US Market Compass 🇺🇸

S&P 500Benchmark -- ----

Nasdaq -- ----

Dow Jones -- ----

One print (MSFT) repaired the AI tape [29][179], but macro didn't confirm: core PCE 3.3%, 30-yr near 2007 highs [175][177]. Two-tier tech, not broad risk-on; month-end flow contaminates it.

What moved and why

Known: The tape rallied on one print. Nasdaq +2.8% as Microsoft's quarter — Azure past $100B, capex framed as self-funding rather than cash-burn — restored the AI bid [29][161][179], delivering the largest single-day market-cap gain in US history, ~$450B [20][17]. Chips ripped alongside (Lam +17%, AMD, Micron) on MSFT plus Lam and Arm results [36][58]. Meta fell ~10% on soft revenue guidance against ramping AI spend [197][157][229]. After the close Amazon added ~10% to $258 as Jassy defended the $200B capex program [12][106]. Note: the panel's 1d readings are all +0.00% — a stale snapshot, not a flat session; the 5d column is where today shows up (MSFT +18.2%, META -9.4%).

I am updating a prior call. I have argued for four sessions that the split is funded vs. financed capex. Today contradicts the clean version: Meta is cash-rich and was still punished, while banks are in advanced talks to lend $15B to Nexus for a Google-backed Anthropic data center [19][39]. The surviving distinction is narrower and falsifiable — capex with revenue already attached (Azure's $100B) is paid for; capex on promise is not [179][157].

Inferred, and I flag it as inference: the melt-up owed something to forced flow. Situational Awareness is unwinding, Citadel bought its stock book [88][174], and the Street is reading that as capitulation [37]. BTIG's counter — that a fast momentum rebound is a trap — is the more disciplined prior given insiders are the most bearish in 20+ years [223][50]. Macro did not confirm risk-on: Q2 GDP 1.5%, June core PCE still 3.3% despite the first headline PCE decline since the pandemic [175][203], the 30-year near its 2007 high [177], Warsh's credibility openly questioned [48][208][211], Goldman on hold all year [172], mortgage rates at one-year highs [102].

Sector reads

Technology: Two-tier, not broad. Hyperscaler-adjacent semis and MSFT bid [36][58]; the multiple, not demand, is the risk in AMD-type names [98]. XLK is still -2.7% over 20 days — one day did not repair the month.
Utilities (-3.5% 5d): Genuinely negative new information. Exelon cut its data-center forecast on consumer pushback against development [7]. If AI load is politically contestable, the entire power-for-AI trade (GEV -11.8% 20d, VRT -24.3% 20d) reprices lower, not just on rates.
Financials (+1.2% 5d, best 20d ex-energy): Mastercard's 14% revenue growth and high-end Q3 guide [220][62], ICE/MarketAxess $6B [248], private-credit outflows bottoming [57][51].
Energy (+10.8% 20d): Supply risk, not demand — Black Sea and Egypt port attacks, Iran retaliation threats [10][225][249], partly offset by a Saudi-led maritime coalition [74].
Industrials (-2.3% 5d, CAT -9.0%): China's US-bound shipments fell in July with factory activity decelerating [16].

What matters next

Friday 7/31 open: does AMZN hold the +10% and does AAPL confirm [12][119]? Month-end flows contaminate the signal — judge it Monday.
Aug 3 ISM / Aug 7 payrolls: with core PCE at 3.3% and the Fed on hold [175][172], a hot print sends the 30-year higher and kills the bounce independently of earnings.
Falsifier for my capex call: more utilities echoing Exelon's data-center cut [7], or the Nexus $15B loan pricing wide [19].

Opportunities

Capex with revenue already attached

The surviving, falsifiable version of my funded-vs-financed call is that capex with revenue already attached gets paid for and capex on promise does not. Microsoft cleared that bar — Azure past $100B and management explicitly avoiding cash-burn mode — and gained $450B in a day; Amazon cleared it after the close, +10% to $258 on Jassy's defense of $200B [179][161][20][12][106]. Meta, cash-rich but guiding revenue soft into ramping spend, did not, and fell 10% [197][157].

Evidence (6)

Semi capex cycle broadening past the GPU

The chip move was not one name: Lam Research climbed 17% on its own results, with Micron and Sandisk carried alongside on Microsoft and Arm [36][58]. Test and custom-silicon demand is confirming independently at Teradyne, and Susquehanna raised AMD's target on server CPU upside [137][191][173]. This is the part of the AI trade with orders on the books rather than announcements, though the multiple — not the demand — is where these names break [98].

Evidence (6)

Payments and market structure, funded by volume not narrative

Mastercard's 14% revenue growth on 10% purchase volume with a high-end Q3 guide is a clean consumer-throughput signal that does not depend on the AI cycle resolving [220][221][62]. ICE paying $6B for MarketAxess says exchange operators are buying fixed-income flow at scale, and Blue Owl reporting private-credit withdrawals have bottomed removes a tail risk that dogged the sector [248][57][51]. This is the least crowded way to be long the tape right now.

Evidence (5)

Key risks

Momentum rebound is a trap, not a bottom

The melt-up owed something to forced flow: Situational Awareness is unwinding and Citadel bought its stock book, which the Street is reading as capitulation [88][174][37][14]. BTIG's counter — that a fast rebound in crushed momentum names opens another selloff — is the more disciplined prior given corporate insiders are the most bearish in over 20 years and Eisman is calling AI 'one trade' [223][50][210]. If this is right, today's +2.8% Nasdaq is the setup for the next leg down, not the repair.

Evidence (7)

Inflation credibility, not earnings, sets the ceiling

Core PCE at 3.3% with Q2 GDP at 1.5% left the Fed on hold and Goldman sees no cuts all year, while long-end yields rose anyway and Wall Street is openly calling it an inflation credibility shock at Warsh's Fed [175][172][211][48][208]. The 30-year near its 2007 high and mortgage rates at one-year highs are the transmission [177][102]. Aug 3 ISM and Aug 7 payrolls can kill this bounce independently of any earnings result.

Evidence (6)

AI capex is contestable — politically and in credit

Exelon cut its data-center forecast because consumers are pushing back on development, the first evidence that AI load faces a political ceiling rather than just a cost of capital [7]. Simultaneously, banks led by Morgan Stanley are in advanced talks to lend $15B for a Google-backed Anthropic data center — promise-funded capex moving onto bank balance sheets [19][39]. OpenAI cutting prices on GPT-5.6 models as customers scrutinize AI spend is a third crack: the revenue side is deflating while the capex side compounds [60][82].

Evidence (5)

11 GICS sectors

Sector ETFs are directly tradable. ETF call -- means the separate call is not generated and never inherits a stock call.

3L axes: relative strength · market breadth · volume confirmation. Axis states are shown live and do not alter the current call or scorecard.

Information Technology GICS sector

Bullish

Heat 9/10 Confidence 6/10

Sector ETF XLK Technology Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Invalid MainlineLagging20D excess -4.7% · Rank 11/11 · breadth 7/11 LogicEvidence-backed7 evidence Price/volumeWaitinghigh -6.2% · MA -1.7% · vol 1.58x

Stock Close 1D 5D 20D ADV Trend

MU Micron Technology 739.00 -9.9% -23.0% -36.0% $42.4B

NVDA Nvidia 190.01 -3.6% -10.4% -5.0% $26.1B

SNDK Sandisk 1015.89 -7.3% -36.5% -55.3% $23.0B

AAPL Apple Inc. 338.19 -0.6% +3.8% +16.9% $15.9B

AMD Advanced Micro Devices 429.56 -5.5% -22.2% -26.1% $14.2B

MSFT Microsoft Buy 390.54 -0.7% +0.1% +4.7% $12.5B

INTC Intel 81.88 -5.1% -20.2% -41.4% $11.9B

Close basis 2026-07-29

Microsoft's quarter — Azure past $100B with capex framed as self-funding rather than cash-burn — drove the largest one-day market-cap gain in US history and pulled the semis with it (Lam +17%, MU, SNDK) [179][20][36][58]. But this is two-tier, not broad: XLK is still down over 20 days, the risk in AMD-type names is the multiple rather than demand [98], and BTIG's warning that a fast momentum rebound is a trap is the more disciplined prior with insiders the most bearish in 20+ years [223][50]. I own the capex-with-revenue-already-attached side of that split.

Evidence (7)

Financials GICS sector

Bullish

Heat 5/10 Confidence 7/10

Sector ETF XLF Financial Select Sector SPDR Fund ETF call --

3LStage: Price/volume confirmation · Waiting MainlineLeading20D excess +4.7% · Rank 2/11 · breadth 7/11 LogicEvidence-backed7 evidence Price/volumeWaitinghigh -1.0% · MA +1.4% · vol 0.93x

Stock Close 1D 5D 20D ADV Trend

JPM JPMorgan Chase 344.71 -3.5% -1.0% +5.3% $3.4B

V Visa Inc. Buy 368.73 +0.6% +4.3% +7.5% $2.9B

HOOD Robinhood Markets 89.84 -3.1% -14.0% -10.4% $2.4B

Close basis 2026-07-29

Mastercard printed 14% revenue growth with purchase volume +10% and guided Q3 to the high end of low-double-digits, which is the cleanest read-through to Visa [220][62][56]. Market structure is consolidating at premium prices (ICE buying MarketAxess for $6B) and private-credit redemptions have bottomed per Blue Owl [248][57][51]. V is the pick over HOOD (record volumes but high beta into a tape where Coinbase just missed a third straight quarter [144][18]) and over JPM, which carries duration risk with the 30-year near a 2007 high [177].

Evidence (7)

Health Care GICS sector

Neutral

Heat 4/10 Confidence 5/10

Sector ETF XLV Health Care Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Waiting MainlineMixed20D excess +3.1% · Rank 5/11 · breadth 7/11 LogicEvidence-backed5 evidence Price/volumeConfirmedhigh -3.0% · MA +0.9% · vol 1.22x

Stock Close 1D 5D 20D ADV Trend

LLY Lilly (Eli) Hold 1210.02 -0.9% +4.0% +0.9% $2.8B

UNH UnitedHealth Group 420.57 -1.9% -2.5% +1.2% $2.4B

JNJ Johnson & Johnson 265.53 -0.4% +3.9% +4.6% $2.1B

Close basis 2026-07-29

My brief carried no health-care view and today's flow is mixed-but-benign: J&J's guidance cut was read as not cause for concern, Bristol raised, Regeneron and Labcorp beat [47][43][247][209]. Nothing here is a market-level signal, so this is genuine abstention rather than a hidden call. Lilly is the name I would actually hold — the $750M US manufacturing expansion with Resilience is capex attached to real demand, which is the same test I am applying to tech [190].

Evidence (5)

Consumer Discretionary GICS sector

Bullish

Heat 8/10 Confidence 7/10

Sector ETF XLY Consumer Discretionary Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Invalid MainlineLagging20D excess -4.2% · Rank 10/11 · breadth 7/11 LogicEvidence-backed4 evidence Price/volumeWaitinghigh -5.7% · MA -2.0% · vol 0.93x

Stock Close 1D 5D 20D ADV Trend

TSLA Tesla, Inc. 298.32 -3.0% -20.2% -29.1% $15.8B

AMZN Amazon Buy 226.65 -1.8% -7.4% -4.9% $9.4B

HD Home Depot (The) 338.27 -1.8% +2.1% -4.1% $1.3B

Close basis 2026-07-29

Amazon jumped ~10% to $258 after hours as Jassy defended the $200B capex program, versus a $235.50 basis close — a known, unpriced gap in the verification window [12][106]. AWS is the second instance of capex with revenue already attached, which is the surviving version of my funded-vs-financed call [129]. The caveat is real: month-end flows contaminate Friday's open and the honest read of whether the gap holds is Monday, not tomorrow [119].

Evidence (4)

Communication Services GICS sector

Bearish

Heat 7/10 Confidence 5/10

Sector ETF XLC Communication Services Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Invalid MainlineLagging20D excess -2.3% · Rank 9/11 · breadth 7/11 LogicEvidence-backed6 evidence Price/volumeWaitinghigh -6.5% · MA -3.0% · vol 1.30x

Stock Close 1D 5D 20D ADV Trend

META Meta Platforms 585.61 -1.3% -6.6% +4.0% $11.9B

GOOGL Alphabet Inc. (Class A) Hold 336.71 +0.9% -1.6% -5.8% $10.0B

GOOG Alphabet Inc. (Class C) 335.76 +1.0% -1.8% -5.0% $6.7B

Close basis 2026-07-29

Meta fell ~10% on soft revenue guidance against ramping AI spend — cash-rich and still punished, which is exactly what forced me to narrow my capex thesis to revenue-attached versus promise-funded [197][157][229]. Alphabet is the better-owned side: it is the counterparty behind the Nexus data-center financing rather than the balance sheet absorbing it, though AI search cannibalizing partner traffic (Reddit may walk from a $60M deal) is a live cost [19][39][96]. I hold rather than buy because Meta's weight makes the sector print negative even if GOOGL works.

Evidence (6)

Industrials GICS sector

Neutral

Heat 6/10 Confidence 4/10

Sector ETF XLI Industrial Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Waiting MainlineMixed20D excess -2.1% · Rank 8/11 · breadth 7/11 LogicEvidence-backed6 evidence Price/volumeWaitinghigh -4.3% · MA -1.4% · vol 1.02x

Stock Close 1D 5D 20D ADV Trend

GEV GE Vernova 900.28 -4.6% -8.6% -23.4% $3.1B

CAT Caterpillar Inc. 782.71 -6.9% -12.0% -26.5% $2.7B

VRT Vertiv Hold 223.04 -17.3% -25.9% -33.4% $1.6B

Close basis 2026-07-29

Genuinely two-sided today. AI power infrastructure got confirmation — Quanta blew out, Rolls-Royce says a hyperscaler nuclear deal is imminent, Vertiv guided to accelerating AI infrastructure spend, and Nexus is raising $15B [213][242][143][19] — but Exelon cutting its data-center forecast on consumer pushback is the first evidence that AI load is politically contestable, which reprices this trade lower on something other than rates [7]. Separately, China's US-bound shipments fell in July with factory activity decelerating, which is CAT's problem [16]. VRT is the pick as the highest-beta expression if capex re-accelerates; low confidence is the honest number.

Evidence (6)

Consumer Staples GICS sector

Neutral

Heat 3/10 Confidence 5/10

Sector ETF XLP Consumer Staples Select Sector SPDR Fund ETF call --

3LStage: Price/volume confirmation · Waiting MainlineLeading20D excess +3.2% · Rank 3/11 · breadth 7/11 LogicEvidence-backed3 evidence Price/volumeWaitinghigh -3.7% · MA +0.9% · vol 1.36x

Stock Close 1D 5D 20D ADV Trend

WMT Walmart Hold 114.22 +1.0% +4.5% +0.8% $2.5B

COST Costco 974.03 +0.8% +5.0% +4.1% $2.1B

KO Coca-Cola Company (The) 89.08 +0.9% +8.4% +9.6% $1.5B

Close basis 2026-07-29

No view in the brief and nothing today changes that; staples had no defensive bid in a melt-up session [29]. The one relevant datapoint is that June spending held up only because households dug into savings after an inflation surge, with core PCE at 3.3% [138][175]. That trade-down dynamic favors Walmart over Costco or KO if the consumer does crack, so it is the name I would own while doing nothing.

Evidence (3)

Energy GICS sector

Bullish

Heat 7/10 Confidence 5/10

Sector ETF XLE Energy Select Sector SPDR Fund ETF call --

3LStage: Price/volume confirmation · Waiting MainlineLeading20D excess +12.3% · Rank 1/11 · breadth 7/11 LogicEvidence-backed6 evidence Price/volumeWaitinghigh -2.5% · MA +3.5% · vol 0.69x

Stock Close 1D 5D 20D ADV Trend

XOM ExxonMobil Buy 156.75 +2.4% +1.5% +14.7% $2.1B

CVX Chevron Corporation 191.86 +2.3% -0.6% +15.7% $1.4B

Close basis 2026-07-29

The 20-day leadership is supply risk, not demand: Black Sea attacks lifted crude in Asian trade, a drone hit two ships at an Egyptian Mediterranean port, and Iran's Revolutionary Guard threatened to punish the aggressor after US strikes [10][225][249]. That premium is headline-reversible — oil fell more than 1% on a proposed Saudi-led maritime coalition — which is why confidence is only 5 despite high conviction on direction [74]. XOM over CVX for downstream breadth with jet fuel and refining margins elevated, and the majors are about to report booming profits [45][133].

Evidence (6)

Utilities GICS sector

Bearish

Heat 6/10 Confidence 6/10

Sector ETF XLU Utilities Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Waiting MainlineMixed20D excess +0.4% · Rank 7/11 · breadth 7/11 LogicEvidence-backed5 evidence Price/volumeWaitinghigh -4.2% · MA -1.7% · vol 1.29x

Stock Close 1D 5D 20D ADV Trend

NEE NextEra Energy Avoid 88.46 -0.9% -1.1% +0.8% $0.9B

Close basis 2026-07-29

Exelon slashing its data-center forecast because consumers are pushing back on development is genuinely new negative information, not a rates story — it means the AI load thesis has a political ceiling [7]. Eversource also missed on revenue, and the rate side is hostile with the 30-year near its 2007 high and mortgage rates at one-year highs [31][177][102]. NEE is squeezed from both ends — bond-proxy duration and AI-load optionality — and the counter-evidence (FirstEnergy bid on data-center demand) is not enough to offset [241][116].

Evidence (5)

Real Estate GICS sector

Neutral

Heat 3/10 Confidence 5/10

Sector ETF XLRE Real Estate Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Waiting MainlineMixed20D excess +3.1% · Rank 4/11 · breadth 7/11 LogicEvidence-backed5 evidence Price/volumeConfirmedhigh -2.5% · MA +0.6% · vol 1.40x

Stock Close 1D 5D 20D ADV Trend

WELL Welltower Hold 241.08 -1.0% -1.6% +6.2% $0.7B

Close basis 2026-07-29

No view in the brief and the sector is being driven by rates rather than fundamentals, with the 30-year near a 2007 high and mortgage rates at one-year highs [177][102]. Today's REIT prints were unremarkable — CubeSmart missed slightly on FFO, VICI and Broadstone guided in line [27][42][40]. Welltower is the only liquid candidate and its senior-housing demographics are at least independent of the AI capex debate, so I hold rather than avoid.

Evidence (5)

Materials GICS sector

Neutral

Heat 3/10 Confidence 5/10

Sector ETF XLB Materials Select Sector SPDR Fund ETF call --

3LStage: Mainline identification · Waiting MainlineMixed20D excess +1.8% · Rank 6/11 · breadth 7/11 LogicEvidence-backed3 evidence Price/volumeWaitinghigh -2.2% · MA +1.3% · vol 0.95x

Stock Close 1D 5D 20D ADV Trend

LIN Linde plc Hold 511.17 -0.0% +0.5% -1.5% $1.0B

Close basis 2026-07-29

Nothing in the brief and the day's evidence is weak-to-mixed: Olin missed on both lines, and China's factory activity decelerated in July with US-bound shipments falling [30][16]. The offset is the dollar's biggest two-day drop since Liberation Day, which helps exporters at the margin [112]. Linde is the only liquid candidate and is the defensive expression of the sector — industrial gas with electronics exposure — but I have no edge here.

Evidence (3)