Technology's bad week fell on the spenders, not the builders
Grant Bishop's view reads the week as the cost of the artificial intelligence (AI) build landing on the companies that pay for it, which sharpens our standing view that AI spending is running ahead of the revenue it brings in, rather than overturning it.
Drafted with Claude Opus 5.5 · read and committed by the publisher How we work →
Bottom line
- Did this week's fall in technology shares say something about the AI build itself?
- Our answer is that the losses landed on the companies paying for the build, while the suppliers, Vertiv, GE Vernova, Quanta Services, KLA and Amphenol, closed higher, which sharpens our view that AI spending is running ahead of the revenue it brings in rather than overturning it.
- Grant Bishop's view: A guidance cut or a shrinking backlog at the suppliers, or a rate rise at the Federal Reserve's December meeting, would change that.
What would change our mind
A cut to guidance or a shrinking backlog in the next results from the companies that supply the build, breadth figures, once posted, showing the decline spread well beyond the companies spending on AI, or a rate rise at the Federal Reserve's December meeting, would change this conclusion.
The question
Technology shares had a hard week, and the losses landed on one side of the AI build. On 29 October the companies that pay for the build, Alphabet and Amazon, closed lower, and so did ServiceNow, a software company that also reported this week. The companies that supply the build, GE Vernova, Quanta Services, Vertiv, KLA and Amphenol, closed higher than the day before.
That split is the finding. Around the days the buyers of the build published their cash bills, the market marked them down and marked up the suppliers that build it. That is not a judgment that the build has stopped. It is a judgment about who carries its cost while revenue catches up.
What we cannot yet say is whether this was the worst week for technology shares since June. The index evidence in this piece stops at 29 October, and the St. Louis Fed's daily series does not establish the highest close since June . The superlative may prove right. It is not yet a fact, and we do not treat it as one.
What is known
The week, by date
28 October. The Federal Open Market Committee, the Federal Reserve's rate-setting body, held the target range for the federal funds rate at 3-3/4 to 4 percent by a 10–2 vote . Both members voting against preferred to raise the range by a quarter of a percentage point . The statement said economic activity "is expanding at a strong pace" and that "Inflation remains elevated" . The Committee last moved in September, as the frame records. The same day, Alphabet, ServiceNow, Vertiv, GE Vernova, KLA and Amphenol released quarterly results .
29 October. Alphabet's shares took the reaction to its results. The company reported revenues up 25% and Google Cloud revenues up 103% . In the same release it reported purchases of property and equipment of $51.4 billion, against $24.0 billion a year earlier . Its shares closed at 283.82 on 29 October, against 324.89 on 28 October. Amazon, Quanta Services and Intel released results that day . The S&P 500 field for that date was blank in the St. Louis Fed's file .
30 October. The closes in this piece run to 29 October. The 30 October closes and index levels had not been confirmed by two price providers, or posted on the Federal Reserve Bank of St. Louis's data service, when this was written, so no 30 October figure appears here. The week's final numbers are not yet in, and anyone quoting them as settled is quoting ahead of the data.
The market under the index
No primary source we hold carries a figure for how many S&P 500 members rose on 29 October. None carries a figure for money moving toward defensive sectors on 30 October. We do not supply either.
What we can show is narrower. Among the companies in the table below, the declines on 29 October came from the AI spenders, from ServiceNow and from Intel. The suppliers, Vertiv, GE Vernova, Quanta Services, KLA and Amphenol, closed higher. That points toward a decline concentrated in a handful of large AI-exposed names. But it is a list of our own choosing, not a measure of the index. A fall led by a few heavyweights and a fall spread across the market look the same in a headline index number. Only breadth figures tell them apart, and those are not yet on a primary source.
How unusual
We have no sourced frequency for declines of this size in the S&P 500 or the Nasdaq indexes. No primary source establishes the peak from which a drawdown, the fall from the highest close, would be measured. We therefore cannot say whether this week counts as a correction, the market's name for the larger declines, or how often weeks like it have come. We decline to estimate. From where the evidence stops on 29 October, a rare week and a routine one would look the same.
Names that bear on this
Share closes are for 28 October and 29 October, each confirmed by two price providers. None of these releases states the date of the company's next results.
| Company | Dated disclosure this week | What it reported | Share close before and after | Next scheduled disclosure |
|---|---|---|---|---|
| Alphabet | Third-quarter results, 28 Oct | Revenues up 25%; Google Cloud up 103%; free cash flow of negative $9.1 billion | 324.89 (28 Oct) → 283.82 (29 Oct) | not yet dated |
| Amazon | Third-quarter results, 29 Oct | Trailing-twelve-month free cash outflow of $42.4 billion, driven primarily by a $94.8 billion year-over-year increase in purchases of property and equipment that the release attributes primarily to AI investment; still expects about $220 billion of cash capital spending this year | 254.60 (28 Oct) → 243.79 (29 Oct) | not yet dated |
| ServiceNow | Third-quarter results, 28 Oct | Raised full-year subscription revenue and free-cash-flow margin outlook; goal for AI annual contract value, the yearly worth of signed subscriptions, raised to $1.7 billion | 150.13 (28 Oct) → 109.67 (29 Oct) | not yet dated |
| Vertiv | Third-quarter results, 28 Oct | Organic orders growth over 100%; raised full-year guidance | 242.94 (28 Oct) → 245.92 (29 Oct) | not yet dated |
| GE Vernova | Third-quarter results, 28 Oct | Backlog of $197 billion, up $20 billion from the prior quarter; raised guidance | 878.03 (28 Oct) → 887.85 (29 Oct) | not yet dated |
| Quanta Services | Third-quarter results, 29 Oct | Record backlog of $63.7 billion | 606.89 (28 Oct) → 635.18 (29 Oct) | not yet dated |
| KLA | Fiscal first-quarter results, 28 Oct | Revenue above the midpoint of its own guidance range | 158.77 (28 Oct) → 167.43 (29 Oct) | not yet dated |
| Amphenol | Third-quarter results, 28 Oct | Record third-quarter results; full-year sales guidance issued | 80.96 (28 Oct) → 82.08 (29 Oct) | not yet dated |
| Intel | Third-quarter results, 29 Oct | Revenue of $16.5 billion, up 21% on a year earlier; chief financial officer expects the personal computer slowdown to continue "at least through the first half of next year" | 95.53 (28 Oct) → 92.07 (29 Oct) | not yet dated |
The case for
The case for reading the week as our standing view showing up begins with where the losses fell and what was filed the day before. Alphabet reported revenues up 25% and Google Cloud up 103% , results most companies would want. Its shares closed lower the next day. The same release shows purchases of property and equipment of $51.4 billion in the quarter, against $24.0 billion a year earlier. It also shows free cash flow, the cash left after investment, of negative $9.1 billion . A quarter in which investment consumed more cash than the business produced is the bill for the build arriving in cash, not in a forecast.
Spending on the AI build is running ahead of the revenue it produces.
Amazon said the same in its own words. Its trailing-twelve-month free cash flow was an outflow of $42.4 billion, "driven primarily by a year-over-year increase of $94.8 billion in purchases of property and equipment". The release says that increase "primarily reflects investments in artificial intelligence" . Amazon held its expectation of about $220 billion of cash capital spending this year, in the same release that puts the year-over-year rise in its trailing-twelve-month purchases of property and equipment at $94.8 billion . Its shares closed at 243.79 on 29 October, against 254.60 on 28 October. That is our standing view that AI spending is running ahead of the revenue it brings in, set out by the spenders themselves.
ServiceNow is the cleanest test of our watchpoint that good results are no longer being paid for. It raised its full-year subscription revenue and free-cash-flow margin outlook. It lifted its goal for AI annual contract value this year to $1.7 billion . Its shares closed at 109.67 on 29 October, against 150.13 on 28 October. Its release offers no reason for the move, and we do not invent one. A raised outlook met by a fall is the pattern the watchpointWatchpoint: a claim we have put on the record with its date, what would prove it wrong, and the date we will check. Each one has its own page. must explain.
The other side of the ledger held. Vertiv reported organic orders growth, which excludes the effect of acquisitions, of over 100%, and it raised its full-year guidance . GE Vernova's backlog, the orders taken but not yet delivered, reached $197 billion, up $20 billion from the previous quarter . Quanta Services reported a record backlog of $63.7 billion . Vertiv, GE Vernova and Quanta Services all closed higher on 29 October than on 28 October, as did KLA and Amphenol. One company's capital spending is another's revenue, and this week's prices drew that line: down for those writing the checks, up for those cashing them.
The case against
The strongest reading against ours is that the week cannot carry the weight of a verdict at all. The index evidence stops at 29 October, and the daily series does not establish the highest close since June . Without a completed week, a peak or a sourced frequency of past declines, the week's ranking and its drawdown are unmeasured. Calling it a signal about the build is as unsupported as calling it routine. What can be seen is a sharp fall in a few large names. That is also what many investors holding the same shares and selling at once would produce, whatever the filings say.
The company evidence is less tidy than our case makes it. Alphabet's net income fell, but its other income swung to a net expense of $8,221 million from $12,759 million of income a year earlier . That is a non-operating item, not the build. A share price marking down lower reported earnings is not the market ignoring good results. Amazon's operating income rose on the year and Amazon Web Services grew : revenue from the build is arriving. Neither Alphabet's capital spending nor Amazon's outflow proves the returns will be poor. They show the burden is heavy, which is a different claim.
ServiceNow sells software rather than building or buying data centers, and its own release carries no explanation of the move . Treating it as proof of a verdict on AI reads more into one session than the filing supports. Vertiv's orders, GE Vernova's backlog and ServiceNow's raised outlook all argue against any collapse in demand . Even so, supplier orders do not by themselves show that the customers will earn a return.
Other explanations sit in plain view. The Federal Reserve held, but called inflation elevated and activity strong, and both dissenters preferred a rise . Shares whose value rests on earnings far in the future are the first to be repriced when the cost of money is in question, build or no build. Intel reported through its chief financial officer that it sees the personal computer slowdown continuing "at least through the first half of next year" . That is weakness in personal computers, unrelated to data centers. The fair conclusion is narrower than ours. Resilient demand sits beside heavy cash demands, uneven end markets and unresolved inflation, and a few days of prices in a handful of names cannot choose among those explanations.
What we don't know, and when
What it does to the standing view
Our standing view that AI spending is running ahead of the revenue it brings in strengthened. Alphabet's negative quarterly free cash flow and Amazon's trailing outflow are stated in the companies' own releases, and Amazon attributes its outflow to AI investment .
Our watchpoint that good results are no longer being paid for is under test, and less cleanly than ServiceNow alone suggests. Part of Alphabet's lower net income came from a swing in other income , so its fall is not a pure case of good results ignored.
Our watchpoint that spare rented computing does not cancel new power and cooling construction was tested by the orders and backlogs at Vertiv, GE Vernova and Quanta Services. Amazon held its spending plan and Vertiv raised its guidance ; guidance from the other hyperscalersHyperscalers: the largest cloud companies (Microsoft, Alphabet, Amazon, Meta); their capital spending is the demand under the AI build-out., the largest cloud-computing companies, is not covered here.
Our watchpoint on whether the Federal Reserve raises rates again is untouched. There was no rise in October. With both dissenters preferring one , December now carries the test.
None weakened.
The dates that will tell
The Federal Open Market Committee's next test is its meeting on 8–9 December , as our watchpoint on whether the Federal Reserve raises rates again records. None of the results releases filed this week dates the company's next one; each company's own announcement will set that date. Breadth figures, once a source carries them, will settle how concentrated the decline was. Each watchpoint's page carries its own review date.
October in review
On the watchpoint on the Federal Reserve, the Committee held in October rather than adding to its September increase, so the test moves to December and nothing tripped. On spending running ahead of revenue, and on good results going unpaid, we read October's filings as consistent with both, and we do not read anything filed this month as tripping either. On the watchpoint on new power and cooling construction, nothing in the filings covered here tripped it; the other largest cloud-computing companies' guidance is not yet read. Amazon held its spending plan and Vertiv raised its guidance .
What would change our mind
Grant Bishop's view is that this week marked down the companies paying for the AI build and marked up the suppliers building it, and that this is not a verdict against the build. The week sharpens our standing view that AI spending is running ahead of the revenue it brings in, and the frame as a whole stands.
What would change that reading: The first is a cut to guidance or a shrinking backlog in the next results from Vertiv, GE Vernova, Quanta Services, KLA or Amphenol, none of which is yet dated; that would turn a story about who pays into a story about whether the build continues. The second is breadth figures, once posted, showing the decline spread well beyond the AI spenders; that would favor that reading over ours. The third is a rate rise at the Federal Reserve's December meeting, which would make the cost of money, not the build, the better explanation for the week.
How this piece was made
Sources. 12 numbered sources, in the Record.
Related claims
Tell the publisher
Something we should look at? Suggest a company, a topic or a feature.
Record
Written according to the Grant Bishop standard — machine-researched, machine-checked, human-judged; then read and committed by the publisher. Every number traces to a filing or a dated market record; every claim carries a date and what would prove it wrong.
- Drafted by
- Claude Opus 5.5 under Grant Bishop's methodology
- Verification
- multi-source verified, machine-drafted, human-gated
- As of
- 2026-10-31
- Published
- Oct 31, 2026, 13:49 ET · 2026-10-31 17:49 UTC · changelog
- Bears on
- What would change this
- A cut to guidance or a shrinking backlog in the next results from the companies that supply the build, breadth figures, once posted, showing the decline spread well beyond the companies spending on AI, or a rate rise at the Federal Reserve's December meeting, would change this conclusion.
- Position disclosure
- The publisher holds index-level U.S. equity exposure and no direct position in any single name discussed.
- Not advice
- General commentary for a general audience. Not advice for your situation.