads with Copilot and OpenAI services. Meta uses the same computing base for advertising, recommendations and generative AI.
Definitions also differ between companies. Meta’s headline capital expenditure for July–September 2026 was $34.9 billion because it includes finance-lease principal payments; its cash purchases alone were $33.9 billion. Amazon also took on $1.5 billion of equipment under new finance leases on top of its cash bill, and Microsoft’s release shows only its cash line. Cash or finance leases can flip the ranking, and we explain why forecasts disagree so much in our report on AI infrastructure.
We split the question three ways. Who has the largest current plan? Who has already paid out the most cash? And whose spending is most clearly tied to AI? Amazon wins the first two. The third still points to Alphabet, but no company publishes an audited AI-only figure.
Has Amazon overtaken Alphabet with its $220 billion AI budget?
Yes, on the headline plan. Alphabet’s July raise had briefly brought its range level with Amazon’s $200 billion, and Amazon’s $220 billion figure put it back in front. It still has the biggest 2026 spending budget of the four.
Amazon first gave the $220 billion figure on its July 30, 2026 call, when Andy Jassy blamed higher memory prices and AWS capacity that still falls short of demand. On October 29 it kept that guidance and said it plans to spend meaningfully more in 2027. Alphabet kept its $195 billion to $205 billion range on October 28, and Meta now guides to $135 billion to $145 billion. Microsoft has given no comparable calendar-2026 figure in its October results.
Amazon has already spent $175.7 billion in cash in the first nine months of 2026, against $132.0 billion at Alphabet. Both are on track to land close to their plans.
Amazon’s lead is about $20 billion over Alphabet’s midpoint, but its number covers more than AI. Alphabet’s plan is smaller and more concentrated on servers and data centers. Amazon wins the budget ranking. It does not prove Amazon is buying more AI hardware.
| Company | Current spending plan | How clearly it is tied to AI |
|---|---|---|
| Amazon | About $220B (cash capex, guidance, kept in October) | Mostly AWS, plus fulfillment, logistics, satellites and other infrastructure |
| Alphabet | $195B–$205B (guidance, kept in October) | Vast majority on AI technical infrastructure: servers, networking and data centers |
| Microsoft | No comparable calendar-2026 figure confirmed; $39.4B cash in July–September 2026 | Mainly cloud and AI infrastructure, with higher component costs included |
| Meta | $135B–$145B (guidance, incl. finance-lease principal, narrowed in October) | AI and superintelligence capacity plus the core advertising business |
This is just a preview page. Don’t worry if a number looks old: the report gets refreshed, and you’ll always get the latest version.
How much did Amazon, Google, Microsoft and Meta spend in Q3 2026?
Amazon spent the most cash in July–September 2026, the latest quarter all four companies have reported: $77.3 billion of property and equipment purchases, or $75.9 billion after sales and incentives. That is 120% more than a year earlier.
Alphabet spent $51.4 billion, up 114%. Microsoft paid $39.4 billion in cash, up 103%. Meta paid $33.9 billion in cash, up 80%, or $34.9 billion including finance-lease principal.
On cash alone, the four spent about $202 billion in three months. That is more than half of what they spent in their latest completed full years, and about $37 billion more than in April–June.
Amazon’s lead is real, but some of that money went to fulfillment and transport. Its quarter also jumped by $23 billion from April–June, the biggest step of the four. For cash already paid, Amazon wins. On a strictly AI-heavy reading, Alphabet’s quarter is the more concentrated one.
| Company | July–September 2026 spending | What was included |
|---|---|---|
| Amazon | $77.3B gross ($75.9B net) | AWS data centers and equipment plus broader Amazon infrastructure |
| Alphabet | $51.4B | About 65% servers and networking, 35% data centers |
| Microsoft | $39.4B cash (lease-inclusive total not shown in the release) | Cloud and AI infrastructure, including higher component prices |
| Meta | $33.9B cash ($34.9B with lease principal) | Purchases of property and equipment and finance-lease principal payments |
Is Alphabet’s giant spending plan really an AI budget?
Mostly yes, though Alphabet will reuse that capacity across Search, YouTube and Google Cloud.
On its October 28, 2026 call, Alphabet said the vast majority of its $51.4 billion quarter went to technical infrastructure for AI. About 65% of that went to servers and networking and 35% to data centers. It kept its 2026 range at $195 billion to $205 billion, raised in July from $180 billion to $190 billion, and repeated that spending will rise significantly in 2027.
The jump from earlier years makes the link clearer. Alphabet spent about $32 billion in 2023, roughly $53 billion in 2024 and $91.4 billion in 2025. Its current midpoint is more than six times the 2023 total. From $32 billion to $91.4 billion in two years: we explain how fast it is growing, and why, in our report covering AI infrastructure.
The cost now shows in cash flow. Alphabet’s free cash flow was negative $9.1 billion in July–September 2026, after negative $5.9 billion in the quarter before. It also reported $924 billion of maximum future purchase commitments, a ceiling it says can shrink, not a spending plan. Calling every dollar an AI expense would go too far. Describing the buildout as ordinary cloud expansion would miss what is driving it.
How much of Amazon’s giant budget really goes to AI?
Most of the increase, but not the whole budget. Amazon does not publish an AI-only share.
The recent jump tells us more than the headline. Amazon’s net property and equipment purchases reached $210.7 billion over the 12 months to September 30, 2026, up 82%. Amazon said the $94.8 billion year-over-year rise primarily reflects investment in AI. That rise alone is bigger than Alphabet’s full 2025 capital spending.
The demand is visible. AWS sales rose 39% to $46.0 billion in July–September 2026. Amazon said on its call that most of its 2026 budget goes to AWS. Anthropic’s October S-1 shows a commitment to buy at least $110 billion of Amazon cloud services between May 2026 and April 2036.
The bill is heavy. Amazon’s free cash flow over the 12 months to September was an outflow of $42.4 billion. Warehouses, delivery and Project Kuiper still sit inside the broad number, so the full budget overstates Amazon’s AI-only spending. Even after that discount, Amazon is at or near the top.
What does Microsoft’s latest quarter say about its AI spending?
That it is spending faster than ever. Microsoft paid $39.4 billion in cash for property and equipment in July–September 2026, up 103% from $19.4 billion a year earlier.
The full-year picture is less clear. After its July 29, 2026 call, a calendar-2026 figure of about $175 billion was reported, linked to a lease accounting change: Microsoft is extending the useful life of data centers from 15 to 25 years, so more future leases count as operating leases, outside capital expenditure. Some coverage tied that figure to fiscal 2027 instead. Microsoft’s October 28 release gives no full-year number, so we do not rank it on a plan.
The actual spending keeps rising. Microsoft paid $115.9 billion for property and equipment in its fiscal year to June 2026, up about 80%. Adding the latest quarter, its cash spending over the 12 months to September comes to about $136 billion, our calculation.
Microsoft has the demand to show for it. Azure revenue grew 48% to $33.2 billion in July–September 2026, and commercial remaining performance obligations rose 73% to $677 billion. On cash already paid, Microsoft sits behind Amazon and Alphabet but ahead of Meta.
Who is betting the biggest share of the company on AI?
Meta is now making the biggest bet relative to its size, with a plan worth about 70% of its 2025 revenue.
Meta’s $140 billion midpoint compares with $201.0 billion of revenue in 2025. Alphabet sits near 50%. Microsoft’s $115.9 billion of fiscal-2026 cash spending equals about 35% of its $331.8 billion revenue. Amazon is around 31% because retail makes its revenue much larger. Budgets near half of revenue make this a huge market; we work out how big it really is today in our report on AI infrastructure.
The latest quarter tells the same story. Meta’s cash purchases equaled 52.5% of its July–September revenue, against about 43% at Microsoft, 40% at Alphabet and 38% at Amazon. Fiscal calendars and definitions differ, so these are scale estimates rather than exact ratios. Meta earned $64.6 billion of revenue in the quarter but had free cash flow of minus $2.1 billion. It also carries the sharper product risk because most of its revenue still comes from advertising, while Microsoft can rent its capacity through Azure.
| Company | Latest annual revenue | Current plan as a share of revenue |
|---|---|---|
| Meta | $201.0B (2025) | About 70% |
| Alphabet | $403.0B (2025) | About 50% |
| Microsoft | $331.8B (fiscal year to June 2026) | About 35% (fiscal-2026 cash spending; no comparable plan confirmed) |
| Amazon | $716.9B (2025) | About 31% |
Who spent the most in the latest completed year?
Amazon spent the most in 2025, with about $128.3 billion of net capital expenditure.
Alphabet followed at $91.4 billion and Meta at $72.2 billion. Microsoft’s fiscal 2025, which ended in June 2025, came to about $64.6 billion. Together, the four spent about $356.5 billion in those years.
Microsoft’s fiscal year to June 2026 has since closed at $115.9 billion. That period runs six months later than the calendar years of the others, so it cannot be ranked against them directly, but it shows how fast Microsoft has accelerated.
Amazon’s total includes the most non-AI assets. Its lead was still wide enough to make it the biggest spender on physical infrastructure. Alphabet’s total was more concentrated in servers and data centers, which keeps its AI-specific claim alive.
This is a preview page. Spotted an outdated number? That’s normal here. The report you get is always the freshest version.
How fast has Big Tech’s AI spending grown?
Big Tech’s AI infrastructure spending has roughly doubled in a year.
Amazon, Alphabet, Microsoft and Meta spent about $356.5 billion in their latest completed years before 2026. The 2026 plans of Amazon, Alphabet and Meta add up to about $560 billion at their midpoints. With Microsoft’s roughly $136 billion of cash spending over the past 12 months, the four come to about $700 billion. The companies use different calendars and definitions, so the total is a scale estimate.
The jumps inside each company are large. Alphabet’s midpoint is more than twice its 2025 spending. Meta’s is about 94% higher. Amazon’s plan is about 71% above 2025. In July–September 2026 alone, cash spending rose 120% at Amazon, 114% at Alphabet, 103% at Microsoft and 80% at Meta from a year earlier.
Higher memory and chip prices explain part of the increase. Amazon and Microsoft both cited component costs in July 2026. But capacity is doing most of the work: more accelerators, larger campuses, more power and custom silicon at the same time. Amazon and Alphabet already say 2027 will be higher again. The race has moved from tens of billions per company to hundreds of billions.
Do OpenAI, Anthropic and Scale AI deals change the ranking?
No. Strategic stakes in OpenAI, Anthropic and Scale AI are small beside today’s infrastructure budgets.
Meta paid about $14.3 billion for its Scale AI stake. Amazon has invested at least $13 billion in Anthropic. Microsoft’s original commitment to OpenAI has been widely reported near $13 billion; the stake later became worth far more. Stakes this large raise the question of who is funding whom, which we follow in our AI infrastructure market report.
The bigger numbers now run the other way. Anthropic’s S-1, filed on October 26, 2026, lists minimum cloud commitments of at least $110.0 billion to Amazon, $114.5 billion to Google and $44.9 billion to Microsoft, spread over seven to ten years. Those are Anthropic’s future purchases, not the providers’ capital spending, and they should not be added to the budgets above.
Stakes also move profits, in both directions. Amazon booked $53.4 billion of non-operating income in April–June 2026, mainly from Anthropic, while Alphabet took an $8.7 billion loss on equity securities in July–September, mainly from its SpaceX stake. Those are paper swings, not AI spending. AI talent is costly too: Alphabet’s research and development expenses rose about 36% in July–September 2026, but filings never isolate a full AI payroll.
Do Apple and Nvidia belong in this AI spending race?
No. Apple follows a lower-spending AI strategy, while Nvidia sells much of the equipment the cloud giants buy.
Apple’s latest annual filing shows $12.7 billion spent on property and equipment, less than one-tenth of Amazon’s 2025 total. It spends heavily on research, custom chips and private cloud systems, and Alphabet named it in October as a customer scaling model training on Google’s TPUs. Its multiyear US investment announcements include manufacturing, suppliers, facilities and entertainment, which makes them far broader than AI budgets.
Nvidia has the opposite profile. In August 2026, it reported $89.0 billion of data-center revenue for the quarter ended July 26, up 117% from a year earlier, within $96.2 billion of total revenue. Much of that money comes from the companies in this article. It is revenue Nvidia receives, not money it spends.
That leaves the real spending race among the four cloud giants that pay for buildings, power and full server fleets.
Preview only. Some numbers may have changed since we made this page, but your report will be up to date.
Who is getting the clearest return from all this AI spending?
Amazon earns the largest cloud profit, Alphabet has the fastest cloud growth, and Microsoft has the biggest contracted backlog. No single company wins on return.
Amazon Web Services generated $46.0 billion of revenue and $17.8 billion of operating income in July–September 2026. That gives it the largest proven channel for turning AI spending into profit.
Google Cloud is growing much faster. Its revenue doubled, up 103% to $30.7 billion in the same quarter, while operating income rose to $11.1 billion from $3.6 billion and backlog reached $561 billion. Microsoft’s Azure grew 48% and its commercial backlog reached $677 billion, though OpenAI accounts for a large part of that backlog. Microsoft does not disclose Azure’s profit on its own.
Meta earns its return indirectly. Its revenue grew 26% in July–September 2026 as ad impressions rose 10% and the average price per ad rose 14%, helped by AI ranking and targeting. Its free cash flow turned negative in the same quarter, which shows the price of the bet.
So which big tech company has spent the most on AI?
Amazon. It leads on the 2026 plan, on cash spent in the latest reported quarter and on the latest completed year.
The caveat is scope. Amazon’s numbers also pay for warehouses, logistics and satellites, while Alphabet’s $195 billion to $205 billion plan is more clearly concentrated on AI servers and data centers. On a strict AI-only basis the two are close, and no filing separates the figures.
Microsoft doubled its quarterly cash spending but gives no comparable full-year plan, so it ranks third on cash rather than on budget. Meta is spending the largest share of its revenue and now feels it in negative free cash flow.
For a reader asking which Big Tech company has spent the most on AI, we would name Amazon, with Alphabet as the strongest challenger on AI-specific spending. Both say 2027 spending will be higher, so the gap to watch is the first 2027 numbers, not this year’s.
Our methodology
This analysis tests which big technology company has made the largest AI spending commitment based on the freshest comparable evidence available. We compare current capital-expenditure plans, cash already deployed, the share of spending companies explicitly connect to AI, growth from previous years, spending relative to revenue and the commercial returns already visible.
We do not treat company-wide capital expenditure as a clean AI budget. Amazon’s figures include logistics, fulfillment and satellites; Alphabet’s infrastructure supports Search, YouTube and Google Cloud; Microsoft mixes AI and broader Azure capacity; and Meta uses the same systems for advertising, recommendations and generative AI.
Current annual plans and completed cash spending answer different questions, so we keep them separate. Guidance shows the scale of the present commitment, while reported capital expenditure shows what has actually been paid during a specific quarter or year. Plans are the latest each company reported in its October 2026 results.
Quarterly figures are compared only when the reporting periods line up. We use July–September 2026, the latest quarter all four companies have reported, and show cash figures separately from figures that include finance leases.
We give more weight to figures that management directly links to AI demand, chips, servers, data centers, networking and power. Broad infrastructure totals receive less weight when a meaningful share funds non-AI businesses. Microsoft gave no full calendar-2026 plan comparable to the others in its October results, so we use its reported cash spending instead.
Spending as a share of revenue is used as a measure of commitment, not financial efficiency. Fiscal calendars and accounting definitions differ, so those percentages are treated as scale estimates rather than perfectly standardized ratios.
Strategic investments in OpenAI, Anthropic and Scale AI are included for context but do not replace infrastructure spending. Gains or losses on those stakes are not treated as spending or as AI revenue, and customer cloud commitments, such as those in Anthropic’s S-1, are not added to the providers’ capital spending. Research and AI talent costs are discussed only where companies disclose them clearly.
We also compare visible returns, including Amazon Web Services revenue and operating income, Google Cloud growth and profitability, Azure growth and backlog, and Meta’s advertising performance. These measures show how spending may be paying off, but they do not determine the spending ranking on their own.
We prioritized official investor-relations releases, annual and quarterly filings, earnings-call transcripts and company announcements. We avoided treating broad investment pledges, repeated press summaries or unattributed estimates as equivalent to audited spending or formal guidance.
Key sources include Amazon’s third-quarter 2026 results, Alphabet’s third-quarter 2026 results, Alphabet’s third-quarter 2026 earnings call, Microsoft’s fiscal 2027 first-quarter results, Microsoft’s fiscal 2026 fourth-quarter earnings call, Meta’s third-quarter 2026 results, Meta’s fourth-quarter and full-year 2025 results, Anthropic’s S-1 registration statement, Apple’s multiyear US investment announcement, and Nvidia’s second-quarter fiscal 2027 results.
Written by the New Market Pitch teamWe track new markets so founders and investors can move faster. About us
