NVIDIA PRINTED $96 BILLION
AND THE NASDAQ STILL CAN'T
GET OUT OF ITS OWN WAY
In June we asked whether the Nasdaq was in an AI bubble and gave what we thought was an honest answer: probably not a bubble in the earnings, definitely a bubble in the positioning. Two months later Nvidia handed us the cleanest test anyone could have asked for, and the result changes our view. Not about whether the numbers are real. About what is actually setting the price.
The Print
| Line | Q2 FY2027 | Year ago | Change |
|---|---|---|---|
| Revenue | $96.2B | $46.7B | +106.0% |
| EPS | $2.22 | $1.05 | +111.4% |
| Data center revenue | $89.0B | $41.1B | +116.6% YoY, +18.3% QoQ |
| Gross margin | 75.0% | 72.7% | +230 bp |
| Q3 guide | $108B ±2% | $57.0B | implies ~89% |
| Q3 gross margin guide | 74.0% | 73.6% | +40 bp |
| FY2028 outlook | ~70% revenue growth, described by management as supply constrained | ||
Read that table again, because the shape of it matters more than any single number. A company doing nearly a hundred billion dollars in a quarter is expanding gross margin while more than doubling revenue. That is not what saturation looks like. That is not what a demand air pocket looks like. And "supply constrained" is not a phrase you use when you are worried about orders. It is a phrase you use when you cannot build fast enough.
The Market Gave It Exactly One Day
The tape reaction is where this gets interesting. Nvidia initially traded down 1.3% on the release, worked its way to roughly +4% by the end of the call, and closed the following session up 8.7%. That was its largest single-day percentage gain since April 2025. A genuinely enormous move.
Then it was over. Two days later Kevin Warsh stood up at Jackson Hole, September hike odds jumped from 35% to 60%, and the Nasdaq closed as the weakest of the three major indices. Again. The same week, chip stocks had already dragged the S&P lower to start the prior week. The best earnings report in the history of the semiconductor industry bought the index roughly one session of leadership.
Our Actual Opinion
Here it is plainly. The AI trade stopped being an earnings trade some time this summer and became a rates trade. Nvidia is no longer priced on whether it will grow. Everyone now agrees it will grow. It is priced on what a dollar of growth arriving in 2029 is worth today, and that is a discount rate question, and the discount rate belongs to Kevin Warsh.
This is why the divergence we documented in July has been so stubborn. It is not a sector rotation in the usual sense, where money moves because one group's fundamentals got worse. Nvidia's fundamentals got dramatically better and the index still lagged. What changed is the denominator. Long-duration cash flows get repriced hardest when the front end of the curve moves, and the front end has done nothing but move since Warsh took the chair.
The Dow keeps winning not because the Dow is growing. It obviously is not. It keeps winning because most of its earnings show up this year and next, which makes them nearly immune to the thing that is actively repricing everything else.
The Strongest Argument Against Us
We would be doing you a disservice if we only made our own case, so here is the best version of the other side, and it is a real one.
- Multiple compression is finite. Growth is not. If the Fed hikes once or twice and stops, the discount-rate hit is a one-time repricing. Nvidia compounding at 89% and then 70% swamps it inside of four quarters. Under that path, every month the Nasdaq lags is a gift.
- "Supply constrained" is forward visibility, not hope. A backlog you cannot fill is the closest thing to a guaranteed revenue schedule that exists in technology. That is a materially different risk profile than a company guiding on pipeline optimism.
- Warsh himself is a partial AI bull. He spent real time at Jackson Hole on AI as a macro variable, citing annualized token sales north of $100 billion, up more than 500% year over year. A Fed chair who believes AI is a genuine productivity shock has a reason to tolerate a hotter nominal economy, not a reason to break it.
- Positioning cuts both ways. The crowded long that unwound in June is not crowded any more. That removes the mechanical seller that made the summer so ugly.
We think that case loses on timing rather than on logic. Everything in it is probably right eventually. None of it helps you in a September that has a live FOMC with a dot plot in the middle of it.
What We Would Actually Do With This
- Stop trading NQ off Nvidia headlines. The correlation that used to make that work has been broken since June. Watch the two-year instead. It has been the better NQ predictor than any semiconductor news for three months running.
- Respect that good news is not currently bullish. Strong AI capex numbers raise nominal growth and give the Fed more room to hike. In this regime a monster print can be a headwind for the index that owns the printer. That feels wrong, and it has been right.
- The divergence is the trade, not the anomaly. YM green with NQ red is the base case until the front end stops moving. Fading it on the theory that "it has to converge" has been the losing side all summer.
- The scenario that hurts everyone: a hot August CPI that forces the long end to move too. Right now only the two-year is repricing, which contains the damage to growth multiples. If the ten-year and thirty-year join, value stops being a hiding place and both legs of the rotation go at once.
Key Takeaways
- Nvidia grew 106%, expanded gross margin to 75%, guided to $108B, and called fiscal 2028 supply constrained. The earnings are not the problem.
- The stock got one great day. The index got roughly one session.
- Our view: this is a discount-rate story now, not an earnings story. The AI trade is a rates trade.
- The honest counter is that rate damage is one-time and compounding is not. That argument probably wins in 2027 and does not help you in September.
- Watch ZT, not SOXX headlines.
Sources
- Kiplinger · Nvidia Q2 FY2027 earnings live updates and commentary, August 2026
- Motley Fool · Nvidia earnings Aug 26 and historical post-earnings moves
- Federal Reserve · Warsh Jackson Hole keynote (AI and token sales section)
- Benzinga · September hike odds spike after Warsh's Jackson Hole speech
- Yahoo Finance · Market live blog, Friday Aug 28, 2026
- CNBC · S&P 500 falls to start the week, dragged down by a sell-off in chip stocks
- Market Maulers · Is the Nasdaq in an AI Bubble? (June 2026)
- Market Maulers · Index Divergence: YM at ATH, NQ/ES Lagging (July 2026)