Chokepoint

Capital rationing is the real story right now — companies are hitting internal funding limits even when projects pencil out. We're in a constrained environment where allocation decisions matter more than opportunity cost. The usual "if it clears our hurdle rate, fund it" logic breaks down when the budget ceiling is hard.

This shows up across the board. Enterprise software spend gets scrutinized harder ($CRM), and even AI infrastructure buildouts face tighter internal gates ($NVDA demand still there, but procurement cycles stretch). The $SPY level reflects it — multiple compression when growth has to fight for dollars instead of riding free capital.

It's not a liquidity crisis. It's disciplined scarcity. The companies that win here are the ones that solve the highest-value problem first, not the ones with the best pitch deck.

Asked the AI one question: where does $BTC actually break?

Not where it is. Where it breaks.

Spot is $76,692. The nearest cluster isn't above — it's 0.76% below at $76,110, with $670.49M in long liquidations stacked there. Lose that level and the downside opens clean.

Positioning is one-sided: $74.83B in long liquidation fuel, nothing material on the short side nearby.

If $76,110 gives way, next flush zone is $69,712 with $888.37M waiting. Below that, the heaviest cascade pocket sits between $62,330 and $61,346 — $1.1B to $1.3B per level. That's where a dip stops being a dip and starts accelerating.

Upside, first squeeze trigger is $81,770. Back above it, shorts start getting pressured.

So: the break is closer underneath than above, and there's nothing hedging it.

Same question works on every asset with a liquidation snapshot. One question, one map.

Four straight sessions down, no bounce. First time since July.

$SPX gamma went from +$725M after $NVDA earnings to -$142M today. The entire post-NVDA rebuild erased in four days, zero sessions rebuilt anything.

The chain shifted hard. 7,500 to 7,650 now holds -$555M of combined negative gamma. 7,500 at -$112M, 7,600 at -$90M, 7,550 at -$88M. Five accelerators in the top ten strikes.

July's selloff ran roughly -$124M per strike. We're close to that now.

Above spot: nothing near. The flip sits 45 points up at 7,677. First positive strike is 94 points up at 7,725. That gap is the widest since mid-August. The index is sitting deep in the accelerator zone with no magnet within a normal session's reach.

$AVGO reports after the close.

$NVDA walked into a +$239M structure and cleared its flip on the gap. $AVGO walks into -$142M with 45 points to cover before it reaches the same boundary.

Same catalyst class, worse starting position. A beat has to be bigger to accomplish less.

Below 7,575, it hits the chain and the amplification compounds.

Range today: 7,575 to 7,700.

Dell just raised full-year EPS guidance by 42% in one quarter — from $17.90 to $25.50. Revenue guidance jumped from $167B to $192B. Street was at $174B.

Q3 guidance is $6.50 vs consensus $4.46. That's 46% above where analysts sat this morning.

Q2 results: revenue up 58% to $46.97B, adjusted EPS $7.04 vs $4.91 expected.

AI-optimized servers doubled to $16.4B in the quarter. Full-year AI server revenue now guided to $74B, up from $60B — triple last year. That's 38% of total company revenue from a product line that barely existed three years ago.

But here's the number that matters: $95B AI server backlog against $74B of AI server revenue guided for the full fiscal year. The order book is larger than a full year of shipments.

Two things follow. Dell is supply-constrained, not demand-constrained — same story $NVDA told us last week. And a backlog that size makes the guidance raise a scheduling exercise, not a forecast.

Traditional servers and networking also grew 122% to $10.53B. The old business is inflecting alongside the new one. PC business did not — commercial up 22%, consumer up 7%.

Watch conversion, not backlog. Everyone in this supply chain now has an order book bigger than their capacity. The question is who can actually ship it.

$DELL $SPY

Trump just said he "couldn't care less about Iran" but likes the U.S. position on the Strait of Hormuz, then asked when Iranians will stand up and fight.

Welcome to September — geopolitical risk is back on the table. The Strait matters because ~20% of global oil flows through it. If tensions escalate, you get supply shocks, crude spikes, and macro repricing fast. That hits everything from inflation expectations to Fed policy bets to risk-off flows into bonds and gold.

Markets have been ignoring Middle East risk for months. If this heats up, vol comes back in a hurry. Watch crude, the VIX, and whether defense names start outperforming. September historically rough for equities anyway — this doesn't help.

Credo's ($CRDO) concentration story just got messier. Top customer now 33% of revenue, second at 28%, third at 13%, fourth right at 10%. Same top three as last quarter, just shuffled. Fourth spot rotated — was a 10% customer before, dropped out, now back in.

That's not diversification. That's four customers driving 84% of revenue, with the top two alone accounting for 61%. Any hiccup in hyperscaler capex timing or allocation and the quarter swings hard.

The real tell: company sentiment cooling while the broader industry average improves. Net positivity still above peers, but the gap is tightening. Translation — $CRDO's narrative edge is eroding. When your customer base is that lumpy and your relative sentiment is compressing, you're not riding the wave anymore. You're hoping the next order doesn't slip a quarter.

This is classic late-cycle optics supplier risk. Revenue looks fine until it doesn't. Watch the next guide closely.

Fable 5.1 just shipped.

Without Alphractal MCP: even the sharpest model is basically trading on vibes.

With Alphractal MCP: 1,500+ live metrics spanning on-chain flows, derivatives structure, sentiment reads, and macro.

Intelligence matters. Context is the edge.

Palantir just won another TITAN production contract from the U.S. Army — 8 new ground station systems (4 Advanced, 4 Basic), plus tech integration and operational fielding.

TITAN is the Army's next-gen AI/ML deep-sensing platform. It stitches together data from space, high-altitude, aerial, and ground sensors to generate targeting intel for mission command and long-range precision fires. $PLTR is the prime contractor — meaning they're not just writing software, they're overseeing manufacturing, delivery, and the entire software stack. Partners include Anduril, L3Harris, Sierra Nevada, and others.

Why this matters: TITAN positions Palantir at the center of the Army's AI-enabled battlefield architecture. It's a prime defense contractor role, not just a software vendor seat. That means deeper ownership of mission-critical systems and a bigger slice of the long-term defense budget.

Akash Jain (Palantir USG President & CTO) nailed it: "TITAN was shaped by the Soldiers who used it, in the conditions they used it in, against the standard they set. That is the only way a system like this earns its place in the field."

This is the kind of contract that expands the TAM and reinforces the moat. If you're long $PLTR, this is the signal you want to see.

Dark pool capital rotated from index dips to single names. Five weeks of buying every $SPY and $QQQ dip just stopped. $SPY -$47M, $QQQ +$6M — both dead flat while S&P dropped -0.71% and Nasdaq -1.03%. The dip-buying disappeared, but the capital didn't. It moved to individual positions.

$NVDA +$500M. Fourth post-beat buy, biggest yet. The accumulation pattern is clear: sold before earnings, went flat on the beat, built incrementally Friday and Monday, now accumulating at scale. Twenty-one trading days since the 97:1 sell. Full cycle documented — sell, silence, re-enter, sell again, beat, flat, build, accumulate. The dark pool is back in size.

$AAPL +$356M. Strong buy. $IWM +$152M on the day Russell dropped -1.16%. Yesterday's equilibrium broke to the upside. The dark pool bought the hardest-hit index. Small-cap equilibrium lasted one session, then resolved to buy on weakness.

$MU -$288M. Fifth session of selling since OPEX. The dark pool is getting selective — stopped buying the basket, started picking names. The capital moved from index hedges to individual conviction. Watching the tape.

$DELL just dropped one of the cleanest AI infrastructure prints you'll see all year.

Revenue hit $47B vs $44.9B expected — up 58% year-over-year and a quarterly record. Adj EPS came in at $7.04 vs $4.91 consensus, more than tripling from $2.32 last year.

The real story is in the AI server segment: $16.4B in revenue, roughly doubling YoY. But the forward look is even sharper — Dell booked $60.9B in AI server orders during the quarter. That pushed the AI backlog to a record $95B, up from $51.3B last quarter. That's one of the largest order books in the buildout right now.

Traditional infrastructure also ripped: servers and networking did $10.5B, up 122% YoY; storage added $4.9B, up 26%. The entire stack is moving.

Management raised FY27 revenue guidance to ~$192B from the prior $165B–$169B range. That's not a tweak — that's a re-rate.

This is what happens when you own the boring middle of the AI supply chain: the racks, the cooling, the integration labor. High-margin downstream gets the headlines, but the infrastructure layer is where the actual capex lands. And right now, that capex is landing hard.

Options flow on $PANW is reading neutral to bearish right now. Not seeing conviction either way — could be positioning for downside or just hedging. Watch the gamma levels and vol structure before leaning too hard into a directional call.

$MDB earnings tonight — Alphatica's signal reads neutral-to-bearish with a 13.87% expected move priced in. That's a wide band for MongoDB, and the setup doesn't favor holding through the print. Risk management mode here: if you're long, consider trimming or hedging. The options market is pricing uncertainty, not conviction. Slightly bearish lean into the release means the path of least resistance might be down if they don't beat cleanly on both revenue growth and margin guidance. Watch how $SPY behaves into the close — if indices are wobbly, that amplifies single-name risk. Not a high-conviction long setup.

Alphatica's earnings risk signal on $DELL ahead of tonight's print: bearish/neutral. Expected move is 9.25%. It's barely bearish, so manage risk accordingly.

$SPY $DELL

$SPY at $761.51, down 0.72%. Oil just spiked 5-7% in one session — WTI at $90, Brent at $94.55 — because Iran and the US are trading strikes. That's an inflationary supply shock landing right on top of Warsh's Jackson Hole warning that the Fed should focus on prices. The macro picture just changed.

The $760 bull flag stop is one dollar away. Six tests in two weeks. Price is closer than it's ever been. Between here and $760, there's over $600M of negative gamma — every strike is negative. The structure is pushing toward the cliff, not away from it.

Morning to afternoon, the engine swung from -1.1M to -87.1M. Call premium collapsed from 76% to 53% — nearly neutral. The afternoon put wave fired at 5.6:1. Geopolitical headlines accelerated the hedging cycle.

But the skew inverted to -0.41%. Calls are more expensive than puts. Twelfth inversion of the series. Same signal every time: institutions are buying downside puts with the left hand and paying MORE for upside calls with the right. They expect this to resolve higher.

The tension is real. Oil at $90+ is an inflationary event. Rate hike probabilities may tick higher. But the AI infrastructure boom is still running — $NVDA guiding 70% growth, the cup and handle targeting $829. Two forces, same market. The structure tells you which one the money is following.

Support layers: $760.52 bull flag stop (one dollar), $751.72 falling wedge breakout ($9.79 below), $750 at -$98M (next accelerator), $745 dark pool floor ($16.51 below). If $760 breaks on a daily close, the 93% bull flag fails. The falling wedge breakout at $751.72 absorbs. Three layers in $16.

ATH drawdown is -2.28%. The 3% level is $755.92 — $5.59 below. The framework is approaching its outer boundary.

$766 is the flip. $761.51 is price. $760 is the cliff. $751 is the breakout floor.

Oil changed the equation. The $760 test is live. The skew says buy. The headline says hedge.

A Salesforce exec just quantified the AI productivity gap on a customer call this morning: AI spend up 30%, measured productivity up roughly 3%. That's a 10-to-1 gap between what customers pay and what they actually get back.

He surfaced it while explaining why $CRM is rebuilding its pricing model. The fix they're selling: outcome-based pricing. Help agents billed per resolution. No resolution, no charge. They confirmed they're extending it to sales and service.

Underneath that is a second bet worth watching. Slackbot runs on Claude. Salesforce sells effectively unlimited use inside a per-seat subscription while the model provider bills per token. An exec described a customer choosing Slackbot specifically to avoid token pricing. They also said routing work to cheaper models is a gross margin lever they manage.

So Salesforce is buying inference wholesale, selling it flat, and pocketing the spread. That works while inference costs fall. It inverts if they stop.

The numbers they gave: Agentforce and Data ARR up 200% to $3.9B. Agentforce ARR up 200% to $1.5B. Consumption up 97% to $7B. Half of Agentforce bookings are customers refilling credits.

Investor Day is September 16 at Dreamforce.

$CRM $SPY

$AVGO reports earnings tomorrow afternoon — and the technical setup is tightening in a way that matters.

Between now and roughly 3:51 p.m. tomorrow, expect some degree of snapback as traders reposition ahead of the print. Earnings are a binary catalyst here, and with implied vol pricing in a sharp move either direction, being heavily short at current levels looks like poor risk/reward.

This isn't about bullish conviction — it's about respecting the gamma and flow dynamics into a known event. Shorts get squeezed into earnings all the time, not because the thesis was wrong, but because the timing was.

Watch how $AVGO moves relative to $SPY and $QQQ into the close tomorrow. If you see coordinated buying pressure across semis and mega-cap tech, that's not random — that's positioning.

Stay tactical.

Red candles killed any momentum we had today — two headlines did the work:

1. Explosions reported across southern Iran: Qeshm island, Bandar Abbas, Chabahar, Konarak port. Chatter about possible new U.S. airstrikes, with Konarak naval base hit multiple times.

2. U.S. embassy in Jerusalem just told Americans in the Middle East to "exercise heightened vigilance" — diplomatic code for "things could get worse."

Crude is sitting just under $90. Last time we saw that level was July 23rd.

This is the kind of supply-side shock that doesn't show up in your quant model until it's already moved. If Iran escalates or supply routes through the Strait of Hormuz get threatened, energy names and inflation hedges wake up fast. Watch how this plays into Fed expectations and whether risk-off flows start hitting tech multiples.

Oil just caught a bid on geopolitical noise — explosions reported in Southern Iran, which matters because any supply disruption in the Strait of Hormuz corridor tightens the global energy stack fast. Watching $USO for follow-through here.

This kind of headline risk usually spikes crude futures first, then bleeds into broader equity vol. If energy names start running and $SPY gamma flips negative, we could see some intraday chop as dealers hedge. Keep an eye on refining margins and whether this is a one-day pop or the start of a sustained supply scare.

After a 23% rip on earnings, $CRM is now hunting for support. Stock tagged $262.32 intraday, now sitting near $256 — down ~1% today.

A Shooting Star just printed on the daily. Not a statistically bulletproof signal, but historical comps suggest:

• ~11% average drawdown
• ~21-day average duration

Price discovery in real time. Worth tracking if you're positioned or watching for re-entry.

$SPY $CRM

$SPY opened Tuesday at $762.01, down 0.66%. First day of September continues August's selloff pattern. The $760 cliff sits 0.3% below — same accelerator that's held five times in August.

The gamma surface rebuilt over the weekend. GEX at -$1.40B, deepest reading since pre-NVDA week. Every strike from $760 to $766 is negative gamma. $762 at -$79M sits right at price. $760 at -$267M is the cliff — two dollars down.

But the flow tells a different story. Premium at 76.4% call, highest call percentage heading into a selloff morning in the entire series. Net +$470M. Institutions aren't selling into September. They're buying the dip. Engine flat at -1.1M — no mechanical selling pressure. The decline is structural positioning, not directional conviction.

IV skew flat at +0.07%. No expansion, no repricing. The options market is calm while price tests the range bottom.

Support stack underneath:
$760.52 bull flag stop
$751.72 falling wedge breakout (1.4% below)
$745 dark pool cost basis (2.3% below)
$708.87 wedge invalidation (7.0% below)

Four independently sourced layers, all tested during July's selloff. The architecture is deep.

The $762-$779 range that's held for three weeks is being tested at the bottom again. Same cliff, same level, sixth test. The gamma surface is doing the work. The money disagrees with the price action.

$766 flip. $762 price. $760 cliff. $751 breakout floor. $779 target. September begins. The range holds. The test continues.