$GOOGL just doubled Google Cloud revenue to $30.7B, and the stock still fell about 7% after hours. Total revenue of $127.58B came in a touch under estimates, GAAP EPS of $2.12 took an $8.7B hit from equity losses, and $51.4B of quarterly capex pushed free cash flow negative. Tonight the market is looking at the spend, not the growth.

$MSFT just put up a clean beat, with revenue of $92.28B vs $90.63B expected and Azure up 48% ex-FX. The surprise is capex at $39.4B, well under the roughly $46B the Street had penciled in. More cloud growth on less spend is exactly what the AI capex worriers wanted to see, and shares are up about 5% after hours.

$GOOGL is going into tonight's print well off its highs. Honestly the capex number is the only line I really care about.

Ten feels like a lot for forever. Your core, growth and momentum picks are mostly the same megacaps in different wrappers, so it's really more like eight.

The Fed held at 3.75% to 4% as expected, but the 10-2 vote is the story. Cook and Hammack both wanted another hike today, and the market now prices about 85% odds of one in December. $SPY $TLT

Datavault AI says it hasn't been able to withdraw 818 $BTC from an exchange since June, and the stock has reportedly been cut in half. A balance on someone else's screen was never the same thing as holding the coins.

Wendy's finally trading like a 4 for $4.

Listing is one thing. Coinbase earns a fat cut on USDC, so I'd be surprised if RLUSD gets anything like the same push in the app.

Yeah, the customer base is the cheat code. Every company already pays for Office, so the upsell is basically one email. Renewals after year one will tell us if it's real.

Two weeks ago the AI trade was a bubble. One SK Hynix report later and the same accounts are posting $MU supercycle charts again. Most people's conviction just follows the last candle.

Negative free cash flow at Meta is a sentence nobody would have believed two years ago. The ad machine is basically funding a data center habit now.

Copilot is the bit I'm least sure about. Azure demand kind of sells itself, but getting companies to pay per seat for an assistant is a much harder sell.

Correction: $0.51 is GAAP EPS vs ~$0.55 est. Robinhood's adjusted EPS was $0.60. Revenue $1.42B, in line. $HOOD

The "just buy the S&P, you don't need anything else" crowd had a decent year. $SPY up 14%. Taiwan is up 69%. Korea is up 62%. Even Japan beat it. Home bias feels safe right up until you check what you missed.

$HOOD missed on both lines tonight, with adjusted EPS of $0.51 vs $0.55 expected and revenue of $1.42B vs $1.43B. Revenue still grew 12%, but shares are down about 5% after hours. When the bar keeps rising, a near miss reads like a real one.

$V beat on revenue by about $10M on a $12B quarter, which is basically in line. EPS of $3.48 vs $3.44 and 74.2B processed transactions look solid, but the stock is slipping after hours anyway. At this size, matching the bar isn't the same as clearing it.

At this point the earnings call is basically a capex confession with some Azure numbers attached.

Forced selling always looks obvious in hindsight. In the moment it just feels like the floor keeps moving.

Hope dies last. Usually right around the guidance call.

SoFi holders have seen this movie before. Strong print, green premarket, then the stock does whatever it wants by the close.