Microsoft vs. Amazon stock: who can fund the AI buildout?

Microsoft and Amazon are both spending heavily on infrastructure, but their shareholders are funding very different cash-flow positions. Over the twelve months ended June 30, 2026, Microsoft generated roughly $67.0 billion after cash purchases of property and equipment. Amazon reported a $7.6 billion outflow on its corresponding free-cash-flow measure.

For investors comparing Microsoft vs. Amazon stock, that difference matters more than a contest between cloud growth rates. It measures how much room the parent company has after paying cash for capacity. It does not establish which cloud platform earns the better return, or which share is cheaper.

Microsoft has the stronger cash cushion on this measure. Amazon needs more of its investment to translate into future cash generation to close the gap. That is the distinction worth carrying into a portfolio decision.

Microsoft vs. Amazon cash flow: match the same twelve months

Microsoft’s fiscal year and Amazon’s calendar year end at different times. Comparing their annual reports directly would mix periods. This comparison instead uses the twelve months ending June 30, 2026, for both businesses.

The inputs come from Microsoft’s fiscal 2026 results and the free-cash-flow reconciliation in Amazon’s June-quarter filing. Amounts below are in U.S. dollars, rounded from figures reported in millions.

• Microsoft: $182.935 billion of operating cash flow less $115.948 billion of cash additions to property and equipment leaves $66.987 billion, calculated by TECHi.
• Amazon: $161.403 billion of operating cash flow less $169.007 billion of property and equipment purchases, net of sales proceeds and incentives, leaves reported free cash flow of negative $7.604 billion.
• Cash spending coverage: Microsoft used 63.4% of operating cash flow for the cash investment measure above; Amazon used 104.7%. These percentages are TECHi calculations.

Put differently, Microsoft retained about $36.60 for every $100 generated by operations after that spending. Amazon spent roughly $104.70 for every $100 generated. The difference is financing flexibility, before other investing and financing uses of cash.

TECHi’s Microsoft financials page provides the quarterly cash-flow series behind the annual view. The Amazon financials page provides operating-cash-flow history; the filing’s reconciliation supplies the net investment measure needed here. Comparing operating cash alone would miss how much of it is being committed.

Why this is not an Azure vs. AWS profitability ranking

The company with the larger cash remainder does not necessarily have the more productive new data center. These statements combine established businesses, working-capital movements and investments made at different stages of development.

Amazon’s second-quarter results show why a negative cash remainder is not the same as a weak cloud franchise. AWS produced $42.2 billion of quarterly sales and $16.6 billion of operating income. Amazon attributed the increase in cash capital investment primarily to artificial intelligence.

Operating income and cash after investment answer different questions. One measures revenue less accounting expenses over a period. The other subtracts cash paid for assets that may serve customers for years. A profitable business can consume cash while expanding capacity. The investment case depends on what that capacity eventually earns.

There is no clean Azure-to-AWS cash-return calculation in these consolidated totals. TECHi’s examination of Microsoft’s Azure revenue and margin disclosures explains the separate disclosure problem. Investors should resist turning a parent-company cash comparison into a claim about the economics of one division.

Microsoft’s cash buffer still has demands on it

Microsoft’s positive remainder is useful, but it is not a sum management can spend twice. In the same fiscal year, cash dividends and share repurchases totaled about $48.7 billion. Subtracting those payments from the calculated $67.0 billion remainder leaves approximately $18.3 billion before other investing and financing flows.

That arithmetic gives the buffer a practical scale. A company can support substantial infrastructure spending and shareholder distributions while still having less room than its headline operating cash flow suggests.

The direction also deserves attention. Microsoft’s calculated cash remainder fell from about $71.6 billion a year earlier even as operating cash flow rose. Capacity spending absorbed the increase and more. Strong collection from customers does not automatically produce a rising amount available after investment.

For a shareholder, the question is how long cash spending needs to outgrow the operating cash it supports. Continued investment can be rational when the resulting capacity earns attractive returns. Repeatedly accepting a smaller remainder without examining those returns would be a much weaker investment discipline.

Amazon’s negative free cash flow is an investment hurdle

Amazon generated substantial operating cash over the matched period. Its negative free cash flow says that the specified cash investment exceeded those inflows; it does not, by itself, establish a liquidity crisis.

Nor does it justify dismissing the spending as temporary and assuming a rebound. A recovery requires some combination of stronger operating cash generation, slower cash investment, or both. Faster revenue growth alone is insufficient if each expansion keeps pulling forward another round of spending.

An investor who favors Amazon should be able to describe that transition. What evidence would show that installed capacity is beginning to generate cash faster than new capacity consumes it? How much continuing investment would count as maintenance of the business rather than an optional expansion?

Those are monitoring questions, not forecasts. The reported figures do not identify the date when Amazon’s cash flow will turn positive or quantify the eventual return on its AI investment.

Cash capex leaves out part of the obligation

The comparison deliberately uses cash purchases from each company’s statements. It is not a measure of all infrastructure commitments.

Amazon nets sales proceeds and incentives against purchases. Microsoft’s line is additions to property and equipment. Financing arrangements can shift the timing and classification of cash payments, so two companies acquiring comparable assets may show different cash-capex profiles.

Amazon also cautions that its free-cash-flow measure excludes items such as debt principal repayments and cash acquisitions. Treating the remainder as unrestricted surplus would ignore those demands.

A fuller funding assessment therefore needs the entire cash-flow statement, existing liquidity, debt maturities and lease obligations. The cash-spending ratio is a useful entry point because it is transparent and reproducible. It is not a substitute for that broader work.

A 10% cash-flow stress shows the funding difference

A simple sensitivity test makes the funding issue tangible. Suppose operating cash flow were 10% lower than the reported twelve-month amount, while the specified cash investment stayed unchanged. This is an illustrative calculation, not company guidance or a prediction of weaker demand.

Microsoft would generate approximately $164.6 billion from operations and retain $48.7 billion after the same cash property and equipment additions. Amazon would generate about $145.3 billion and have a $23.7 billion shortfall against the same net purchases. Each result comes from multiplying reported operating cash flow by 0.9 and subtracting the investment figure already shown.

The exercise does not say that either company must borrow those amounts. Existing liquidity, asset sales, payment schedules and adjustments to spending would affect the actual financing decision. It shows why equal percentage changes in cash generation need not create equal pressure on capital allocation.

The same logic works in the other direction. Holding Amazon’s reported cash investment fixed, operating cash flow would need to rise about 4.7% from the reported level just to bring this free-cash-flow measure to zero. That is an arithmetic threshold, not a forecast: additional spending would move it higher, and a reduction in spending would move it lower.

For Microsoft, even the stressed positive remainder would be roughly equal to the year’s cash dividends and repurchases. The apparent cushion therefore depends partly on which commitments an investor includes. Flexibility comes from the combination of cash generation and choices about uses of cash, not from one favorable subtotal.

Valuation needs a separate comparison

A cash-flow winner is not automatically the investment winner. To compare prices sensibly, an investor must decide which cash earnings are sustainable and what future reinvestment they require. Using the present cash remainder as a permanent earnings base would assume away the spending cycle this comparison is trying to examine.

A negative denominator also makes a simple price-to-free-cash-flow multiple unhelpful for Amazon over this period. Replacing it with operating cash flow avoids that arithmetic problem but ignores the investment bill. Any valuation based on a recovery should disclose its assumed spending path, timing and required return rather than hide those judgments in a single multiple.

Which stock fits the investment case?

Microsoft has the clearer case for an investor prioritizing internally funded expansion. Its matched-period cash generation covered the cash investment measured here with a substantial positive remainder. That is an observable advantage, although it does not establish an attractive purchase price.

Amazon asks the investor to accept more dependence on future investment payoffs. Its operating business generates cash, but the current spending pace exceeds it on the company’s reported measure. The argument improves when cash generation catches up without undermining growth.

Neither conclusion settles valuation. A strong business can be an expensive stock, while a company in a heavy investment phase can offer value if the price adequately reflects the risk. This comparison does not assign a buy rating or use a current share-price multiple.

The next useful comparison should repeat the same twelve-month arithmetic and then investigate what changed. A narrowing gap caused by better operating cash generation means something different from one caused by delaying capacity purchases. Shareholders need that distinction before treating a higher free-cash-flow number as proof of better economics.