The latest technology earnings did not produce one artificial-intelligence trade. They produced several businesses with similarly impressive growth and very different amounts of cash left after building the infrastructure behind it.
Alphabet increased second-quarter revenue 24% and Google Cloud revenue 82%. Microsoft reported 18% quarterly revenue growth and 43% growth in Azure and other cloud services. Meta’s revenue rose 28%. Amazon’s AWS sales grew 37%, its fastest pace in 18 quarters.
Those are powerful numbers. They are also only the first line of the investor test.
In the same quarter, capital spending absorbed about 115% of Alphabet’s operating cash flow, 98% of Meta’s and 65% of Microsoft’s. Amazon reported that trailing-12-month free cash flow had moved from an $18.2 billion inflow to a $7.6 billion outflow, primarily because property-and-equipment purchases increased with AI investment.
The conclusion is not that infrastructure spending is a mistake. The conclusion is that “AI exposure” no longer describes a uniform earnings profile. For investors who already own these companies through broad index funds, the next useful question is not who is spending the most. It is who can turn that spending into durable, per-share cash generation.
The earnings page contains three different stories
A headline earnings release combines operating performance, investment timing and accounting items that may have little to do with the cash economics of the quarter.
Alphabet is the clearest example. It reported net income of $112.2 billion, almost four times the prior-year level. But $98.0 billion of other income was primarily an unrealized gain on equity securities. The operating business was strong—operating income rose 30%—yet the reported net-income jump was not a repeatable measure of Google’s underlying earning power.
Amazon’s second-quarter net income included $53.4 billion of non-operating pre-tax other income, primarily from its investment in Anthropic. Again, that does not make the gain unreal. It makes it different from recurring profit generated by retail, advertising or AWS.
Meta supplied the opposite warning. Revenue grew 28%, while net income declined 14%. The quarter included $2.4 billion of legal charges and $1.18 billion of severance expense, but the larger cash-flow fact was capital intensity: operating cash flow of $31.86 billion was almost entirely consumed by $30.12 billion of property purchases and $962 million of finance-lease principal payments.
An investor reading only revenue growth would miss the capital burden. An investor reading only net income would confuse investment gains, litigation and restructuring with the recurring economics of the platforms.
Cash flow is the bridge, not the verdict
Free cash flow is often treated as a final score. In a buildout this large, it is better used as a bridge between today’s demand and tomorrow’s return.
Alphabet generated $39.07 billion of operating cash in the quarter and spent $44.92 billion on property and equipment, leaving free cash flow of negative $5.86 billion. Meta generated $31.86 billion and reported only $784 million of free cash flow after its property purchases and finance-lease payments. Microsoft generated $55.44 billion of operating cash and spent $35.80 billion on property and equipment, leaving roughly $19.64 billion before other investing and financing activity.
These figures are not perfectly interchangeable. Companies use different financing structures, asset lives and definitions. A finance lease may put infrastructure into service before the associated cash leaves the business. Quarterly working-capital movements can also distort a single period.
Still, the contrast is economically useful. Microsoft’s current operations covered its cash infrastructure additions with more room to spare. Alphabet’s quarterly property spending exceeded operating cash generation. Meta was close to the line. Amazon’s trailing-year measure had crossed below it.
That does not predict which stock will outperform. It identifies where the burden of proof now sits.
What must the new capacity earn?
Capital expenditure becomes valuable when the assets produce returns above their cost of capital for long enough to compensate for obsolescence, competition and execution risk.
AI infrastructure makes that test unusually demanding. Data-center shells and power connections can serve several generations of equipment. Accelerators, servers and networking hardware can age much faster. Microsoft has said a large share of its recent spending is for shorter-lived assets such as GPUs and CPUs, while another portion supports sites expected to operate for many years.
The revenue response is already visible. Google Cloud’s revenue rose 82% and its operating income more than tripled to $8.81 billion. Azure and other cloud services grew 43%. AWS sales grew 37% and AWS operating income increased to $16.6 billion.
But growth alone does not reveal the return on the next dollar of capacity. The practical sequence is:
Demand → utilization → revenue → operating cash → maintenance and expansion spending → free cash flow → per-share value.
A bottleneck anywhere in that sequence can weaken the investment case even while headline revenue remains strong. Capacity may arrive before customers are ready. Price competition may lower the return. Electricity, memory and construction costs may rise. Depreciation may catch up with capital spending and pressure future margins. Or demand may remain strong enough that the current cash squeeze proves temporary.
The portfolio issue is larger than four stock symbols
Many households do not think of themselves as making a concentrated AI bet. They own a broad U.S. index fund, a growth fund, a technology fund and perhaps one or two individual leaders. The company names can repeat across every account.
The SEC’s Investor.gov notes that an ETF or mutual fund does not necessarily provide diversification when it is narrowly focused. The same logic applies when several funds hold the same dominant companies. A portfolio can contain many line items and still depend on one earnings engine: large platforms converting AI infrastructure into profitable demand.
The useful audit is exposure by company and risk driver, not by account label.
- Add the overlap. Estimate each company’s weight across taxable accounts, retirement plans, index funds and individual positions. Look through fund names to the underlying holdings.
- Separate reported earnings from operating earnings. Identify investment gains, legal charges, severance and other items that can obscure the recurring business.
- Calculate the cash-coverage ratio. Divide operating cash flow by property and equipment spending over both the latest quarter and trailing year. A single quarter is a signal; the trend is the evidence.
- Track the return indicators. Cloud revenue, operating margin, backlog, utilization, depreciation and free cash flow should begin to move in a coherent direction.
- Set a concentration ceiling before volatility sets it for you. The right limit depends on time horizon, tax consequences and the role of the portfolio—not on confidence in a single earnings call.
The Perspective Log’s Cash-Flow Gate
For capital-intensive growth stories, we will return to a simple gate in future editions:
Can recurring operations fund the build, can the build earn an acceptable return, and does that return reach each share without requiring ever-larger financing?
Today, Microsoft clears the first part of that gate with the widest quarterly cushion among the three directly comparable releases. Alphabet and Meta show much thinner coverage after investment. Amazon’s trailing-year free-cash-flow outflow shows that even enormous operating cash generation can be overtaken by the scale of the build.
This is a snapshot, not a ranking. The next several quarters will determine whether today’s negative or narrow free cash flow represents productive capacity arriving ahead of revenue—or a permanently more capital-intensive business model.
What would change the view
- Sustained cloud and AI revenue growth accompanied by expanding operating cash flow and recovering free cash flow.
- Capital spending moderating without renewed capacity constraints or lost market share.
- Depreciation rising faster than gross profit, indicating that the accounting cost of the build is catching up with reported margins.
- Greater reliance on debt, equity issuance or finance leases to fund infrastructure that operations previously covered.
- A material reduction in overlapping Big Tech exposure across the reader’s own funds and individual holdings.
Jasper Kellan is editor of The Perspective Log. This article is general information, not personal investment, tax or legal advice. Publication cutoff: 4:30 a.m. ET, August 11, 2026.
Primary sources: Alphabet, second-quarter 2026 earnings release; Microsoft, fiscal fourth-quarter 2026 earnings release; Meta, second-quarter 2026 earnings release; Amazon, second-quarter 2026 results; Investor.gov, asset allocation and diversification.
