Two companies spent record sums on artificial intelligence and reported on the same evening. One got credit for it. The market did not reprice the AI capital-expenditure story Thursday night — it repriced the question of who is converting the spending into revenue the fastest, and answered it in opposite directions for two names the book has been short since the same week.

I. The Session

Equities rallied hard into the close. The Standard & Poor’s 500 rose 1.66 percent to 7,437.63, the Nasdaq Composite gained 2.78 percent to 25,122.18, and the Dow Jones Industrial Average added 1.19 percent to 52,208.06. The CBOE Volatility Index (“VIX”) collapsed 17.3 percent to 17.09, its steepest single-session decline in weeks, as the week’s two largest earnings uncertainties — Microsoft and Meta — both resolved and the market moved on to pricing the outcomes rather than the binary risk of them.

The index-level gain masks a session that, at the single-name level, was defined by one earnings night: Microsoft and Meta Platforms reported after Wednesday’s close, and the market drew a sharp line between them Thursday.

II. Microsoft and Meta Split the AI Capex Story

Microsoft closed at $451.10, up 15.53 percent from Wednesday’s $390.54 — the largest single-session move in the stock’s recent history. The market read the quarter as confirmation that Azure and Copilot are converting the company’s AI capital expenditure into consumption revenue at a pace that justifies the spending, the opposite of the capex-without-payoff concern this publication has tracked across the sector since late June.

Meta Platforms closed at $539.03, down 7.99 percent from Wednesday’s $585.61. Meta’s quarter told a structurally similar spending story — record AI infrastructure investment — without Microsoft’s offsetting evidence of monetization showing up in the same period’s numbers. Same input, opposite market verdict on the output.

The book was positioned on both sides of this divergence simultaneously, not by design as a single thesis but because Tranche 11 was built across several distinct GNRC-anchored and infrastructure-anchored pairs entered on different days in late July. Two pairs are short MSFT (P42, P43); two are short META (P47, P49). The MSFT move hurt the short side of P42 and P43 hard; the META move helped the short side of P47 and P49. The pair-by-pair detail is in Section IV, but the book-level arithmetic is worth stating plainly here: P42 and P43 combined for -$22,293.40 on the day; P47 and P49 combined for +$17,213.13. The earnings night that produced the largest two individual pair swings of the cycle net to a comparatively modest book-level drag, which is the diversification the multi-pair Tranche 11 structure was built to provide — though it is worth being direct that this was not fully anticipated when P42/P43 and P47/P49 were sized independently across different sessions; the offsetting behavior is a favorable outcome of how the tranche happened to be built, not a hedge that was deliberately engineered pair-against-pair.

What would prove this wrong going forward: if Microsoft’s Azure/Copilot monetization narrative fails to repeat next quarter, or if Meta’s spending starts showing comparable monetization evidence in Q3, the current opposite-sign response embedded in P42/P43 versus P47/P49 would not persist, and the book would need to reassess whether it is running four independent single-name theses or an implicit pairs-of-pairs spread that happens to net out.

III. P8 Closes at the New York Open

Earlier in the session, ahead of any of the day’s earnings-driven volatility, the desk closed P8 — long Apollo Global Management (“APO”), 922 shares at a $108.42 entry, against short the Themes Global Systemically Important Banks ETF (“GSIB”), 2,044 shares at a $48.93 entry — at New York’s 9:30am open. APO’s opening print was $119.28; GSIB’s was $62.30, its only trade of the session at capture time given the name’s customary thinness. The long APO leg contributed +$10,012.92; the short GSIB leg cost -$27,328.28; the pair closed at a net realized loss of -$17,315.36, moving the realized register from +$487,933.25 to +$470,617.89.

This close was executed on Mode B, PROVISIONAL marks — GSIB’s single opening print was cross-checked against the Yahoo Finance v8 chart API but did not receive the full two-source-to-the-cent confirmation the desk applies to settled closes, since a second independent trade never printed at the moment of capture. The figure stands as recorded; no re-mark against the 4:15pm settled close was requested.

IV. Pair Book — 33 Active Pairs, Settled Close

Book of record: 2026-07-30.brc-pair-book-canonical-v12.md. Realized register +$470,617.89 (frozen at this edition). Active unrealized -$81,088.03. Inception-to-date $389,529.86. All marks below are the July 30, 2026 New York settled close, two-source confirmed to the cent, ranked by day’s P&L, high to low.

Realized register (16 closed pairs, frozen at this edition): +$470,617.89. Active unrealized: -$81,088.03. Inception-to-date: $389,529.86.

A tranche- or portfolio-level equity-curve chart remains a blocked infrastructure item — no tradable instrument exists for “Tranche 11” or the book as a whole, and an internal chart would require a persisted daily P&L snapshot log the desk does not yet maintain. Flagged, not fabricated.

Looking Ahead

The desk holds 33 active pairs into Friday’s session, with the realized register at +$470,617.89 following today’s P8 close and active unrealized at -$81,088.03. STNG, VLO, AAPL, and AMZN all reported this week; their next-quarter reporting dates had not yet rolled forward in the data provider’s calendar as of this edition and will be confirmed in a future edition rather than estimated here.