[Feature: Market Wrap] Alphabet Beat Its Numbers, and Fell Overnight — Because It Said It Would Spend Even More on AI

Yesterday, in this space, I wrote that Alphabet would open the first report card tonight. That report card came back last night. And the result flipped.

On Wednesday, July 22, the New York market barely moved. All four major indices finished flat, and the chips that had jumped hard the day before were quiet too. "Flat" means neither up nor down — essentially unchanged. The real event came after the close.

Alphabet beat market forecasts on both revenue and cloud. Yet when it actually expanded its 2026 AI investment plans, the stock fell more than 3.6% in after-hours trading — the session that opens once the regular market closes. Tesla sold more cars in a single quarter than ever before, and yet fell as its profit collapsed.

Here is the summary. Last night both companies beat the market's expectations — on revenue, on cloud, on deliveries. And yet both stocks fell. It was a day when the lens the market looks through shifted from growth (revenue) to profitability and cash. But the direct reason each fell differs — for Alphabet, the worry was about timing, about when that spending pays back as profit, as it expanded its AI capital spending; for Tesla, the main reason was that its automotive margins collapsed even as it sold plenty. And what got pushed down last night was not the universal proposition of "AI investment" — because under the same investment plans, the side that receives that money (the chipmakers) actually rose. This is one line threading the data together, not a nailed-down cause — what was observed, and how it is read, are set out separately below.

🇺🇸 United States — Indices Flat; the Real Event Was After-the-Close Earnings

Start with the indices. The day itself was quiet.

Indices (July 22 close)

• Chips were quiet too — the Philadelphia Semiconductor Index (SOX) +0.44%, Micron -1.17%, Intel -2.68%. Nvidia rose +2.29% and Broadcom +2.65%
• ⚠️ Don't mix up the dates — the day chips jumped hard was the day before, July 21 (Micron up around +12%). Micron slipping about 1% yesterday was a small giveback of that surge. Looking at yesterday alone, there was nothing remarkable in the US market

In the regular session, Alphabet closed -1.46% and Tesla -1.30%, both down modestly before reporting earnings. And after the close, the real news of the day arrived.

★ Alphabet — The Report Card Was Good, but the Stock Fell

On the numbers alone, these are good results. For reference, consensus below means the average of Wall Street forecasts — the bar the market had set in advance.

Cloud is the key here. Revenue grew 82% in a year, and the margin the business keeps rose to 35.6%. It is a signal that selling AI actually makes money.

So why did it fall — Alphabet actually expanded its 2026 capital-spending plan that day. Here, capex means the money poured up front into facilities like data centers and chips, and guidance is a company's own forward forecast. Alphabet raised that capex guidance again, from $180–190B in April to $195–205B. The current quarterly pace is about $45B, accelerating further into the second half. The finance chief added that it would "rise substantially in 2027" as well. The result → even though it was a mixed report card — revenue and cloud beat, but adjusted EPS came in a touch below consensus — the stock was down about -3.6% after hours. Raise guidance and the near-term costs grow, while the moment that investment returns as profit gets pushed further out. So when the market shifts its weight from "how much you invest" to "when you pay that money back," the stock often gets pressed down like this. (This far is observation; the "why" is commentary from several outlets — we don't assert cause.)

There is one number to watch. Alphabet's accounting net income comes to $112.1B and earnings per share to $9.11 — a 294% surge from a year earlier. But this figure carries a large unrealized gain on equity holdings — not money earned from the actual business, but the rising value of stocks it holds. So the right read on the core business is the adjusted EPS of $2.85 above.

★ Tesla — Sold the Most Ever, and Still Its Profit Collapsed

Tesla's direction was the opposite. It sold plenty, but kept less.

The two lines that stand out most are below. Operating margin is the share of profit left after subtracting cost of goods and expenses from revenue, and it shrank from 4.1% a year earlier to 1.4% — about a third. Free cash flow (FCF) is the cash actually left in hand after subtracting investment from the cash earned. This came to -$1.09B — it turned negative, for the first time since early 2024. That is because capital spending grew to more than double a year earlier (+142%).

Tesla's fall had a different character from Alphabet's. The main force dragging it down was not a plan to "spend more" but the profitability of the car business itself — as seen above, margins sank to a third and adjusted EPS came in well below the bar. The increased capital spending cannot explain this margin collapse. Capex is investment for the future, not this quarter's cost — what it explains is not margin but cash flow (FCF turning negative). The one point where the two companies overlap is this: they beat on revenue and deliveries, and the market still looked first at profitability and cash. After the results, the stock fell after hours. (The size of the drop swung with the timing of the release, so here we note only the direction.)

The one line for last night — Alphabet and Tesla fell for different reasons (one a plan to "spend more," the other a collapse in automotive margins). But the rhyme is alike — both beat on revenue, cloud and deliveries, and both stocks fell anyway. It was a day when the lens the market looks through shifted from "how much you sell (growth)" to "how much you keep and when you pay it back (profitability, cash)." Yet what got punished was not the universal proposition of "AI investment" — because under the same capex raise, the side that receives the spending, Nvidia (+2.29%) and Broadcom (+2.65%), actually rose, and this morning the KOSPI 200 overnight futures pointed to the +4% range too. What got pushed down last night was the near-term profitability of the side doing the spending (the spenders).

Rates, FX, commodities

• The US 10-year Treasury yield 4.659% (around +3bp); the dollar index at 101.14, little changed
• Oil (WTI) rose about 3% to near $86 a barrel. Gold also climbed to around $4,100 an ounce. Tension around Iran is the backdrop

🇰🇷 Korea — Up More Than 6% Intraday, Then Gave Most of It Back

The heart of Korea's session yesterday (July 22) was not the close but the reversal within the day.

Indices (July 22 close)

• KOSPI 6,797.70 (+0.74%) — but intraday it surged as high as 7,166 (+6.20%), then gave back most of the gain into the close
• KOSDAQ 751.09 (-0.30%) — it split direction from the KOSPI
• Buy sidecar — tripped around 9:06 a.m., just after the open. A sidecar is a brake that halts program-trading orders for five minutes when futures swing hard. Despite the name, a "buy sidecar" does not assist buying either — it is a device that slows the pace of a surge

It jumped hard right at the open, even tripping the sidecar, but gave most of that gain back over the day — that is the point of Korea's story yesterday. Outlets read it as profit-taking, as caution over Alphabet's earnings due that night, higher oil, and uncertainty around tariffs all converged.

• USD/KRW 1,480.10 (+6.70 won) — the won weakened slightly

📅 Today's Watch (Thursday, July 23) — Tonight It's Intel's Turn

• 🇰🇷 9 a.m. open (watch #1) — Seoul opens today into last night's US earnings. The overnight futures (KOSPI 200) traded through the night pointed to the +4% range. But given yesterday's experience of giving back an intraday surge, whether early strength holds to the close is worth watching. Korea's Q2 earnings season for automakers and banks also opens today. Outlets flagged whether big tech keeps raising AI investment as the variable for domestic chip sentiment
• 🇺🇸 Intel Q2 earnings (watch #2) — out after the US close today. Wall Street's bar is revenue of $14.4B and adjusted EPS of $0.22. If Alphabet and Tesla graded the "floor of AI investment" yesterday, Intel is the report card on the logic-chip side
• The grading stand for memory itself hasn't come yet — yesterday's results only looked at "how much is being spent on AI," not memory prices themselves. Those numbers come next Wednesday (July 29) from SK Hynix's Q2 earnings

🗓️ Key Calendar

In One Line

Last night, Alphabet's and Tesla's report cards came back. Both beat the bar on revenue, cloud and deliveries, yet both stocks fell after hours. The reasons they fell differed — for Alphabet, the main force was a plan to spend even more on AI; for Tesla, a collapse in automotive margins. What they share is that the lens the market looks through shifted from "how much you sell" to "how much you keep and when you pay it back." But under the same capex raise, Nvidia and Broadcom — the ones receiving that money — rose, so what got pushed down last night was not "AI investment" as a whole but the near-term profitability of the side doing the spending. This far is observation, and the "why" is media commentary, so we don't nail it down as cause. The logic-chip report card comes tonight from Intel; the report card on memory itself comes next week from SK Hynix.