#谷歌云增长103%股价却跌近10% Cloud business doubled, but profits plunged! Alphabet shares fell 8% after hours as the cost of AI “burning cash” comes due?

After the market close on October 28, U.S. Eastern Time, Google parent Alphabet reported its third-quarter 2026 results.
Revenue was $127.58 billion, up 25% year over year, slightly below the $127.93 billion estimate. Google Cloud revenue soared 103% year over year to $30.71 billion, while operating income surged to $11.1 billion from $3.6 billion in the same period last year.
But that’s not what the market focused on.
Earnings per share came in at just $2.12, far below expectations of $3.02 and down from $2.87 in the same period last year. Advertising revenue of $82.01 billion and Services revenue of $96.3 billion both missed expectations. Operating income was $40.58 billion, below the $43.16 billion estimate.
After the results were released, Alphabet shares plunged nearly 8% at one point in after-hours trading.
Cloud business doubled and revenue came close to expectations, yet shares fell nearly 8% after hours—this report puts the central contradiction of the AI era on full display: growth is real, but so is the cost.

1. First, clarify a key point: Half of the “plunge” in profits comes from investment losses reversing prior gains!
The market’s fears about profits need to be unpacked.
Alphabet’s Q3 earnings per share were $2.12, far below the $3.02 expected. But the biggest single drag on EPS was not its core business—it was equity investments.
This quarter, Alphabet recorded a net loss of $8.7 billion on equity securities investments, reducing net income by $6.7 billion after tax and cutting earnings per share by about $0.55. In the same period last year, this item was a huge unrealized gain.
In other words, the high base from the same period last year was “propped up” by one-time investment gains; this quarter’s low base was “dragged down” by one-time investment losses. The direct reversal from a “large profit” to a “large loss” is one of the main reasons EPS appears to have “plunged”—not a sudden deterioration in the core business.
What really deserves attention is that “operating income was $40.58 billion, below the $43.16 billion expected.” But the other side of the story matters too: Group operating income grew 30% year over year, and the operating margin expanded from 31% to 31.8%. Operating income “missing expectations” means it fell short of analysts’ very aggressive forecasts; its absolute growth and improved profitability remain strong.
The real question isn’t “how much did profit fall?” but “how fast are costs rising?” Depreciation, energy use, and operating costs from AI infrastructure investment are materially eroding the income statement. The cloud business boom and margin pressure are two sides of the same coin.

2. Cloud business doubles: AI demand is real, but “supply constraints” are more challenging than “lack of demand”!

Google Cloud’s 103% growth is the hardest-hitting figure in this report.
Revenue was $30.71 billion, far above the market estimate of $29.1 billion; operating income was $11.1 billion, and the operating margin jumped to above 36% from about 24% in the same period last year.
This shows one thing: AI compute isn’t “failing to sell”—there isn’t enough to sell.
CEO Pichai said explicitly that AI investment is driving growth in Search queries, and that Cloud growth is driven by huge demand for AI infrastructure and AI solutions, with strong demand for TPU systems across a broad customer base. Earlier, Pichai even acknowledged that Cloud had given up some short-term revenue to prioritize compute capacity for frontier AI research.
Cloud backlog reached $514 billion, with more than 50% expected to be recognized as revenue over the next 24 months. This is the clearest evidence on the demand side. The question isn’t “are there customers?” but “can capacity keep up?”
Google’s response is to “walk on two legs”: build its own compute capacity while increasing third-party compute purchases as a transitional solution to support customer expansion. But third-party purchases will weigh on Cloud margins in the short term—an active trade-off between “growth” and “profitability,” not a sign of a demand problem.

3. Capital expenditure of $51.4 billion: below expectations, but the free cash flow shortfall is about 250 times what was expected!

The market expected capital expenditure of $55.78 billion; the actual figure was $51.36 billion, about 8% below expectations.
Coming in below expectations is good in itself, but the absolute figure is still staggering. Capital expenditure was just $23.95 billion in the same period last year; this quarter’s $51.4 billion was more than double year over year.
The real warning sign is free cash flow: this quarter, free cash flow was negative $9.065 billion, while analysts had expected just negative $36 million—a shortfall about 250 times larger than expected. In the same period last year, free cash flow was a net inflow of $24.461 billion.
A roughly 250-fold gap versus expectations is the most glaring figure in this report, and explains the 8% after-hours drop better than “profits missing expectations.” Google is using cash generated by its advertising business to build AI infrastructure at an unprecedented pace. Its full-year 2026 capital expenditure guidance remains at $195 billion to $205 billion, and management explicitly said that it will continue to “grow significantly” in 2027.
Capital expenditure came in below expectations, yet the stock didn’t rise. This shows that the market’s concerns have moved beyond “how much did it spend this quarter?” to a more fundamental question: when will the free cash flow shortfall narrow?

4. The “crack” in advertising: Search ad growth is slowing, and monetization from AI upgrades hasn’t caught up!

Advertising revenue was $82.01 billion, up 11% year over year, but below the expected $83.57 billion.
The breakdown makes it clearer: Search and other advertising revenue was $63.3 billion, up 17% year over year, but below the expected $64.96 billion; YouTube advertising revenue was $11.1 billion, also slightly below expectations.
This is the easiest signal to overlook in this report, but may be the most important. Advertising is Google’s core cash cow. AI is reshaping Search, but the incremental monetization from that transformation has yet to fully offset the friction costs along the way.
Pichai emphasized that “AI investment is driving growth in Search queries, including commercial queries.” Query growth is good, but if queries grow faster than advertising monetization efficiency improves, short-term ad revenue growth will be diluted. The market’s concern is precisely that AI Search may erode the traditional advertising monetization model before its benefits arrive.
Slower-than-expected ad growth, combined with a widening free cash flow shortfall, points to the same conclusion: AI’s “boost” to core business has yet to be fully realized, but its “erosion” of the cost base has already begun.

5. Industry position and moat: TPU is Google’s most undervalued card
Google’s AI competitiveness isn’t about whether it has the “strongest model”; it’s that it is one of the very few companies in the world with full-stack capabilities spanning “models + chips + cloud.”
TPU is at the heart of this full-stack advantage. Google has more than a decade of experience developing its own TPUs, and its eighth-generation chip is about to launch. External demand already exceeds supply. More importantly, it has a cost advantage: the total cost of ownership (TCO) of a TPUv7 server is about 44% lower than NVIDIA’s GB200, and even through cloud rentals, TCO is still about 30% lower. This advantage comes not only from the chip itself, but also from Google’s long-term investment in systems-level capabilities, financial engineering, and large-scale deployment.
TPU’s commercial potential is moving from “in-house use” to “external sales.” Google Cloud’s CEO has said that the TPU business is “more than twice the size of the second-largest hyperscaler,” and that it is monetized through three business models. There are reports that Google and Meta are discussing direct sales of TPU chips, with leasing as a bridge in 2026.
If TPU shifts from an “in-house cost-cutting tool” to an “AI chip sold to external customers,” Google’s valuation story will fundamentally change—it will no longer be just an advertising and cloud company, but a platform company with an AI chip ecosystem.

6. What should we watch next? Three key indicators

Indicator 1: Can Cloud growth stay above 80%? The 103% growth rate benefits from a low base. If Cloud growth remains above 80% in Q4, it would show that AI demand is holding up; if it quickly falls below 60%, concerns that “AI demand has peaked” will intensify rapidly.
Indicator 2: Can the free cash flow shortfall narrow? This quarter’s negative $9.065 billion was about 250 times what was expected. If it narrows in Q4 as revenue recognition accelerates and the pace of capital expenditure slows, concerns about an “endless money pit” will ease; if the shortfall continues to widen, guidance that capital expenditure will “continue to grow significantly” in 2027 will weigh heavily on the valuation.
Indicator 3: Can external TPU sales move from “talks” to “orders”? TPU’s TCO advantage (44% lower than GB200) is a real strength, but “saving money” and “making money from sales” are two different things. If orders for external TPU sales or rentals materialize in late 2026 or early 2027, Google’s valuation story will be rewritten.

Conclusion
This report from Google is a story about “the reality of growth and the lag in profits.”
Cloud business doubled, operating income grew 30% year over year, and TPU demand is strong—the AI industry trend remains intact.
But earnings per share were far below expectations, the free cash flow shortfall was about 250 times what was expected, and advertising growth slowed—the financial cost of AI is becoming concrete.
The nearly 8% after-hours drop doesn’t mean “Google is finished.” The reversal from large investment gains to large investment losses ($8.7 billion in investment losses, dragging EPS down by about $0.55) was the direct reason EPS appeared to plunge; but what the market is really pricing in is the “spending far faster than expected” exposed by the free cash flow shortfall (negative $9.065 billion vs. an expected negative $36 million, about 250 times larger), as well as the suggestion from slower ad growth that “AI monetization hasn’t caught up with AI investment.”
TPU is Google’s most undervalued card. Its 44% TCO advantage over GB200 gives it the confidence to move from “cutting costs in-house” to “generating revenue externally.” But until that card is truly played, the market will vote quarter by quarter through the income statement and cash flow statement.
The real test isn’t the $30.7 billion in Cloud revenue, but whether the free cash flow shortfall can narrow, advertising growth can stabilize, and external TPU sales can move from “talks” to “orders.”#Gate广场万圣节 $GOOGL