AI Is Moving Up The Stack
The S&P closed up 0.6% from Monday’s open to Friday’s close.
The bottom of the flag has been reclaimed at 767.5 and is now a must hold level moving forward. If lost, the daily trend threatens flipping to down with Friday marking a lower high vs the all time high.
As we’ll see, the supporting evidence still leans risk on, however its impossible to ignore the drag of the the Nasdaq. Technology is notably weaker with semiconductors breaking down on Friday after Kevin Warsh’s Jackson Hole speech that basically confirmed a September rate hike.
Jensen Reminds Us Memory Is Expensive
Nvidia just generated $96.2 billion in quarterly revenue, up 106% from lat year, while data center rev surged 117% to $89 billion. Clearly demand isn’t the problem.
The problem might be paying for it all eventually.
As noted last week Nvidia’s largest customers have been warend that AI-server prices are going to surge 15% for systems shipping in early 2027.
The reason? Higher memory costs.
The chart above illustrates a more than 80% increase in the estimated cost of next generation memory for AI workloads.
Colette (CFO) was unusually direct on the call, warning that the company is experiencing ‘extreme pricing conditions in memory,’ with increases exceeding prior expectations and moving even higher into next year.
Even with this mention, prices of memory stocks unwound almost immediately the next morning, and closed the week below the unaffected price from Wednesday’s close.
Why does it matter? It continues to reinforce that expectations have likely topped in the memory trade, and expecting more +100% gains in MU, SNDK, SKHY, WDC, STX are unlikely.
Our thesis remains that memory stocks along side semiconductors more broadly likely chop instead of trend straight down, while new software leaders emerge as the next profit centers of AI. On August 9th, our newsletter wrote about ‘The AI Trade Is Moving Up The Stack,’ and the price action surrounding Nvidia’s earnings call seems to confirm that.
The only counter argument to this thesis I see is if the market has dramatically underestimated the exponential increase in infrastructure required to meaningfully realize the potential of agentic AI systems.
Software Strikes Back
Continuing down this thesis, its worth focusing on the not so dominant headlines from Wednesday night. Salesforce, CrowdStrike and Okta all crushed earnings, sending CRM +22.6%, CRWD roughly +20% and OKTA nearly +29% on Thursday while IGV rallied 2.5% and strength spread across names like NOW and ADBE.
Marc Benioff essentially used the quarter to declare the SaaSpocalypse over.
More importantly, the argument is shifting from ‘AI replaces software’ to ‘AI makes valuable software platforms more valuable.’ Models can generate intelligence, but Salesforce still owns the data, permissions, workflows and systems where that intelligence actually has to operate.
Back in February the concern was that a bunch of vibe coders were going to replace Salesforce… Not going to happen.
I do still believe that the more generic ‘commodity’ software platforms will be punished… Think MNDY, ASAN. Platforms that generally have task/project management as the primary offer have low switching costs and weaker data moats.
Jobs Could Decide What Happens Next
Warsh made the Fed’s position pretty clear on Friday’s Jackson Hole keynote speech. Inflation, not labor, remains the problem.
He also explicitly mentioned that interest rate policy is the primary tool for achieving the dual mandate. This essentially sets us up for a hike in September.
Markets bumped the odds of a September rate hike up from 35.9% to 59.7%.
If he doesn’t hike, boy oh boy will we have a credibility problem.
This upcoming Friday we get the final payroll report for August before the September FOMC meeting. Here’s what I’m thinking:
- HOT – 100k+ NFP & high wage growth – Yields up, market down
- PERFECT – 25-75k NFP & soft wage growth – Yields flat to down, market flat to up
- COLD – Negative NFP & low wage growth – Yields down, market up initially, then fades
Essentially, Friday isn’t just about the health of the labor market, the impact on yields will ultimately have a pretty substantial impact on the cost of financing AI as mentioned in last weeks newsletter: Price of Money Problem
As always the charts will light the way, so lets take a look!
Trade Brigade Live Squawk
The tape just changed. Did your strategy? Join the intraday squawk to trade with a plan, not emotion.
Check out this recording for a taste of what each morning session is like: 06/24/2026 Live Squawk
Broad Market Analysis
SPY – Weekly
SPY printed a green inside inverted hammer. That structure is neutral, but the broader trend remains constructive because the pullback is coming in from the highest high in the weekly count. This entire consolidation remains above the 38.2% retracement.
The previous all-time high near 760.00 remains the first major support. An overshoot into 754.75, where the weekly 10 SMA and daily 50 SMA converge, would still qualify as a weekly higher low. If structure is going to break, price would need to breakdown past 754.75 and then set a lower high anywhere on the weekly beneath 760. Likely something that takes more than one week to develop.
SPY – Daily
The expected move is neutral, with an upper bound of 778.57 and a lower bound of 761.70. Thursday’s post-Nvidia gap reclaimed 767.50, the daily 8 EMA, the daily 20 SMA and the prior bull flag range low. Friday rejected the top of that range, but it did not make a new bar-to-bar low or lose the reclaim level.
The daily trend remains up as long as 767.50 holds. A look below and fail would keep the door open for another rotation toward the all-time high. The bearish sequence begins with acceptance below 767.50, followed by an hourly lower high and a break of 762.00. That would flip the short-term trend down and open a deeper weekly higher-low test near 760.00 or 754.75.
SPY – Hourly
The hourly trend flipped back up after breaking above the multi-day balance range, and Friday produced another higher high before pulling back. Price is now testing whether it can establish a higher low near 767.50, where several daily moving averages and anchored VWAPs converge.
An early push below 767.50 followed by a quick reclaim would be constructive. A gap down, sustained acceptance below the level and an hourly lower high underneath it would invalidate the immediate bull case. The market can overshoot this area briefly, but it cannot spend meaningful time below if I’m going to remain bullish.
Market Internals – NYSE
Friday’s intraday reversal looked worse in price than it did underneath the surface. Volume flows were only modestly negative, the advance-decline line never reached a sustained trend-lower condition and cumulative TICK bottomed near negative 3,000, well short of the negative 5,000 threshold associated with substantial selling.
Financials and the Mag 7 helped support the S&P, which explains why the NYSE internals were significantly healthier than the NASDAQ internals. I can’t call Friday broad liquidation based on this.
Market Profile – /ES
Value moved higher throughout most of last week. Friday produced an engulfing value area with a slightly lower point of control, but value did not meaningfully break away to the downside. The main divergence was that value failed to follow price into Friday’s highs.
I do not view the lower value placement as automatically bearish. Buyers (if participating on the violent pullback) are positioned lower in the range, which leaves less trapped overhead supply if the market begins to rally. Think about the Toppling Effect™. Sustained strength could make those buyers whole and encourage support on subsequent pullbacks.
SMH – Semiconductor ETF
Semiconductors produced the market’s most concerning chart, with a lower high, a lower low and a declining 50-day SMA creating downside risk toward 551.00.
QQQ – Impact on S&P
Qs are in a weaker position vs SPY because they never reached a new all-time high, produced a lower low in the daily trend count and rejected an unfilled overhead gap despite having room remaining inside the expected move.
707.00 is the important dividing line. Above it, neutral chop remains possible. Below it, the potential lower high and head-and-shoulders structure become much more threatening. I would need to see a daily higher low above 722.00 to become meaningfully more bullish.
If you want the full take, including all of the nuance of breadth, risk appetite and much, much more, check out the weekend episode here:
Economic & Earnings Calendar
Top watches: RZLV, CRDO, DELL, PANW, MDB, GTLB, FCEL, AVGO, HPE, SNOW, NTAP, CIEN, PATH, AMBA, DOCU, ZS, IOT, ASAN, PL.
Swing Stock Scans
ALM – Daily Chart
Bull flagging after breaking the weekly downtrend. Ideally this would look below and fail on the range lows to offer a pullback entry. If the pullback is deeper than just a quick look below and fail, a structural higher low can emerge over 14.65 and keep this in play.
APD – Daily Chart
Ascending triangle getting tight near the apex of the move. Rising 20 SMA support below us and breakout level potentially over 311.
CXW – Daily
Flagging at highs after a really strong linear move from. June. Ideally this would shake out the flag lows and then go after kissing the daily 50 SMA, or just breakout over 34.50 with momentum… a catalyst would make this an even stronger candidate.
DE – Daily Chart
Wild earnings reaction, but holding hammers on Friday over the HVC and prior range high. Over Friday’s high on Monday could be a simple pullback long entry tactic to work back towards local highs after setting a higher low.
LW – Daily Chart
Weekly cup and handle showing up with unfilled earnings gap above still. Over 55.65 and theres room to get into the 58.75 gap close level. Measured move from the bull flag or cup and handle takes price far beyond that target if it does start to move.
LYV – Daily Chart
Chart isn’t perfect, but we have hammers at the 50 SMA in a nicely uptrending stock. Bias here, but I do believe that the ‘in person experience’ trade for consumers to connect outside of technology is a compelling one. Think SPHR (although last weeks close is getting ugly there).