On the evening of June 1, 2026, Alphabet (NASDAQ:GOOGL | GOOGL Price Prediction) priced one of the largest equity offerings in corporate history: an upsized $84.75 billion raise spanning Class A common, Class C capital stock, and mandatory convertible preferred shares, alongside a $10 billion private placement from Berkshire Hathaway. The stock fell 6% over the next month. Four days later, Meta shares dropped 5% to 7% after a Financial Times report that the company was weighing a multi-billion-dollar share sale to fund 2026 capex of $125 billion to $145 billion. Hyperscalers raise equity when they have to.
The Crossover
Research firm Epoch AI has put a number on the squeeze. Across the major AI builders, capital expenditures are growing roughly 70% per year while operating cash flow grows roughly 23% per year. In Epoch’s model, aggregate free cash flow across the group crosses zero around Q3 2026. The individual timelines diverge sharply: Oracle has already crossed, Amazon is crossing now, Alphabet hits the line around Q1 2027, Meta around Q3 2027, and Microsoft around Q3 2028.
Oracle (NYSE:ORCL) is the cautionary tale. Free cash flow for fiscal 2026 came in at negative $23.69 billion. CFO Hilary Maxson told analysts on June 10 that Oracle “expects to raise around $40 billion in debt and equity in our fiscal year 2027, and that includes our already announced $20 billion at-the-market equity issuance.” Oracle raised $43 billion in debt in fiscal 2026 alone.
The scale is what makes this different from prior capex cycles. The Big Four of Alphabet, Microsoft, Meta, and Amazon are projected to pour more than $700 billion into capex in 2026, with Wall Street estimates topping $1 trillion in 2027. Combined, the group is expected to spend roughly 90% of operating cash flow on capex in 2026, up from about 65% in 2025. Amazon (NASDAQ:AMZN) already tipped its hand: Q1 2026 operating cash flow was $26.0 billion against $44.2 billion in capex, and long-term debt has climbed to $119.1 billion.
How Hyperscalers Are Funding the Unfundable
When capex overwhelms operating cash flow and equity raises dilute shareholders, hyperscalers are turning to a toolkit of financing structures that move debt off balance sheets and shift capex to operating leases.
Off-Balance-Sheet Vehicles and Joint Ventures. Microsoft created a highly-leveraged $100 billion off-balance sheet special purpose vehicle in September 2024, the AI Infrastructure Partnership, which includes $70 billion in fund-level debt with data centers and power infrastructure assets as collateral, with BlackRock, Global Infrastructure Partners, and MGX in equity and targeting $30 billion total equity. Hyperscalers have turned to off-balance-sheet arrangements through special purpose vehicles and joint ventures, often in partnership with private credit firms. These structures involve a dedicated vehicle that acquires data center assets, capitalized with equity from multiple sponsors, while the hyperscaler holds a minority stake and commits to long-term operating leases or capacity offtake agreements, keeping associated debt off the hyperscaler’s balance sheet.
Finance Leases and Long-Dated Commitments. Moody’s found that lease commitments not yet on hyperscaler balance sheets total 113 percent of their most recent adjusted debt, with the firm expecting material increases in adjusted debt and lease-related cash outflows in coming years. Long-dated operating leases, power purchase agreements, joint ventures, and capacity reservation contracts embed substantial fixed obligations off balance sheet, with roughly 70-75% of new data center capacity pre-leased under long-term contracts.
Customer Prepayment and Bring-Your-Own-Hardware Models. Oracle’s prepaid and customer-supplied hardware portions of large AI contracts now total $75 billion, substantially reducing the amount of capital Oracle must raise to build out its AI datacenters. This structure shifts the capex burden to customers (like OpenAI) who either prepay for GPUs or supply them directly. Oracle signed $67 billion in AI infrastructure contracts in Q4 FY2026, the majority of which was either bring-your-own-hardware or prepaid, with those contracts carrying no degradation in margin compared to other contracts.
Internal Redeployment and Portfolio Cuts. Meta has identified potential to cut Reality Labs spending by 30 percent, which could free up $56 billion for redeployment toward AI. Reality Labs’ budget was cut by 30 percent in rolling layoffs throughout 2026, with the division absorbing cumulative losses exceeding $90 billion since Meta began breaking out the unit. Meta’s pivot from metaverse to AI infrastructure has been swift: the company identified potential 30 percent Reality Labs cuts, freeing tens of billions for redirection toward AI data centers and custom silicon.
Morgan Stanley expects hyperscaler borrowing to top $400 billion this year, more than double 2025’s $165 billion. Hyperscalers have engaged in substantive rounds of investment-grade debt issuance, private credit, special project vehicles, and leasing structures, with Meta, Alphabet, Amazon, and Oracle’s collective weighting in the Bloomberg U.S. Corporate IG Index nearly doubling from 2.2 percent to 4.1 percent over the year ending April 1, 2026. Capital markets are now financing the buildout, with retained earnings no longer sufficient. The bond market remains willing—for now.
Why NVIDIA Sits on the Other Side of the Trade
NVIDIA (NASDAQ:NVDA) is collecting cash at an extraordinary clip. Operating cash flow hit $102.7 billion in fiscal 2026, and the company authorized an $80 billion buyback. If Microsoft, Amazon, Alphabet, and Meta Platforms hit financing limits and trim orders, NVIDIA’s demand curve bends with them. That is why the stock is down 8% over the past month even after a blowout Q1 report.
The Caveat That Matters
Epoch’s model assumes stagnant productivity gains from the capex. It does not account for whether the AI investments themselves accelerate cash flow growth. If Azure, AWS, and Google Cloud convert these GPU farms into durable revenue at the margins Satya Nadella keeps describing (the AI business now runs at a $37 billion annual rate, up 123%), the crossover timeline gets pushed out or never arrives. The number to watch next quarter is straightforward: free cash flow at Amazon, Alphabet, and Meta, and whether the bond market keeps absorbing hyperscaler paper at investment-grade spreads. Hyperscalers have committed to off-balance-sheet obligations nearing $1 trillion, primarily in lease obligations and chip purchase commitments, which represent contractual fixed payments that may impair balance sheet strength if monetization of AI infrastructure stalls. If either breaks, the AI capex cycle stops being a story about ambition and becomes one about access.
Top Gaining Stocks
Top Losing Stocks