Five tech giants' earnings released: Amazon, Alphabet, and Meta post negative free cash flow for the quarter

As of now, five of the "Magnificent Seven" U.S. tech giants have released their latest quarterly earnings, and three of them reported negative free cash flow for the quarter: Amazon, Google parent Alphabet and Meta posted net outflows of about $33.5 billion, $9.1 billion and $2.1 billion, respectively. Microsoft and Tesla, by contrast, generated positive free cash flow of about $18.6 billion and $1.5 billion, respectively. Apple and Nvidia are expected to report their latest quarterly results next month.

After the earnings releases, Microsoft won positive market reviews on the back of better-than-expected cloud business growth and cash flow performance; Google parent Alphabet fell 12.6%, its largest single-day drop since going public; Amazon shares fell nearly 7% in after-hours trading following its earnings release; and Meta and Tesla shares also declined after their earnings reports.

"In the past, the market was willing to pay for AI's growth prospects and technology story, but the latest round of earnings from tech giants means that AI investment is moving from an 'arms race' over the scale of capital expenditure to a stage of testing commercial returns," Zhang Yi, CEO of iiMedia Research, told a Yicai reporter. Free cash flow has become an important indicator for measuring investment safety, but judging whether negative cash flow is dangerous still hinges on real orders, the ability to convert revenue, and sustained financing capacity.

Tech giants show diverging performance

The latest earnings of the three major cloud computing giants show similar growth trends but deliver different cash flow report cards: revenue at Amazon AWS, Microsoft Azure and Google Cloud grew 39%, 48% and 103% year on year, respectively, with AI demand continuing to drive cloud business expansion, but differences in capital expenditure, profit performance and investment payback periods are affecting how the market evaluates each company.

Amazon's latest earnings show third-quarter AWS revenue of $46 billion, up about 39% year on year, the fastest growth in nearly five years, with operating profit reaching $17.8 billion. The company disclosed that spending by Fortune 500 customers on Amazon Bedrock grew nearly sevenfold year on year, and the number of tokens processed by the platform increased nearly 17-fold.

The other side of AWS's rapid growth is massive infrastructure investment. In the 12 months through the end of September, Amazon's free cash flow swung from an inflow of about $14.8 billion in the same period a year earlier to an outflow of $42.357 billion. The company expects cash capital expenditure of about $220 billion in 2026, with a further significant increase in 2027. The after-hours decline in Amazon's share price shows that although cloud business growth exceeded expectations, it still failed to fully offset investors' concerns about cash burn.

Microsoft's performance was different. In the first quarter of fiscal 2027 through the end of September, Azure revenue was about $33.165 billion, up 48% year on year, higher than the 43% growth in the previous quarter. Microsoft's capital expenditure and finance lease-related investment in the quarter totaled about $45.2 billion, below its previous plan of about $50 billion; free cash flow reached about $18.6 billion.

However, Microsoft's capital expenditure coming in below plan was not a deliberate pullback in AI investment, but rather delays in the delivery of some third-party computing facilities that affected the pace of investment in the quarter. The company's commercial remaining performance obligation reached $677 billion, up 73% year on year, providing some contractual support for future revenue; Microsoft expects Azure revenue growth to further reach 49% to 50% in the next fiscal quarter.

Google parent Alphabet showed a more pronounced contrast between business and cash flow. Third-quarter Google Cloud revenue was about $30.7 billion, up 103% year on year; operating profit was about $11.1 billion, with an operating margin of 36.2%. Part of the growth came from external sales of Google's self-developed TPU systems, but the hardware business also requires advance procurement of equipment and inventory building, increasing working capital needs.

Alphabet's capital expenditure in the quarter reached $51.4 billion, free cash flow was negative $9.1 billion, and full-year capital expenditure guidance is $195 billion to $205 billion. The company's cash flow pressure stems not only from AI infrastructure construction but also from factors including increased TPU inventory and the payment of a $5.2 billion EU fine. At the same time, Google's advertising revenue fell short of market expectations, and the company's net profit fell about 25% year on year, mainly due to fair value losses on equity securities investments.

These factors made Alphabet one of the tech giants with the most dramatic market reaction in this earnings season. On October 29, Alphabet's share price fell more than 12%.

For investors, cloud business growth rate is no longer the only evaluation standard; how quickly massive investment can be converted into profit and cash flow is becoming equally important.

Meta's business model differs from the three major cloud providers, but it also faces a balance between investment and returns. The company's third-quarter revenue was $64.6 billion, up 26% year on year; operating profit was about $12.988 billion, down 37% year on year, with the operating margin falling from 40% to 20%. The decline was mainly affected by $9.89 billion in litigation-related expenses and cannot be entirely attributed to AI investment. Excluding that item, Meta's operating profit in the quarter was about $22.88 billion, up about 11.4% year on year.

Meta's operating cash flow in the quarter was about $32.8 billion, capital expenditure was about $34.9 billion, and free cash flow was negative $2.12 billion, compared with positive about $10.6 billion in the same period a year earlier. The company expects full-year 2026 capital expenditure to reach $135 billion to $145 billion.

Meta's AI returns are mainly reflected in improved advertising efficiency, rather than direct sales of computing power. The incremental revenue brought by algorithms has already entered the advertising business, but the research and development of future-oriented models, agents and superintelligence still requires sustained investment.

Tesla's third-quarter revenue was about $28.05 billion, essentially flat compared with the same period a year earlier; net profit attributable to ordinary shareholders was about $790 million, down 43% year on year, and free cash flow was about $1.53 billion. The company also adjusted its 2026 capital expenditure forecast to more than $20 billion from its previous expectation of more than $25 billion.

Tesla is still continuing to invest in AI computing power, robots and manufacturing facilities. In addition, the potential merger between Tesla and SpaceX that is under evaluation further increases the complexity of its capital allocation.

What will the next round of competition be about?

Amazon expects capital expenditure of about $220 billion in 2026, Alphabet expects $195 billion to $205 billion, Meta expects $135 billion to $145 billion, and Microsoft has slightly lowered its figure to $170 billion. The scale of capital expenditure at these companies has reached unprecedented levels, and the industry expects it to reach $1.3 trillion next year.

Research published this year by S&P Global Ratings forecasts that the six hyperscale companies it covers (Alphabet, Amazon, Microsoft, Meta, Oracle and SpaceX) will post negative free operating cash flow in 2026 and 2027, with cash flow expected to recover only by 2029. The agency also lists the ability to monetize AI investment, the sustainability of demand, the risk of overcapacity, and debt-like contractual commitments as key areas for monitoring.

However, negative free cash flow does not directly mean investment failure. Data centers, AI servers and network equipment usually require concentrated cash payments during the construction and procurement stages, and costs are then gradually recovered over many years through cloud services, model inference and enterprise applications. An AI industry practitioner told the reporter that negative free cash flow in a single quarter or even several consecutive quarters may reflect an investment cycle and does not necessarily mean the business model is unsustainable. What really needs to be observed is whether future revenue growth can cover equipment depreciation, operating costs, financing expenses and the next round of hardware upgrades.

"The key to the next round of competition is not how much computing power is invested, but whether computing power can be converted into profit on the financial statements," Zhang Yi said. Unlike the previous development stages of the internet and mobile internet, in the AI heavy-asset era two indicators deserve particular attention: first, order visibility, such as whether remaining performance obligations (RPO) and backlog orders can provide support for future revenue; second, capital discipline, that is, whether companies can maintain the cash flow bottom line while sustaining necessary investment, ultimately forming a positive cycle from computing power investment to product and application implementation to commercial monetization.

Even with orders, tech giants still need to solve funding and asset management problems. Companies with stable cash flow, relatively high credit ratings and long-term customer contracts can obtain funding at relatively low cost and lock in data center, power and chip resources in advance; but the more complex the financing structure, the more cautiously investors need to assess a company's true leverage and long-term commitments.

Technological iteration may also change investment returns. AI chips and servers are updated quickly, and new-generation products are often significantly better than previous-generation equipment in performance and energy efficiency. For companies investing tens of billions of dollars to build data centers, computing power utilization, the actual economic life of equipment and depreciation periods will all affect future profits; if hardware updates faster than expected, companies may also face greater depreciation or asset impairment pressure.

The market is not uniformly pessimistic about AI investment returns. The rapid growth of AWS, Azure and Google Cloud has already shown that enterprise demand for computing power is still expanding. As model inference and the use of enterprise AI applications increase, existing infrastructure may improve returns by raising utilization. Microsoft sells AI services through Azure and enterprise software, Amazon relies on AWS and Bedrock to provide computing power and model services, Alphabet is advancing both cloud services and TPU commercialization, while Meta mainly uses AI to improve advertising efficiency. The investment payback periods of different business models are not the same, but these businesses have already provided real demand support for massive infrastructure investment.

Next, Nvidia will release its new quarterly financial report in November, providing a new basis for observing upstream demand for computing power.