How to Read Big Tech Earnings: What to Look for Beyond Revenue Growth

For US-listed companies, filings are available free through the SEC’s EDGAR database, and most companies publish their earnings releases, presentations and call details on their investor relations websites. The chief executive and chief financial officer must provide the required certifications for Forms 10-K and 10-Q.

If you are more familiar with ASX reporting, the main difference is frequency. Many Australian companies report half-year and full-year results rather than quarterly results. Appendix 4D covers half-year results, while Appendix 4E provides preliminary final results. Together with the annual report, these disclosures serve a similar purpose to US earnings materials, although the reporting requirements and formats differ.

Read the Segments, Not Just the Totals

Group results combine businesses with different growth rates, margins and capital requirements. Looking only at consolidated revenue and profit can obscure where earnings are generated and where investment is concentrated.

Amazon illustrates why segment analysis matters. In the June 2026 quarter, AWS generated USD 16.6 billion of the group’s USD 27.5 billion in operating income, despite recording USD 42.2 billion in revenue against total group sales of USD 200.6 billion. AWS therefore generated about 60% of group operating income from roughly 21% of revenue.



When you work through the segment table, look at:

• Growth. How fast is each segment growing, year on year and against the previous quarter, and is that pace picking up or slowing? Where a company gives constant currency growth, use it, since exchange rates can distort the trend.
• Profit. Check operating income and margin by segment. Growth is worth less if each extra dollar of sales brings in less profit.
• Mix. A fast growing, lower margin business can lift revenue while pulling group margins down.
• Definitions. Companies sometimes shuffle products between segments, which can make year on year comparisons misleading. Some headline figures, like Microsoft Cloud, also cut across several segments rather than being one.

Key question: Which parts of the business are driving growth, and are they getting more profitable as they grow?

The AI Investment Checklist

AI infrastructure is the biggest spending commitment these companies have ever made. The costs and the payoff show up in different parts of the accounts, and at different times. These are the five places worth checking.

1. Capital expenditure

     This is the clearest read on what the build out is costing. You'll find it in the cash flow statement as purchases of property and equipment.

• Compare what was actually spent with what management guided to last time, and with any update this quarter.
• Keep cash capex separate from equipment acquired through finance leases and from future commitments. Those may not appear in the capex line at all.
• Listen to how management explains the spending. Is it meeting demand customers have already signed up for, or building capacity for demand they expect later?

2. Free cash flow

     Profits can rise while cash generation falls, because capex goes out the door upfront. Amazon's last result shows how this plays out. Operating cash flow over the 
     trailing twelve months rose 33% to USD161.4 billion, but free cash flow swung to an outflow of USD7.6 billion. The main reason was a USD66.1 billion jump in spending 
     on property and equipment compared with a year earlier.

• The simple version: free cash flow is operating cash flow minus capital expenditure.
• Compare it with net income. A large gap that keeps showing up is worth understanding.
• Check how each company defines it. Some publish more than one measure, including versions that take out finance lease repayments, so take care comparing across companies.

3. Depreciation and useful lives

     Capex reaches the income statement slowly, through depreciation spread over the years a company expects each asset to last. So the assumed life of servers and 
     network gear matters. Change the assumption and reported profit moves, even though the cash spent is exactly the same.

• Find the accounting policy note on property and equipment, and see what useful lives are assigned to servers, networking equipment and data centres.
• Look for any change to those estimates and the dollar impact disclosed. A longer life means lower depreciation and higher reported profit. A shorter one does the reverse.
• Expect depreciation to keep climbing for a few years after a spending surge, even if capex itself levels off.

4. Remaining performance obligations and commitments

     Remaining performance obligations (RPO) are revenue a company has under contract but hasn't yet recognised. A growing balance points to demand, but it isn't 

     revenue yet, and it's not cash in the bank either.

• See how much of it is expected to turn into revenue over the next twelve months.
• Think about contract length and cancellation terms. Each company also defines its backlog a little differently.
• Read the notes on data centre leases and purchase obligations. They show what the company has already committed to spend.

5. Is AI actually making money?

     User numbers, product launches and adoption figures are encouraging, but none of them prove the investment is paying off.

• Look for disclosed AI revenue, paid usage and what management says about cloud demand.
• Be careful with run rate figures and other AI metrics. Management defines them, they're usually unaudited and they differ from company to company, so they can't be compared directly.
• Work out whether AI is winning new customers, getting existing ones to spend more, or mostly protecting market share.
• See whether management offers any evidence of returns on the money being spent, or just describes activity.

The central question is whether AI investment is building lasting value or whether spending is running ahead of the evidence of returns.

Earnings Quality: What Adjusted Results Can Hide

Net income can swing a long way for reasons that have nothing to do with how the business performed. Gains on investments are the obvious example right now. In the June quarter, Alphabet booked a USD98.0 billion net gain in other income, mostly unrealised gains on equity holdings. Amazon's net income included USD53.4 billion of other income before tax, mainly from its investments in Anthropic. Amounts like these can be bigger than a quarter's operating profit, so earnings per share on its own can tell you very little.

Things to check every time:

• GAAP versus non-GAAP. Read the reconciliation and see what's been left out of the adjusted figures. Companies treat items differently. Microsoft, for instance, strips the impact of its OpenAI investment out of its non-GAAP results.
• Stock based compensation. It's a real cost to shareholders even when it's excluded from adjusted earnings, and it adds to the share count.
• Unusual items. Restructuring costs, legal settlements and tax effects can make a single quarter look better or worse than it really was.
• Investment gains and losses. Take fair value movements on shareholdings out before you judge how the quarter went.
• Buybacks and dilution. Check whether buybacks are actually shrinking the share count or mostly soaking up new shares issued to staff.

Key question: Is reported profit backed by how the business is actually running, and by cash?

Guidance, Consensus and the Call

What management says about the next quarter often matters more than the quarter just gone. Results are measured against what investors were already expecting, so a company can post solid growth and still see its shares fall if the outlook disappoints.

• Pick one consensus source and stick with it. Estimates differ between data providers, and switching between them can turn a beat into a miss.
• Read the guidance carefully. Look at next quarter revenue and margins, and full year capex plans. Note any change from last time and how management explains it.
• Listen to the Q&A, not just the prepared remarks. The questions show what the market is focused on. Pay attention to comments on cloud demand, pricing, capacity, AI revenue and spending, and to how straight the answers are.
• Tell supply problems apart from weaker demand. Cloud growth slowing because a company can't get capacity online fast enough is a very different story from growth slowing because customers are spending less.

A beat doesn't guarantee the share price goes up. The reaction depends on results against expectations, the guidance, where the valuation started and how investors were positioned going in.

Key question: Does the outlook match, beat or fall short of what the market already expects?

Applying This to Alphabet, Microsoft and Amazon

Here's the framework applied to each company reporting next week. It shows where an analyst would start, and what deserves a closer look against prior results and expectations.



These are areas to investigate, not conclusions about the companies or their shares. Each result should be assessed against previous performance, current guidance and market expectations.

The Big Tech Earnings Checklist

Keep this handy and work through it for each company as the results come in.

• Revenue growth by segment, year on year and quarter on quarter
• Cloud growth and segment operating margins
• Group gross and operating margin trends
• Evidence of AI monetisation, and how each metric is defined
• Capital expenditure against previous guidance
• Operating cash flow, free cash flow and the gap to net income
• Depreciation policy and any change to useful lives
• Remaining performance obligations and contractual commitments
• GAAP versus non-GAAP reconciliation, including investment gains
• Stock based compensation and share count
• Guidance against consensus from one consistent source
• Management's answers in the Q&A
• Results against expectations

Look Past the Headline Numbers

A beat or miss is where the analysis begins, not where it ends. Revenue growth, segment profitability, capital spending, cash flow, earnings quality and guidance need to be considered together.

For Big Tech, the test is whether investment in AI can generate enough additional revenue and profit to justify the cost. The next set of results will offer more evidence, but the longer-term picture will depend on how effectively each company turns that spending into sustainable business growth.