Wall Street goes into Wednesday with a simple question and no simple answer. Amazon, Microsoft and Alphabet report on October 28, and Reuters reported on October 27 that investors expect faster cloud growth from all three, even as a turbulent selloff and rising costs of AI financing dampen the mood. This year’s Big Tech earnings season has become a stress test for the AI boom, and the market is grading it on spending as much as on sales.
The setup is lopsided. The three companies reportedly carry nearly $1.7 trillion in combined backlog and plan to spend as much as $600 billion in 2026. Meta Platforms already showed what happens when spending outruns patience, with its shares falling 7% after its report. If you want the coding-tool side of the same week, our earlier piece on Big Tech earnings, Copilot and Gemini Code Assist covers it. This article looks at the cloud numbers, the financing squeeze and the model pricing war underneath them.
One caution before the numbers. Nobody can say in advance that these reports will prove or disprove the AI boom, and nothing here claims otherwise. What the figures can do is show where the pressure sits, which lines investors will read first and how the market has already reacted to the first AI-heavy reports of the season.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What Wall Street Expects From Big Tech Earnings on October 28
Amazon, Microsoft and Alphabet were all expected to report on Wednesday, October 28, 2026. According to Reuters, the consensus calls for faster cloud growth at every one of them. That alone would be a clean story in a calm market. It is not a calm market.
Reuters described a selloff and worries about the rising cost of financing AI infrastructure as the two forces tempering optimism. Higher U.S. interest rates and rising bond yields were adding pressure on AI cloud providers, with CoreWeave named among them. In plain terms, building data centers costs more to fund than it did a quarter ago, and investors know it.
That shifts what counts as a good quarter. A company can post strong cloud growth and still see its stock fall if the capital expenditure guide jumps faster than the revenue guide. Meta’s 7% drop is the fresh example. Treasury Partners chief investment officer Richard Saperstein put the bar this way: “We generally expect tech earnings to come in line with or modestly above expectations, as the bar is rising for tech companies to deliver against a backdrop of increasing questions about the return on investment from their AI spending.”
The Numbers Behind the Test: $1.7 Trillion Backlog, $600 Billion Spend
Start with the two largest figures in the story. Reuters reported that Amazon, Microsoft and Alphabet together hold nearly $1.7 trillion in backlog, which is contracted revenue not yet recognized. The same three companies planned to spend as much as $600 billion in 2026.
Divide one by the other and the backlog covers roughly 2.8 times the planned 2026 outlay. That is this article’s own arithmetic, not a reported ratio, and it comes with a catch. Backlog converts to revenue over several years, while the spending lands now. The gap in timing is exactly why investors ask about return on investment.
A second data point widens the frame. Hyperscalers as a group were expected to spend more than $750 billion on capital expenditures in 2026, roughly double their 2025 spending. The $600 billion and $750 billion figures cover different groups of companies, so they should not be added or compared line by line. Together they show the scale: a build-out that doubled in a year.
| Metric | Reported or expected figure | Context |
|---|---|---|
| Combined backlog (Amazon, Microsoft, Alphabet) | Nearly $1.7 trillion | Reported ahead of Oct. 28 results |
| Planned 2026 spending (same three) | Up to $600 billion | Reported ahead of Oct. 28 results |
| Hyperscaler capex, 2026 | More than $750 billion | Roughly twice 2025 spending |
| Google Cloud Q3 revenue | $29.57 billion | Analyst expectation, 95.1% growth |
| AWS revenue growth | 38.2% | Versus 36.8% in the prior quarter |
| Azure growth guidance | 45% | Versus 42% in the prior quarter |
| Alphabet total revenue | $128.18 billion | Expected up 25.2% |
| Google Search revenue growth | 14.7% | Versus 16.8% in the prior quarter |
Cloud Growth Expectations: Three Clouds, Three Speeds
Cloud is where the AI story either shows up as revenue or does not. Each of the three providers enters the quarter with a different kind of expectation, and the differences matter more than the shared headline of acceleration.
Google Cloud: the 95.1% Outlier
Analysts expected Google Cloud to report September-quarter revenue of $29.57 billion, a 95.1% jump, according to Reuters. That would make it the fastest-growing of the three cloud businesses by a wide margin. A growth rate near double leaves little room for a miss, because the stock will already price in a lot of good news.
The other half of Alphabet’s report carries a different tone. Reuters said Google Search revenue growth was projected to slow to 14.7%, down from 16.8% in the prior quarter. Alphabet’s total revenue was expected to rise 25.2% to $128.18 billion. Investors will weigh whether cloud strength offsets a cooling search business.
AWS: Faster, but From a Larger Base
Amazon Web Services was expected to post revenue growth of 38.2%, compared with 36.8% in the prior quarter. The step up is small, about 1.4 percentage points. Yet AWS is the largest of the three cloud businesses, so even a modest acceleration implies a big absolute dollar gain.
Azure: Microsoft Set Its Own Bar
Microsoft had forecast Azure revenue growth accelerating to 45% from 42% in the prior quarter. Unlike Google and Amazon, this is company guidance rather than an analyst estimate, which makes it a harder target to miss without explanation. Management gave the number, and the market will hold it to that number.
Why Financing Costs Now Matter as Much as Revenue
For two years, the AI trade rewarded growth and forgave cost. Reuters now describes a market where the cost of money is a headline variable. Higher U.S. interest rates and rising bond yields raise the price of the debt that funds data centers, chips and power contracts.
The pressure shows up first in the smaller, more leveraged names. CoreWeave, an AI cloud provider, was cited by Reuters among the companies feeling the squeeze from higher yields. For a look at how that squeeze runs through the neocloud tier, see our breakdown of the AI funding crunch and the pressure on neoclouds.
The same Reuters report said the cost of insuring Meta’s debt reached record levels, a sign of investor concern about AI-related spending and leverage. It also noted that the market stress had led some companies to delay planned initial public offerings. Ken Griffin, the Citadel founder, was among the investors warning about possible excesses in the AI market.
None of that says AI demand is weak. It says the funding side of the trade has tightened while the spending plans have not. If Wednesday’s guidance shows capital expenditure rising again, the market will ask who pays for it and at what rate.
Meta’s 7% Drop Sets the Template for This Week
Meta Platforms reported before the cloud trio, and the reaction became the reference point. Its shares fell 7% after the earnings report. The company also reported negative free cash flow for the first time since its 2012 initial public offering, a milestone nobody at Meta wanted to hit.
Free cash flow is the money left after capital spending, and negative means the company spent more on its business than it generated. Meta has been funding an aggressive AI build-out, and the new report put a number on the cost. Our earlier coverage of Meta’s reported $97.5 billion equity raise lays out how the company has been approaching the funding question.
Meta differs from the other three in one key way. It sells no cloud capacity, so its AI spending has no direct cloud revenue line to point to. Amazon, Microsoft and Alphabet can show customers renting that capacity and a backlog that stretches forward. Meta has to argue the payoff through advertising and products, which is a harder case to make in a quarter.
Nasdaq’s Biggest One-Day Drop Since June
The market context is not decoration. The Nasdaq Composite recorded its largest one-day decline since June, and the selloff was tied to revived doubts about AI. The Globe and Mail’s market wrap, titled Most stock indexes end sharply lower as revived AI doubts ignite tech selloff, tied the move to Meta and to heavy buying of U.S. Treasuries.
Bonds rising as stocks fall is a classic flight to safety. It also pushes yields down, which is the one piece of this picture that could help AI financiers. The tension is that investors fled for safety precisely because they doubted the spending story. That is a poor backdrop for companies about to raise their capital expenditure guidance.
Business Insider framed the next two weeks as a crucial stretch, with AI earnings arriving alongside a Fed meeting and the midterm elections, in its piece AI Earnings, the Fed, Midterms: Stocks Are Kicking Off Crucial 2-Weeks. Investopedia’s market update added the calendar: Alphabet, Amazon and Microsoft this week, Apple the week after and Nvidia a bit later. Meta, Tesla and SpaceX have already reported.
What the CEOs Have Said About the Bubble Debate
The executives running these companies have addressed the bubble question directly during 2026. Kiplinger collected their comments, and they split along predictable lines. The cloud sellers sound confident. The one who does not sell cloud sounds the most cautious.
Amazon CEO Andy Jassy, in comments dated April 9, 2026, said he had followed the debate over whether AI is over-hyped, whether it is a bubble and whether margins and return on invested capital will be appealing. His verdict: “My strong conviction, at least for Amazon, is that the answers are no, no, and yes.” That is a flat statement, and Wednesday’s AWS numbers will test it.
Microsoft CEO Satya Nadella offered a more conditional view. He said a telltale sign of a bubble would be if all anyone discusses is the technology firms, and that avoiding a bubble requires the benefits to be “much more evenly spread.” He also warned about “unsustainable silliness of random, new neocloud popping up,” claiming compute build-outs the revenue cannot support. Notice who he targets: smaller infrastructure players, not the hyperscalers.
Meta founder and CEO Mark Zuckerberg gave the most candid assessment. He said there are compelling arguments for why AI could be an outlier, and that if model capability and demand keep growing there may be no collapse. Then he added that there is “definitely a possibility” based on past large infrastructure build-outs that something like a bubble could happen here. Meta’s shares then fell 7% on its own report.
What Investors Say: Patience Is Running Out on Capex
Institutional voices have grown blunter. John Belton, a portfolio manager at Gabelli Funds, said, in comments carried by the Los Angeles Times, that there is “Absolutely no patience right now for higher capex.” The quote ran in the paper’s October 26 story, Big Tech’s strong earnings may not be enough to ease fears over AI spending.
The headline of that story carries its own message. Strong earnings, it argues, may not be enough. Treasury Partners’ Saperstein reached a similar place from another direction, expecting results in line with or modestly above expectations while the bar for delivering on AI return keeps rising.
Put the two together and the trade is clear. A beat on revenue is table stakes. The stock reaction will depend on what each company says about 2027 spending, how it describes returns on the capacity already built and whether the financing story stays calm. Investors have stopped treating capex as a sign of ambition and started treating it as a cost to be justified.
The Model Race Running Under the Earnings
Cloud revenue only grows if customers keep buying model access and compute, and the model market has been moving fast. One release tracker counted 150 AI releases from 22 companies in October 2026 as of October 27, after 213 releases in September. October is already trailing September’s total with days to spare, yet the pace stays relentless.
The newest generation matters here because each family sets what customers pay per token. Trackers list Anthropic’s Claude Opus 5.5 from September 22, Claude Sonnet 5.5 from September 28 and Claude Haiku 5.5 from October 7. OpenAI shipped GPT-6.1 Sol on September 29 and GPT-6.1 Terra on October 15. Google released Gemini 4 Argon on September 30. For a side-by-side of the frontier tier, read our GPT-6.1 Sol vs Sonnet 5.5 vs Gemini 4 Argon comparison.
| Model | Maker | Release date | Listed price per 1M tokens (input/output) |
|---|---|---|---|
| Claude Opus 5.5 | Anthropic | Sept. 22 | $4 / $20 |
| Claude Sonnet 5.5 | Anthropic | Sept. 28 | $2 / $10 |
| GPT-6.1 Sol | OpenAI | Sept. 29 | Not listed in sources reviewed |
| Gemini 4 Argon | Sept. 30 | $2 / $10 (1M-token context) | |
| Mistral Large 4 (public preview) | Mistral | Early Oct. | $0.68 / $2.09 |
| Claude Haiku 5.5 | Anthropic | Oct. 7 | $0.10 / $0.50 (prompts up to 100,000 tokens) |
| GPT-6.1 Terra | OpenAI | Oct. 15 | $1.00 / $5.00 (1.1M context) |
| Qwen3.6 27B | Alibaba | Oct. 21 | Not listed in sources reviewed |
| Kimi K3.1 Pro | Moonshot | Oct. 23 | Not listed in sources reviewed |
Prices come from third-party trackers, including the Opper AI model tracker and a monthly LLM news roundup, rather than from the vendors’ own price pages. Treat them as the listed rates on those trackers. They show a wide spread, from $0.10 per million input tokens at the low end to $4 at the top.
How Cheaper Tokens Change the Revenue Math
The pricing table points to a tension that sits beneath the earnings. Providers keep adding capacity at higher cost while the cheapest tier of models gets cheaper. A Haiku-class model at $0.10 per million input tokens gives customers a low-cost route for simple tasks, which can pull volume away from pricier tiers.
That does not make the cloud businesses less valuable. Lower prices tend to raise usage, and usage drives compute demand. But the revenue per unit of capacity falls, and the investors worried about capex will watch that ratio closely. If tokens get cheaper faster than volume grows, the return on a new data center shrinks.
Microsoft’s side of that equation is visible in how it prices third-party and in-house models for enterprise customers. Our look at Microsoft Foundry pricing from Luna to Astra shows how wide the tiers run inside a single platform. Expect analysts to ask how Azure’s 45% growth guide splits between volume and price.
Competitive Comparison: Who Carries the Most Pressure
Each reporting company faces a different test. Alphabet has the highest growth bar and the softest search line. Amazon has the largest base and the smallest acceleration. Microsoft has the most precise public target. Meta has already taken the hit, and Nvidia, which has not yet reported, supplies the chips behind all of it.
| Company | Status as of Oct. 27 | Key item to watch |
|---|---|---|
| Alphabet | Reports Oct. 28 | Google Cloud at $29.57B (+95.1%), Search growth at 14.7% |
| Amazon | Reports Oct. 28 | AWS growth of 38.2% versus 36.8% |
| Microsoft | Reports Oct. 28 | Azure growth against its own 45% forecast |
| Meta Platforms | Reported, shares down 7% | First negative free cash flow since the 2012 IPO |
| Tesla | Already reported | Named in coverage as part of the earnings run |
| CoreWeave | Not on the Oct. 28 slate | Financing pressure from higher yields |
| Apple | Next week, per Investopedia | Follow-on read on consumer tech |
| Nvidia | A bit later, per Investopedia | Demand signal for AI chips |
Our AI and machine learning coverage tracks the model and infrastructure side of this story as it develops. The comparison above is a snapshot, and it will change by Thursday morning.
Historical Context: Capex Cycles and Investor Patience
Big infrastructure build-outs have a pattern. Spending runs ahead of revenue, investors cheer the ambition and then a quarter arrives when they ask for proof. Zuckerberg named this himself, pointing to past large infrastructure build-outs and how they led to bubbles.
What is different this time is the funding structure. The 2026 version leans heavily on debt and financing, which is why the cost of insuring Meta’s debt reaching a record carries weight. When a company’s borrowing costs become a market signal, the build-out has entered a new phase.
The Meta free-cash-flow milestone fits that history. A company that went public in 2012 has now spent past what it earns. Whether that proves a smart bet or a costly one is still an open question, and this article will not pretend to settle it.
Five Predictions for the Earnings Cycle
These are this publication’s judgments based on the reported setup. They are not forecasts of whether AI spending will pay off or whether a bubble will burst, which no one can know yet.
- Capex guidance will move stocks more than cloud growth. With Gabelli’s Belton voicing no patience for higher capex, a company that raises its spending outlook faces a harder reaction than one that holds steady.
- Cloud growth will land near the consensus numbers. Expectations of 38.2% at AWS and 45% guidance at Azure leave little surprise room, so the story will turn on commentary rather than the headline growth rate.
- Analysts will press harder on free cash flow. Meta’s first negative reading since 2012 made the metric a headline, and questions on the calls will follow.
- Cheap model tiers will draw more attention. With low-end pricing as low as $0.10 per million input tokens on trackers, expect questions about revenue per unit of capacity.
- Neocloud financing will stay the weak link. Higher yields hit leveraged providers first, and Nadella’s own warning about neoclouds points the same direction.
The takeaway: this week tests how much patience is left more than it tests whether AI works. Demand looks strong on paper, with a backlog near $1.7 trillion. The harder question is how much the market will pay for the capacity needed to serve it.
Frequently Asked Questions About Big Tech Earnings and the AI Boom
When do Amazon, Microsoft and Alphabet report earnings? All three were expected to report on Wednesday, October 28, 2026, according to Reuters.
How big is the combined backlog at the three companies? Reuters reported nearly $1.7 trillion in combined backlog, against planned spending of as much as $600 billion in 2026.
What cloud growth do analysts expect? Google Cloud is expected at $29.57 billion in September-quarter revenue, up 95.1%. AWS is expected to grow 38.2%, up from 36.8%. Microsoft forecast Azure growth of 45%, up from 42%.
Why did Meta’s stock fall after its report? Meta Platforms shares fell 7% after earnings. The company reported negative free cash flow for the first time since its 2012 initial public offering, which sharpened concern about AI spending.
Do these earnings prove or disprove an AI bubble? No. Nothing confirms that the reports will settle the debate, and predictions about future AI profits or a bubble bursting remain unconfirmed. They will add data points on cloud growth, spending and cash flow.
How did the market react to the AI doubts? The Nasdaq Composite recorded its largest one-day decline since June, and most major U.S. indexes closed sharply lower as AI doubts revived.
Which AI models launched most recently? Trackers list Claude Haiku 5.5 on October 7, GPT-6.1 Terra on October 15, Qwen3.6 27B on October 21 and Kimi K3.1 Pro on October 23. Gemini 4 Argon arrived on September 30.
When do Apple and Nvidia report? Investopedia’s market update said Apple comes the following week and Nvidia a bit later, without giving exact dates.
