Big Tech’s $3.2 trillion commitments and AI financing | Magica
Big Tech’s reported $3.2 trillion commitments test AI financing
Editorial Team
••📖8 min read
Five technology groups have accumulated an estimated $3.2 trillion in future purchase and lease commitments. Filings show the funding pressure, but broad spending categories, long payment schedules and a methodology change limit what the total says about AI debt and growth.
Two reported analyses put five US technology groups’ future purchase and uncommenced lease commitments at roughly $3.2 trillion.
The reported 90% rise in six months is complicated by an apparent methodology change; the filings also include spending beyond AI.
Long contracts help suppliers borrow; power constraints and the timing of customer revenue affect the investment return.
Five US technology groups have accumulated approximately $3.2 trillion in future commitments as they expand AI infrastructure, according to a report relaying Nikkei’s analysis of their latest earnings disclosures. The tally, partly estimated, is reported to have risen 90% between the end of March and the end of September. An apparent change in Google parent Alphabet’s calculation methodology limits that growth comparison.
The buyers are Alphabet; software and cloud provider Microsoft; retail and cloud group Amazon; Meta Platforms, whose apps include Facebook and Instagram; and Oracle, a supplier of enterprise databases, applications and cloud infrastructure. The companies are securing equipment, facilities and outside computing capacity to develop and run AI services.
The significance extends beyond their own cash budgets. Purchase and lease contracts give suppliers revenue commitments against which they can raise money to build. That links infrastructure developers and lenders to a small set of customers—and makes the timing of those customers’ payments and returns consequential across the financing chain.
Long-term purchases account for 60% of Nikkei’s reported tally; leases not yet commenced account for 40%. The categories include spending beyond AI. Source: BusinessKorea.
A large commitment stock, with uneven boundaries
Nikkei’s tally assigns 60% of the total to long-term purchases, including chips, servers and electricity, and 40% to equipment and facility leases that have not commenced. The report puts the total at about 80% of the five companies’ combined assets. These are commitments disclosed in financial-statement notes, not a single category of outstanding borrowing; the asset comparison illustrates scale rather than a conventional debt ratio.
PwC finds growth in a subset of AI-exposed junior job postings demanding senior skills, alongside pressure on entry-level opportunities. Workplace research suggests AI can support learning, but leaves employers to demonstrate how recruits acquire independent expertise.
Editorial Team
A separate report on Goldman Sachs credit research reaches roughly the same $3.2 trillion total. Its components make the accounting boundary clearer: approximately $1.3 trillion in leases that have not begun, plus $1.9 trillion in other purchase commitments. It also identifies about $400 billion of leases already commenced, outside that off-balance-sheet sum. Of the purchase commitments, $685 billion is due by the end of 2027.
The Goldman Sachs research says some commitments are already reflected in consensus capital-expenditure forecasts. Adding the entire stock to those forecasts would risk double counting. Nor does either reported total establish an exclusively AI spending bill: the underlying corporate disclosures cover broader categories.
Alphabet’s September filing, for example, lists $924.4 billion of material purchase commitments and $130.5 billion of payments for leases not yet commenced. Those two categories add to approximately $1.055 trillion, consistent with Nikkei’s rounded $1.05 trillion figure.
But the purchase category includes content licenses and open purchase orders alongside technical infrastructure, inventory and energy contracts. The filing says the future cash commitments depend on counterparties performing their obligations. Of the $924.4 billion in purchase commitments, $213.4 billion is due within 12 months. An $802.1 billion multi-year component is to be fulfilled predominantly through 2031.
Amazon’s contractual-commitment disclosure illustrates the limits of comparisons across companies. It excludes generally cancellable open orders supporting normal operations, while its unconditional purchase obligations include energy, digital-media content and software licenses. For variable contracts, it generally counts only specified minimums or termination penalties; energy agreements without fixed or minimum volume commitments are excluded. Contract definitions help determine the reported totals.
Alphabet also discloses contingent support separately: $75.7 billion of data-center backstop obligations and $22.8 billion of financial guarantees, primarily supporting counterparties’ procurement of equipment for future energy agreements. Payment depends on the relevant contractual triggers, so these contingent exposures require separate treatment.
Company-reported additional leases not yet commenced, primarily for data centres. The snapshots are 15 months apart; some September 2026 arrangements depend on contractual conditions. Source: Microsoft fiscal 2025 SEC annual report.
Lease payments extend well beyond the construction phase
Selected filings show the scale of leases that have not started. These figures cover that lease category alone, rather than each company’s total commitments, and Oracle’s reporting date is a month earlier.
Company
Reporting date
Payments or commitments for leases not yet commenced
Microsoft’s additional leases, primarily for data centers, are scheduled to commence between fiscal 2027 and fiscal 2034 and have terms of one to 20 years. Some arrangements depend on contractual conditions being met. Its comparable disclosure was $92.7 billion at June 30, 2025 in its fiscal 2025 annual report. The September 2026 amount is roughly 3.9 times that level over 15 months, independently demonstrating growth over a different period from the reported six-month aggregate.
Oracle’s additional commitments are substantially all for data centers. They are generally expected to commence from the second quarter of fiscal 2027 through fiscal 2029, with terms of 15 to 19 years. Amazon’s schedule places approximately $114.8 billion of its $164.9 billion in uncommenced lease payments after 2030—about 70%, calculated from the filing’s payment buckets. Signing a contract, starting the lease and paying its full lifetime cost are distinct events.
The companies can purchase infrastructure, lease facilities or buy capacity from outside cloud providers. Each route changes when cash is needed and who finances construction. None removes the physical constraints: Microsoft warns in its September filing that power availability, grid connections and permitting can restrict expansion. It also says competitors use some of the same scarce components, affecting prices and availability.
Buyers’ contracts support suppliers’ credit
Meta’s commitments were approximately $810 billion, up 90% since March, according to the reported Nikkei tally. It cites a data-center leasing joint venture with BlackRock and capacity purchases from outside cloud providers.
One such supplier is CoreWeave, an AI cloud provider established in 2017 that completed its Nasdaq listing in March 2025. In an April 9 announcement filed with the SEC, CoreWeave said it had expanded its existing relationship with Meta through an agreement for approximately $21 billion of AI cloud capacity through December 2032. The announcement identifies scaling inference—the operation of AI models—as the intended use.
The financing chain also reaches chip suppliers. SK hynix, a memory-chip maker, said it had signed long-term supply agreements with 14 major customers covering more than half its high-bandwidth memory production capacity for 2027 and 2028, according to the same report. That report also describes a sharp fall in CoreWeave’s bond prices in October and warns that weaker cloud providers could struggle to sustain investment if financing becomes harder to obtain.
The Bank for International Settlements, whose mission includes supporting central banks’ pursuit of monetary and financial stability, published research on this structure on October 27. The bulletin’s authors found that commitments from a few large firms improve the credit standing of infrastructure debt and draw in institutional capital. Banks often lend to both financing vehicles and investors holding their debt.
This redistributes exposure through the financial system without necessarily diversifying the customers behind it. But the researchers found that spillovers to US Treasury and investment-grade bond markets on AI-financing news days in 2026 were small and short-lived. They warned that sustained issuance could change that. The bulletin expresses its authors’ views, not necessarily those of the institution or its member central banks.
Goldman Sachs’ credit research offers another qualification: it says much of the AI ecosystem has ratings near the high end of investment grade, leaving room for downgrades without necessarily losing debt-market access. That assessment concerns the availability of funding; it does not establish an adequate investment return.
Bank for International Settlements researchers’ stylised financing map. Solid arrows show cash and contracts; dashed arrows show guarantees and backstops. The analysis expresses the authors’ views. Source: Bank for International Settlements.
The next test is turning capacity into cash
Alphabet’s cash-flow statement shows how quickly the funding balance can change even when operating cash grows. In the first nine months of 2026, it generated $127.1 billion of operating cash and spent $132.0 billion on capital expenditures. The comparable 2025 amounts were $112.3 billion and $63.6 billion. Using the unrounded cash-flow statement amounts, operating cash minus capital spending gives approximately $4.8 billion of negative free cash flow in 2026, versus $48.7 billion positive a year earlier, before other investing and financing demands.
Oracle has offered a timetable for easing its own funding pressure. At its October 28 investor day, it targeted positive free cash flow in the second half of fiscal 2028 and at least $30 billion in fiscal 2029, supported by lower capital spending in fiscal 2029. It said vendor financing, customer-provided hardware and customer prepayments should reduce upfront cash requirements. These remain management expectations.
Oracle also said it would limit additional capital-market funding through fiscal 2030 to $20 billion, excluding refinancing: approximately $10 billion in equity and $10 billion in gross debt. That is on top of a $20 billion equity raise already completed in the first quarter of fiscal 2027. The November 4 report on AI financing says Oracle had sold another $10 billion in bonds and $9.9 billion in equity earlier that week, close to the additional amount outlined at investor day.
The remaining test is therefore more than access to fresh capital. Buyers and suppliers must deliver usable capacity despite power and equipment constraints, then generate customer cash on a schedule that supports years of contracted payments. Future disclosures of lease commencements, cash expenditure and operating cash generation will show whether those schedules align.
CoreWeave bonds fell after the election of candidates favoring data center restrictions. Texas governor-elect Gina Hinojosa wants to extend an existing approvals pause until lawmakers pass protections, adding delivery uncertainty for an industry whose expansion depends on power and financing.
Walmart plans new permissions and a gateway for outside shopping agents, while Amazon blocks unauthorized tools and sells ads through its own assistant. The stakes are product discovery, customer data and whether delegated purchases add sales or displace advertising opportunities.
Apple reportedly cut some engineers’ daily AI allowances from $300 to $200 in October and is weighing deeper restrictions. The limits tighten control over model use, but unused allowances make the savings uncertain.
Factory is acquiring coding-agent collaboration platform HumanLayer after raising $200 million in September. It plans to integrate shared tasks and planning tools, while HumanLayer promises open source, refunds and at least 30 days of continued service.
A minister identifies Google-commissioned research behind the action plan’s £400 billion AI claim. The government’s central estimate is £120 billion, but different methods and adoption assumptions prevent a straightforward forecast-cut comparison.
Research has separated language models’ detection of altered internal activity from their ability to identify its content. Mechanistic studies and skeptical controls leave the interpretation disputed—and one intervention that improved detection made other measures of self-knowledge worse.
The IEA maps 134 policies governing data-center integration, arguing that credible project queues, coordinated tariffs and flexible connections can protect consumers while accommodating AI growth. Many measures remain too new to judge.
Three watchdog groups seek to halt Project Meridian, alleging that its contractor-heavy membership and closed meetings violate federal advisory committee law. MITRE’s invitation for outside ideas complicates the dispute over public participation.
Meta is seeking up to €9 billion in its first euro bond sale, drawing record orders while offering wider spreads than Alphabet. Its advertising business and large liquid reserves cushion an AI investment push that has overtaken quarterly operating cash flow.
Pangram is expanding its text-detection business into images and video. Browser checks and developer access broaden its reach, while company benchmark scores leave real-world error rates and costs unresolved.
About two dozen restrictive data center measures appeared to pass in initial returns. Texas governor-elect Gina Hinojosa’s promised construction pause adds a state-level constraint, though the capacity ultimately delayed remains uncertain.
Riot Platforms says its 15-year Corsicana lease with Anthropic will generate about $28 billion in revenue. The tenant owns and operates the compute, while Riot develops power infrastructure and funds module reimbursements through debt or credits against future rent.
Sonos plans limited U.S. early access for its conversational assistant, with a reported November 6 start. Users can return to older voice controls—a choice that also separates local audio processing from the new assistant’s cloud processing.
Tesla, SpaceX and AI infrastructure shares fell after Democratic election wins, while broader stocks rallied. Texas governor-elect Gina Hinojosa’s proposed data center pause puts a major construction pipeline in focus, but Republicans retain control of the Texas Legislature and federal agencies.
Capital One Ventures leads Sentra’s $150 million Series C and gains a board seat as the company expands AI-agent data controls. Sentra reports roughly 100 major enterprise customers and revenue in the tens of millions, but exact revenue, valuation and comparative costs remain undisclosed.
Google agreed to support Anthropic’s lease obligations at TeraWulf’s Kentucky campus, strengthening its bid for project financing. The support starts after delivery milestones, while debt terms and a guaranteed construction price remain unsettled.
OpenAI plans to acquire Pierre Computer and integrate its Code Storage service into Codex. The purchase would extend an existing supplier relationship into code infrastructure, but its terms, customer economics and relationship to OpenAI’s own Pando system remain unresolved.
OpenAI says India leads ChatGPT’s markets by weekly active users. Earlier free offers, strong app engagement and planned enterprise deployments explain the stakes, but the company has not disclosed comparable paid-user or revenue figures.
A national exit poll found broad opposition to new local data centers and a Democratic tilt among voters who prioritized AI. But the issue ranked below four other tested concerns, industry-backed candidates also won, and the next fight is over siting and power costs.
