Did Google's capital expenditure in its financial report come in below expectations?

Did Google's capital expenditure in its financial report come in below expectations?

Below expectations, but only 6.9% lower, and this item itself props up free cash flow rather than being the one that smashes cash flow into negative territory. In the September quarter Alphabet's capital expenditure was $51.36 billion, versus the sell-side expectation of $55.2 billion on the FactSet basis, a gap of $3.85 billion, i.e. -6.97%.

What really makes the market uncomfortable is not the money already spent, but that the full-year guidance was not raised once more. The Q2 earnings call had just raised 2026 from $190 billion to $195–205 billion, and the market assumed by default that Q3 would raise it another notch, but it stayed unchanged. So the phrase "capital expenditure below expectations" is actually talking about the slope of the guidance, not the absolute amount of spending.

The total is rising, the acceleration is falling

The quarter-on-quarter growth rate fell from +28.1% in Q1 to +25.9% in Q2, and only +14.3% remained in Q3; year-on-year it is still +114.4%. Capital expenditure as a proportion of revenue was 40.3% in Q3, versus 23.4% in the same period last year. Money is still being poured into compute, it is just that the part by which each quarter pours in more than the previous quarter is getting smaller.

This is the real source of the three words "below expectations"—the absolute amount hit a new high, and the increment is narrowing.

Guidance unchanged, Q4 must make up $63.05–73.05 billion in one go

The cumulative total for the first three quarters is $131.95 billion. Subtracting that from $195–205 billion, the December quarter must deliver $63.05 billion to $73.05 billion, another 22.8% to 42.2% higher than Q3's $51.36 billion. It did not change guidance, which amounts to pressing all the pressure onto those three months of December.

The statement about 2027 was also unchanged. CFO Anat Ashkenazi said on the call that day:

We continue to expect our capex to increase significantly in 2027, and will provide further details on our Q4 call.

Free cash flow -$9.07 billion, this account is not caused by capital expenditure

Operating cash flow was $42.29 billion, $6.12 billion less year-on-year. Backing out from the sell-side expected capital expenditure of $55.2 billion and expected free cash flow of -$2.27 billion, expected operating cash flow is $52.93 billion—actual was $10.64 billion less than that.

$10.64 billion less inflow, $3.85 billion less spent, netting the two ends, free cash flow is $6.8 billion worse than expected, exactly matching the difference between -$9.07 billion and -$2.27 billion.

Calculated on the LSEG basis (expected capital expenditure $55.09 billion, expected free cash flow -$2.7 billion), the conclusion is the same: the operating cash flow gap is about $10.1 billion, and free cash flow is $6.37 billion short. Both bases point to the same thing: the cash flow gap is not on the equipment side.

The missing $10.6 billion is in working capital

Unit: $ billion. The largest piece is "other assets" tying up $7.04 billion more and "accrued expenses and other liabilities" flipping from +$8.08 billion in the same period last year to -$1.17 billion, $9.25 billion going back and forth.

"Other assets" is not broken down in the earnings release. The Q3 10-Q is not out yet; the previous one was the Q2 report on July 23, and only the original text will tell whether it contains prepaid income tax, prepayments to suppliers, or something else. Until that is in hand, attributing it to "prepaying compute fees to chipmakers" is just a guess.

The capital expenditure line understates real compute investment

Two points, both with hard numbers.

• Inventory rose from $2.44 billion at the start of the year to $13.31 billion, tying up $10.96 billion of cash in the first three quarters. The earnings release explicitly states that external sales of TPU systems are an important source of revenue growth, and this stockpiling does not go into "purchases of property and equipment".
• As of the end of June, fixed assets "not yet placed in service" were $122.81 billion, versus $78.59 billion at the end of last year, an increase of $44.22 billion in half a year. In addition, there are signed but not yet commenced data center leases with future payments of $85.2 billion, which are also not in capital expenditure.

The financing side is moving in tandem: long-term debt rose from $46.55 billion at the start of the year to $125.1 billion, with net bond issuance of $85.22 billion in the first three quarters; in June there was also a $49.6 billion equity offering, of which $10 billion was a private placement from a Berkshire-affiliated institution. Capital expenditure can no longer be self-funded from operating cash flow.

Where is the output from the money already invested

The cloud piece has some. Q3 Google Cloud revenue was $30.71 billion, operating profit $11.12 billion, versus only $3.59 billion in the same period last year, with an operating margin of 36.2%. Revenue doubled while the margin moved up, showing that the servers poured in over the previous quarters are already collecting money.

Nothing on the advertising side. Search and other revenue was $63.35 billion, up only 12%, below the sell-side expectation of $64.96 billion; YouTube advertising was $11.21 billion, with growth falling to 9%. The increment is in cloud, the existing base is in advertising, and advertising hasn't kept up—which is also why "spending $3.8 billion less on capital expenditure" wasn't read as good news.

Depreciation hasn't started to bite yet. Third-quarter depreciation was $8.37 billion, +49% year-on-year, while capital expenditure was +114% year-on-year; depreciation growth is less than half of spending growth, and on top of that there is still $122.8 billion of assets not yet transferred to fixed assets. That batch of money turning into depreciation and entering operating costs is a matter for the coming quarters.

Peer positioning: largest absolute amount, most conservative guidance

A note on the basis: Alphabet and Microsoft both use the cash-flow basis of "purchases of property and equipment," excluding finance lease principal; Meta's $34.91 billion includes finance lease principal, so it is slightly inflated and cannot be directly compared across companies. Amazon's most recent report is its second-quarter report dated July 30; the third quarter has not yet been released.

In absolute amount, Alphabet is the largest. On guidance, Meta raised its lower bound from $130 billion to $135 billion, while Alphabet did not move at all—this is the most concrete comparative for "below expectations."

The strongest rebuttal, and the conditions under which it holds

The strongest rebuttal is: the $3.85 billion difference is delivery timing, not demand. The money is already in contracts ($85.2 billion of leases not yet commenced, $122.8 billion of assets not yet put into use), and since guidance was not changed, it must be made up in the fourth quarter. Under this explanation, the capex miss has no negative implication for 2027 capacity; what should really be worried about is the reverse—a single-quarter jump from $51.3 billion to $63–73 billion, quarter-on-quarter 23% to 42%, no hyperscaler has ever done such a leap.

This explanation holds under only one precondition, and it must wait for next month's original text: in the third quarter, fixed assets "not yet put into use" must continue to increase quarter-on-quarter. If it stops, that means it is not that the equipment hasn't arrived, but that capacity cannot be brought up, and then this $3.8 billion miss is not a good thing.

Falsification conditions

• Third-quarter 10-Q (expected late October to early November): if "not yet put into use" assets no longer increase quarter-on-quarter, the "spending delayed" explanation is void.
• Fourth-quarter actual capital expenditure below $63 billion: the full-year lower bound of $195 billion is breached, and "maintaining guidance" becomes empty talk.
• The fourth-quarter earnings call no longer reiterates a "significant increase" in 2027: the judgment that this round of capital expenditure has not yet peaked must be withdrawn.

Sources: Alphabet Inc., "Alphabet Announces Third Quarter 2026 Results," Exhibit 99.1 to Form 8-K, 2026-10-28, filing number 0001652044-26-000080; Alphabet Inc., "Form 10-Q" (second quarter of 2026), 2026-07-23, filing number 0001652044-26-000071; Microsoft, "FY2027 First Quarter Earnings Release," Exhibit 99.1 to Form 8-K, 2026-10-28, filing number 0001193125-26-436122; Meta, "Third Quarter 2026 Earnings Release," Exhibit 99.1 to Form 8-K, 2026-10-21, filing number 0001628280-26-066700; sell-side consensus and earnings call quotes: CNBC, "Alphabet sinks 10% as it holds 2026 capex unchanged, expects significant increase in 2027," 2026-10-28; TradingKey, "Alphabet Shares Plunge 8% After Hours…," 2026-10-28 (FactSet basis).