The Morning Bell No. 3: Ninety Six Billion in a Single Quarter

Branden Bell
2 days ago
6 min read

The Morning Bell broadsheet cover, Vol. I No. 3: six painted blooms in a row with the hero bloom in amber, for Nvidia printing $96.2B and guiding to $108B while the S&P 500 managed +0.30%.

Nvidia did $96.2 billion in a single quarter, up 106% year over year, and guided the current quarter to $108 billion with zero data center revenue from China in the number. The stock is up 6.30% premarket. The S&P 500 is up 0.30%. That gap is the whole story of the morning.

I am not a financial advisor and nothing in The Morning Bell is financial advice. This is entertainment and education based on research my AI system produces and things I may or may not be doing personally. The account featured here is a small agentic test account, fully separate from my personal trading. Always do your own research.

The tape before the bell

Every number below is marked between 8:39 and 8:51 AM ET, premarket, from live broker quotes against Wednesday's official closes.

Premarket tape strip for August 27 2026: SPY +0.30%, QQQ +0.89%, IWM -0.08%, NVDA +6.30% at $222.88 breaking the top of the band, regime call amber.

S&P 500 (SPY): 768.35, +0.30%. A blowout print from the largest company in the index bought thirty basis points.

Nasdaq 100 (QQQ): 717.68, +0.89%. Where the actual reaction lives.

Small caps (IWM): 298.69, -0.08%. Negative. On this news.

Bonds and gold: TLT 83.14, -0.19%. GLD 420.80, -0.12%. Both quiet.

Bitcoin: $79,367, +0.69% against the midnight mark. Ether $2,501, +0.33%. Crypto did not care.

Volatility proxy: VIXY 17.76, -0.39%. Lower, which is what you would expect the morning after a binary resolves.

My system's regime call is amber, and I want to defend it rather than just print it, because on the surface this looks like a green morning. Semis are ripping. SMH is +2.57%, Micron +3.90%, Marvell +3.34%, Broadcom +1.27%. But look at what is not participating. Apple is -1.08%. Microsoft is -0.88%. Small caps are flat to red. HP is -11.92% after its own print. This is not the tape lifting. This is one sector and a handful of software names lifting while everything else sits out.

Amber means narrow. A market that can only rally through six tickers is a market with one point of failure, and the new Fed chair's debut keynote lands tomorrow at 10:00 ET.

What the system flagged overnight

The guide was the story, not the beat. Data center revenue came in at $89.0 billion against roughly $86.3 billion expected, which is a beat but not a shock. The $108 billion current quarter guide against about $104 billion expected is the number that moved the stock, and management specifically noted it assumes no China data center sales. My ledger scores a raise that excludes a whole country as higher quality than a raise that needs one.

The software tape is louder than the chip tape. Okta is +18.13%. Salesforce is +11.13%. CrowdStrike is +9.36%. Veeva is +8.06%. Snowflake is +3.69%. Five enterprise software names all gapping up on the same night is not a coincidence, it is a read on AI spend finally showing up in someone else's revenue line.

HP is the counterexample and nobody is talking about it. HPQ is -11.92%. Same night, same AI narrative, and the PC and print business got taken apart. When a theme is real, it is still only real for the companies actually inside it.

Retail beat across the board this morning and it is being ignored. Dollar General $2.48 against $1.99 expected. Dollar Tree $1.39 against $1.03. Burlington $2.37 against $1.98. Best Buy $1.47 against $1.37. Four for four, all before the open, all buried under Nvidia.

The setups on my screen

Research notes, not recommendations. Graded here tomorrow.

Momentum long, Marvell (MRVL). $253.30 premarket, +3.34%. It reports after the close tonight, which makes this a dated binary rather than a drift trade. Invalidation is a close back under $240.

Fade watch, Okta (OKTA). $158.79, +18.13%. An 18% gap on a software print is the kind of move that historically gives back a third of itself by Friday. My fade band is $160 to $166 and I have no interest in it below that.

Pullback watch, Astera Labs (ALAB). $300.88, +3.64%. Third day running that I have written the $268 to $278 accumulate zone and third day it has moved away from me. At some point a zone that never fills is just a wrong zone, and I will say so if it happens again.

Research short, Strategy (MSTR). $125.75, +2.08%. This is yesterday's short and it is going against me. I am leaving it on the board rather than quietly retiring it, and the invalidation I said out loud was $130.

The account behind all of this holds $114.13 in cash. None of these are fundable at that size and I will not pretend otherwise. The point of publishing them is the record.

Today's calendar, all times ET

8:30, done. Weekly jobless claims. Retail earnings already out: Dollar General, Dollar Tree, Best Buy, Burlington, all beats.

Through the day. Jackson Hole continues.

After the close. Marvell, Workday, Autodesk, Ulta Beauty, Affirm, Elastic, SentinelOne, Rubrik, Gap. Marvell is the one the semi tape watches.

Tomorrow, 10:00. The new Fed chair's debut keynote. With Nvidia behind us, that is now the only scheduled event left that can reprice the week.

Week timeline Monday August 24 to Friday August 28 2026: NVDA earnings reported Wednesday beat and raised, new Fed chair debut keynote Friday at 10:00 ET.

A trading lesson I keep relearning

My best call yesterday made money and my worst call yesterday lost money, and the interesting part is that I had far more conviction in the loser.

The Nvidia call spread was a study. I wrote it up as an options exercise, published the mark, and treated it as a thing to observe. The Strategy short was the one I argued for. I stacked three independent bears on it: management's own words about credit outranking buybacks, the analyst my system weights highest, and an index consultation with real passive selling attached. Three unrelated sources pointing the same way. That is supposed to be the good kind of evidence.

It is up 2.08% against me this morning.

Here is the lesson, and it is not that the thesis was wrong. It might still be right. The lesson is that conviction is a feeling about how much work you did, and the market prices something else entirely. Three sources agreeing tells you the reasoning is sound. It tells you nothing about the timing, and timing is the entire difference between a good idea and a good trade.

This is exactly why I made the system grade itself in public. If I only tracked the calls I felt strongly about, I would have quietly dropped this one by Friday and remembered myself as being right about Strategy. The report card does not let me do that. It is the same discipline I build into client systems, and I wrote about the mechanics of it in how scheduled tasks actually run while your computer is asleep.

The report card

Four setups from edition No. 2, marked premarket this morning:

Report card for edition No. 2 setups marked premarket: NVDA call spread A, TSLA long +0.5% C, ALAB watch no entry C, MSTR short -3.8% D. Test account $114.13.

Options study, Nvidia. The August 28 215/225 call spread, published at a $2.72 mark. With Nvidia at $222.88 premarket the spread is deep in the money and worth multiples of that. I want to be precise about one thing: the live option mark timed out on me this morning, so I am grading this on the underlying and not printing a price I did not verify. Grade A on direction, price to be confirmed.

Momentum long, Tesla. $344 reference, now $345.60. That is +0.5% and the entry band never really got tested. Grade C.

Pullback watch, Astera Labs. $284.61 to $300.88. It moved up 3.64% and never came near the $268 to $278 zone. No entry, so nothing to grade well. Grade C.

Research short, Strategy. $121.16 to $125.75. That is 3.8% against the short. Grade D. First losing mark of the run and I am not going to bury it in a paragraph about the other three.

The account is $114.13, down $1.18 from yesterday's $115.32, with no positions and no orders I can find. That is the second unexplained drop in two days on an account that is not trading. Small money, still a real hole in my own books, and I am chasing it before I add anything to that account.

One piece of honesty about this edition. It first published at 10:05 AM ET, not before the bell, and the painted boards were added after that. The market was already open. Every price in it is a real premarket mark taken between 8:39 and 8:51, and I am not relabelling those, but the edition itself was late and you deserve to know that rather than infer it from a timestamp. Two mornings running it is the pipeline, not the writing, that has made me late. That is the thing I am fixing before I add anything else to this newsletter.

Where this leads

Building agentic systems like this one is what I do at Bell Consulting Solutions. Mostly for client delivery operations rather than trading, but the architecture is identical: connected data, skills that encode the judgment, a schedule, and a paper trail you can audit afterward. If you want to see a finished skill instead of a description of one, the brand kit skill is free and ungated. If you want to talk about a system for your own business, reach out directly.

And the question for today. I had three independent sources agreeing on a trade and it still went against me. What is something you were confident about recently for good reasons, that turned out to be right about the world and wrong about the timing?

One more time for the people in the back: not financial advice, entertainment and education only, based on what I may or may not be doing personally. See you tomorrow, boards included.