This page covers the budget-control layer that turns AI demand into funded infrastructure. A hyperscaler capex budget is the approval, financing, procurement, and utilization plan for data-center campuses, accelerator clusters, custom silicon, networking, storage, power contracts, leases, land, and construction. The budget creates shareholder economics only when that capacity becomes cloud revenue, ad or productivity monetization, backlog or remaining performance obligation conversion, operating income, utilization, and free cash flow after capex, depreciation, leases, power, financing claims, and component costs.
What the stack is: corporate capital-allocation, procurement, financing, and utilization controls for AI data-center capacity at Microsoft Azure, Google Cloud, Amazon Web Services, Meta Family of Apps, and Oracle Cloud Infrastructure.
What it does in the chain: it is the demand origin for downstream accelerator, server, networking, memory, power, cooling, construction, lease, and financing nodes. Supplier revenue depends on these budgets staying funded.
Main pieces: capex plans, property and equipment additions, finance leases, purchase obligations, cloud backlog, remaining performance obligations, customer prepayments, data-center leases, power contracts, GPU/server/network/storage orders, and cloud or ad monetization metrics.
Where it sits: inside platform capital-allocation and cloud procurement teams, before the physical build reaches suppliers, utilities, land, construction, and financing markets.
How the theme uses it: the node asks whether AI infrastructure spending is converting into revenue, operating income, and cash rather than only expanding assets, leases, debt, depreciation, and commitments.
Terms used below: capex means capital expenditures; property and equipment is the balance-sheet asset base from the build; remaining performance obligations are contracted revenue not yet recognized; Oracle Cloud Infrastructure is shortened to OCI after first use; free cash flow after capex is the cash conversion test.
Report boundary: this node is a tactical report layer. It ranks the current hyperscaler and AI-cloud budget basket, chart routes, confirmation triggers, invalidation evidence, and source provenance. Durable research, claim IDs, raw source registries, and sector thesis maintenance stay in the linked knowledge pages. Right-rail charts use the report API backed by read-only discovery daily_ohlc through 2026-06-12.
Current Setup
Budget conversion gateFunded AI capacity has to become revenue, margin, and cash.
MSFT and GOOGL still have the strongest reported bridge from cloud demand to operating income. AMZN, META, and ORCL keep the theme investable, but each needs cleaner cash conversion or funding-quality evidence after the build.
Stronger evidenceCloud revenue and backlog are real.
Azure, Google Cloud, AWS, and OCI show demand, while Meta's ad business funds the build.
Conversion testCapex must stop outrunning cash.
AMZN, META, and ORCL need better proof that infrastructure spend lifts margin and FCF.
Main invalidationFinancing fills the gap before utilization does.
Debt, leases, preferred claims, equity issuance, or customer concentration can weaken per-share conversion.
Step 1Budget approved
Step 2Capacity built
Step 3Workload monetized
Step 4Cash proof
Hyperscaler and AI-cloud budgets set the spending pool for the whole AI capex chain. When MSFT, GOOGL, AMZN, META, and ORCL approve or finance data-center, accelerator, networking, storage, power, lease, and construction spending, that demand can reach semiconductors, servers, EMS, electrical equipment, cooling, software, utilities, and financing markets. The budget itself is not the payoff. The payoff has to show up in cloud, ad, productivity, or database-cloud revenue, margin, backlog or RPO conversion, utilization, and FCF after capex.
Current sources show real demand signals. MSFT reported Q3 FY26 Azure growth of 40%, Microsoft Cloud revenue of $54.5B, commercial RPO of $627B, and nine-month operating cash flow above PP&E additions. GOOGL reported Q1 Google Cloud revenue of $20.028B, Cloud operating income of $6.598B, a 32.9% Cloud margin, and Cloud backlog above $460B. AMZN reported Q1 AWS sales of $37.587B and AWS operating income of $14.161B. META reported Q1 revenue growth of 33%, ad impressions up 19%, and average price per ad up 12%. ORCL reported Q3 FY26 OCI revenue of $4.888B, up 84%, and RPO of $552.6B. The next proof point is capacity spending tied to stable margins, backlog or RPO conversion, and FCF after capex.
The setup weakens if capex, PP&E additions, leases, purchase commitments, data-center contracts, component pricing, depreciation, power costs, debt, preferred claims, or equity issuance rise faster than cloud, ad, productivity, OCI, operating-income, and FCF evidence. GOOGL has strong Cloud evidence but the local lane now puts financing and per-share conversion in the watch queue; AMZN has TTM PP&E purchases slightly above TTM operating cash flow; META raised 2026 capex while still lacking a clean AI infrastructure to FCF KPI bridge and has unconfirmed external-financing headlines to monitor; ORCL has the largest funding-quality gap because OCI growth and RPO sit beside negative trailing FCF, high borrowings, large lease commitments, and customer-concentration uncertainty. Oracle's Q4 FY26 release arrived after the local ORCL lane's Q3 snapshot, so the canonical lane needs refresh before treating Oracle fundamentals as fully current.
The visual summary preserves the three source setup fields in hidden structured HTML. Fundamental claims route to the AI Capex theme page, linked security lanes, and the official releases cited in Source Trail.
Basket
This basket is inherited from the parent theme's Capital funders group and ranked by source-backed capex economics before technical timing. The order is MSFT, GOOGL, AMZN, META, then ORCL: first the names with the clearest cloud-to-cash evidence, then the names where AI spending still needs a cleaner bridge to free cash flow, funding quality, or customer breadth.
Enterprise cloud and productivity funder with the cleanest reported bridge from Azure demand to cash after capex.
Market cap$2.90T
Next earningsNot confirmed
Latest qtr revenue$82.886B
Role in stack
Microsoft approves and absorbs AI infrastructure through Azure data centers, Microsoft Cloud services, Copilot, GitHub, and Microsoft 365 distribution. The node works if Azure capacity and commercial backlog convert into cloud revenue, margin, and free cash flow after property and equipment additions, power, leases, and depreciation.
Revenue mix
Cloud and subscription revenue dominate the node exposure. Productivity and Business Processes plus Intelligent Cloud were about 84% of Q3 FY26 revenue and more than 90% of segment operating income in the local security lane.
Latest qtr revenue
Q3 FY26 total revenue was $82.886B from Microsoft's FY26 Q3 release and metrics tables. Microsoft Cloud revenue was $54.5B, Azure and other cloud services grew 40%, commercial remaining performance obligations were $627B, and Microsoft Cloud gross margin was 66%.
Search-funded cloud and TPU spender with strong Cloud margin and backlog evidence.
Market cap$4.36T
Next earningsNot confirmed
Latest qtr revenue$109.896B
Role in stack
Alphabet funds Google Cloud, TPUs, Gemini, Workspace, security infrastructure, and data centers from Search, YouTube, subscriptions, and Cloud cash flows. The node works if Cloud backlog converts into revenue and margin while capex, leases, AI research, power, and regulatory costs stay funded.
Revenue mix
Google Services remains the funding engine and Google Cloud is the direct AI infrastructure route. Q1 2026 Services revenue was $89.637B and Google Cloud revenue was $20.028B, with Cloud operating income of $6.598B.
Latest qtr revenue
Q1 2026 total revenue was $109.896B from Alphabet's Q1 2026 release. Cloud margin was 32.9%, Cloud backlog exceeded $460B, Q1 capex was $35.674B, and Q1 free cash flow was $10.116B after capex.
AWS is the direct cloud profit pool, but current cash conversion is weaker than the top two ranks.
Market cap$2.57T
Next earningsNot confirmed
Latest qtr revenue$181.519B
Role in stack
Amazon owns AWS capacity, cloud pricing, custom silicon, and a large retail and advertising cash engine. The node works if AWS growth, AWS operating income, utilization, power access, and retail or advertising cash flow rebuild free cash flow after heavy property and equipment spending.
Revenue mix
Retail and marketplace activity supply most revenue, but AWS is the profit center. Q1 2026 AWS was about 20.7% of sales and 59.4% of segment operating income in the local security lane.
Latest qtr revenue
Q1 2026 consolidated net sales were $181.519B from Amazon's Q1 2026 release. AWS sales were $37.587B, AWS operating income was $14.161B, Q1 cash capex was $43.2B, and trailing PP&E purchases slightly exceeded trailing operating cash flow.
Ad-funded AI infrastructure spender where the payoff must show up through ad ranking, engagement, and FCF.
Market cap$1.44T
Next earningsNot confirmed
Latest qtr revenue$56.311B
Role in stack
Meta funds data centers, servers, models, AI talent, and infrastructure commitments from Family of Apps advertising. The node works if those assets improve ad conversion, ranking, engagement, business messaging, agents, devices, operating income, or free cash flow rather than only raising depreciation and commitments.
Revenue mix
Family of Apps supplies nearly all revenue and profit. Q1 2026 FoA revenue was $55.909B and FoA operating income was $26.900B; Reality Labs remained small and loss-making.
Latest qtr revenue
Q1 2026 total revenue was $56.311B from Meta's Q1 2026 release. Revenue grew 33%, operating margin was 41%, ad impressions rose 19%, average price per ad rose 12%, and free cash flow was $12.386B after capex.
OCI and database-cloud capacity route with large RPO upside and the highest funding-quality gap.
Market cap$530B
Next earningsNot confirmed
Latest qtr revenue$19.18B
Role in stack
Oracle funds OCI regions, database cloud capacity, and large AI infrastructure contracts. The node works if OCI revenue, RPO conversion, customer prepayments or customer-supplied GPUs, and cloud margins cover capex, leases, power, debt, depreciation, and possible dilution.
Revenue mix
Cloud and software are the base; OCI is the fast-growth AI infrastructure leg. The local security lane is still anchored to Q3 FY26 and needs a post-Q4 refresh, so this card treats the latest quarter as a bounded node update rather than a canonical lane replacement.
Latest qtr revenue
Q4 FY26 total revenue was about $19.18B from Oracle's June 2026 fiscal fourth-quarter release and financial-calendar coverage. The prior local lane snapshot reported Q3 FY26 revenue of $17.190B, OCI revenue of $4.888B, RPO of $552.6B, negative trailing FCF, high borrowings, and large lease commitments.
Market caps use local discovery daily_ohlc closes and instruments.weighted_shares_outstanding as of 2026-06-12; they are rounded and are not live exchange quotes. Official IR event pages and financial-calendar searches on 2026-06-13 did not provide confirmed next earnings dates for the basket, so every card marks next earnings as not confirmed. Expanded source-routed addition: ORCL. Rejected or routed elsewhere: CRWV, NBIS, IREN, APLD, CORZ, and other capacity-financing names belong in AI-cloud capacity; DLR and EQIX belong in data-center capacity or physical-delivery routes; BABA, BIDU, TCEHY, TCTZF, SAP, SFTBY, and IBM stay watch-only until local discovery coverage, durable lanes, and budget-setting evidence support inclusion.
What Confirms Or Weakens
AreaWhat confirmsWhat weakens or invalidatesWatch next
1Node thesisBudget conversion
What confirms
Capex stays tied to Azure, Google Cloud, AWS, Family of Apps, or OCI revenue, backlog or RPO conversion, operating income, and FCF after capex.
What weakens or invalidates
Capex guides, PP&E additions, leases, infrastructure commitments, debt, preferred claims, or equity issuance rise without matching revenue, utilization, margin, or cash-conversion evidence.
Watch next
MSFT FY27 capex language
GOOGL Q2 Cloud and financing detail
ORCL post-Q4 funding quality
2Economics and recovery mechanismCloud to cash
What confirms
Azure holds near the 39%-40% constant-currency guide, Google Cloud backlog converts at stable margin, AWS keeps growth and operating income moving together, and ORCL converts RPO into OCI revenue.
What weakens or invalidates
Cloud revenue slows, backlog or RPO mix is lower margin, utilization is weak, or AI workload pricing fails to cover depreciation, power, component costs, and financing costs.
Watch next
Azure and RPO
Google Cloud margin
AWS operating income
OCI growth and RPO
3Customer and demand proofAd, productivity, cloud
What confirms
Copilot seats, Search paid clicks and cost per click, Meta ad impressions and price, Amazon retail productivity, and cloud workload usage show measurable revenue or operating-income gains.
What weakens or invalidates
AI usage grows while ad economics, Copilot ARPU, retail productivity, business messaging, agents, devices, or cloud margin weaken.
Watch next
Copilot ARPU or usage
Search CPC
Meta ad price
Amazon retail operating income
4Funding capacityFCF after capex
What confirms
Operating cash flow funds more of the buildout and FCF after capex improves while revenue and operating income keep growing.
What weakens or invalidates
Capex consumes operating cash flow for multiple quarters, commitments expand faster than revenue, or external financing fills a gap before utilization and margin are visible.
Watch next
MSFT OCF less PP&E
GOOGL CFO/capex bridge
AMZN PP&E purchases
META and ORCL financing
5Policy and regulationRules and remedies
What confirms
Export, antitrust, privacy, and data-center policy outcomes leave cloud AI demand, ad monetization, and infrastructure deployment intact.
What weakens or invalidates
Antitrust remedies, privacy constraints, export restrictions, or data-center permitting rules reduce Search/ads monetization, cloud demand, AI hardware access, or usable capacity.
Watch next
Alphabet remedies
Export-control updates
Data-center permitting
Privacy and ad targeting
6Operating and supply constraintPhysical delivery
What confirms
Leases, purchase commitments, data-center contracts, power, and component supply are matched by visible customer demand, utilization, and margin.
What weakens or invalidates
Power constraints, memory or networking bottlenecks, export controls, lease backstops, customer-concentration issues, or supplier delays reduce usable capacity or margin.
Watch next
Google leases
Meta commitments
AWS-linked commitments
Oracle leases and customer funding
7Stale conditionRefresh trigger
What confirms
Discovery daily_ohlc remains current, linked security lanes cover the latest filings, guidance, capex, commitments, and cloud or ad metrics, and financing reports are separated from company-confirmed filings.
What weakens or invalidates
A new trading session, earnings release, 10-Q, capex guide, lease or power commitment, financing update, regulatory remedy, source-lane refresh, or ORCL post-Q4 update arrives before this page is refreshed.
Watch next
Discovery coverage
Source lanes
API chart package
Node metadata
Source Trail
Canonical Thesis
AI Capex Cycle defines the platform funder bucket and the capex-to-cash proof burden.
Information Technology supports the broad AI infrastructure demand route across cloud, data-center systems, semiconductors, networking, servers, and software.
Communication Services supports the ad-funded platform route for GOOGL and META and the regulatory-remedy caveat.
Security Lanes
MSFT supports Azure growth, Microsoft Cloud revenue, commercial RPO, Copilot, PP&E additions, capex, and FCF conversion questions.
GOOGL supports Search, Google Cloud revenue and margin, Cloud backlog, capex guidance, commitments, leases, backstops, financing watch items, and regulatory risk.
AMZN supports AWS sales and operating income, Q1 cash capex, TTM OCF versus PP&E purchases, commitments, retail productivity, and FCF recovery risk.
META supports Family of Apps ad revenue, capex guidance, infrastructure commitments, Reality Labs drag, FCF after capex, and the missing AI-to-FCF KPI bridge.
Power Scarcity And Grid Load owns the power, cooling, interconnection, utility, and physical delivery constraints that can delay usable capacity.
Memory And Component Scarcity should be reopened when HBM, DRAM, NAND, packaging, or component pressure changes deployment timing or margin.
Discovery And Chart Provenance
The selected-security right rail calls /api/securities/{ticker}/chart?frequency=weekly&window=3y&as_of=latest. The report API returns weekly packages from read-only discovery daily_ohlc using first open, maximum high, minimum low, final close, and summed volume.
Discovery status checks on 2026-06-13 showed local daily_ohlc rows through 2026-06-12 for MSFT, GOOGL, AMZN, META, and ORCL. Latest closes were MSFT $390.74, GOOGL $359.68, AMZN $238.55, META $566.98, and ORCL $184.13.
A read-only DuckDB market-cap query multiplied 2026-06-12 closes by instruments.weighted_shares_outstanding. Rounded market caps were MSFT $2.90T, GOOGL $4.36T, AMZN $2.57T, META $1.44T, and ORCL $530B.
Lineage status checks for MSFT and ORCL showed provider massive, adjusted daily bars, and succeeded runs completed on 2026-05-11 with source freshness 2026-05-08. The canonical table has newer rows through 2026-06-12, so provider-lineage freshness should be reconciled before using lineage as live provider freshness.
Representative API validation for MSFT on 2026-06-13 resolved as_of=latest to 2026-06-12 and returned source table reports.weekly_ohlc+discovery.daily_ohlc, weekly OHLC rows, 20-period mean, 100-period EMA, Bollinger fields, and volume averages. If an API server is not running, static HTML still renders and the selected-security panel shows its empty state.
The removed static setup table used 2026-06-05 weekly thresholds. Those thresholds are stale relative to the 2026-06-12 API package and must be refreshed before use as current trading evidence.
Parent expanded-basket metadata records 5,274 active common stocks and 4,991 active common stocks with at least 64 daily bars and latest OHLC coverage through 2026-06-05.
Earnings And Revenue Sources
Next-earnings checks were run on 2026-06-13 against company IR event pages and financial-calendar pages for Microsoft, Alphabet, Amazon, Meta, and Oracle. No confirmed next earnings dates were available from the inspected sources, so the basket cards use Not confirmed rather than estimated dates.
MSFT latest-quarter revenue is Q3 FY26 total revenue of $82.886B from Microsoft's FY26 Q3 press release and metrics table. The same source supports Microsoft Cloud revenue of $54.5B, Azure and other cloud services growth of 40%, commercial RPO of $627B, and Microsoft Cloud gross margin of 66%.
GOOGL latest-quarter revenue is Q1 2026 total revenue of $109.896B from Alphabet's Q1 2026 release. The same source supports Google Services revenue of $89.637B, Google Cloud revenue of $20.028B, Cloud operating income of $6.598B, Cloud backlog above $460B, Q1 capex of $35.674B, and Q1 FCF of $10.116B.
AMZN latest-quarter revenue is Q1 2026 consolidated net sales of $181.519B from Amazon's Q1 2026 release. The same source supports AWS sales of $37.587B, AWS operating income of $14.161B, Q1 cash capex of $43.2B, and the TTM operating-cash-flow versus PP&E-purchases comparison in the local lane.
META latest-quarter revenue is Q1 2026 total revenue of $56.311B from Meta's Q1 2026 release. The same source supports FoA revenue of $55.909B, FoA operating income of $26.900B, ad impressions up 19%, average price per ad up 12%, capex including finance leases of $19.840B, and FCF of $12.386B.
ORCL latest-quarter revenue is Q4 FY26 total revenue of about $19.18B from Oracle's June 2026 fiscal fourth-quarter release and financial-calendar coverage inspected on 2026-06-13. The local ORCL knowledge lane still uses the Q3 FY26 release, so Oracle's Q4 details should be clipped into the canonical lane before using the lane as fully refreshed research.
Known Gaps
MSFT does not separately disclose Azure AI revenue, AI capex, AI depreciation, or Copilot profitability in the local lane.
GOOGL backlog mix, TPU timing, Search AI monetization, financing mix, and full peer primary-source review remain incomplete.
AMZN does not disclose standalone profitability for advertising, Prime, AI services, custom silicon, LEO, or private AI investments; AI infrastructure ROIC is externally hard to verify.
META has no clean KPI bridge from AI infrastructure spending to incremental revenue, margin, or FCF in the local lane.
ORCL has a source-refresh risk because Q4 FY26 earnings have now been released, while the local ORCL lane remains anchored to Q3 FY26. OCI growth, RPO conversion, customer funding, debt, lease commitments, and FCF should be refreshed in the canonical lane.