Portfolio Q2 Earnings Sector IC Memo

Executive Summary & Action Plan

Verdict

NEUTRAL / REBALANCE — operating evidence is strong, but price expectations and cash-conversion risk are uneven. Keep the portfolio thesis and do not chase the aggregate basket. Top Pick: Microsoft (MSFT), an external peer, after comparing all 26 names on the same five-part scorecard.

Loading chart

Abstract

This review covers every operating company in the refreshed portfolio and six major peers, with ITA excluded and no confidential position data used. Twenty-one companies have reported the June-aligned quarter; NVIDIA, Micron, Marvell, Broadcom, and Planet Labs remain pending, so their latest available quarters are labeled rather than passed off as Q2 results. Eleven reporters raised or clearly strengthened guidance, and demand expanded across compute, memory, optics, power, cloud, and space. Valuation is the constraint: only four of 26 names have positive upside to the audited Base fair value, and the median gap is -16.2%. Several fast growers still consume cash, depend on customer advances, or carry peak-cycle margins after exceptional YTD gains. Microsoft remains the coherent selection because revenue rose 18%, Azure grew 43%, commercial RPO reached $678B, the next-quarter Azure outlook accelerated to about 45%, and FCF stayed positive. The aggregate stance is selective, not broadly bullish.

Forecast Path

The scoreable target is a frozen 20-company basket: 5% initial weight per security, distributions reinvested, no rebalancing, and corporate actions handled by the resolution rule. The Base path assumes the four pending reports broadly validate current demand; the Bear path reflects simultaneous guide cuts and multiple compression after a median +33.6% YTD portfolio move; the Bull path requires broader cash conversion, not merely revenue beats. Probabilities vary by horizon in the prediction sidecar; the chart displays the 12M vector.

Loading chart

Action Plan

Hold aggregate risk near neutral until the five pending reports settle; do not infer trades from actual portfolio weights, which were not used.
Stage incremental exposure in thirds: after the August 26–September 3 report window, after the first Q3 guide reset, and only after cash conversion confirms the revenue signal.
Rebalance single-name concentration created by extreme YTD winners; use the equal-weight basket as the research benchmark, not as a mandate to sell every outperformer.
Add MSFT to the focused research list as Top Pick; require Azure growth near the guided level, ex-frontier-customer backlog breadth, and positive FCF before upgrading conviction.
Treat constituent Base fair values as 12-month decision anchors, not as the portfolio forecast; rebuild each pending name after its next release before acting on the gap.
Re-underwrite immediately if two of the five pending companies cut demand guidance or the basket reaches the 1M Bear level.

State of the Sector

The common factor is AI infrastructure, but the earnings quality is not common. Merchant compute and foundry remain strong; memory and storage show peak-cycle economics; optics and power are converting backlog into revenue; hyperscalers are growing while FCF absorbs capex; space names are scaling revenue but still rely on external capital or customer advances. The correct portfolio question is therefore not “did AI demand grow?” It did. The question is which companies convert that demand into durable cash without requiring a perfect multiple.

Forecast Drivers & Cycle Map

Node	Current evidence	Next confirmation	Principal failure mode
Compute / foundry	NVDA latest revenue +85%; TSM Q2 revenue +36%	NVDA, MRVL, AVGO pending prints	guide deceleration or packaging bottleneck
Cloud / software	Azure +43%; PLTR +93%; cloud backlogs expanded	Q3 cloud growth and Copilot/agent monetization	capex outruns revenue and FCF
Memory / storage	MU +346%; SNDK +372% from a low base	next pricing, capacity, and gross-margin guides	peak-margin mean reversion
Optics / power	COHR +34%; GLW core +17%; BE +166%; GEV orders +88% organic	backlog conversion and working-capital quality	deposits reverse or capacity slips
Space	SPCX +92%; RKLB +62%; RDW +90%	Starship/Neutron milestones and task orders	capex, dilution, fixed-price execution

Selection & Dispersion

Score = reporting integrity (20%), revenue/earnings quality (25%), guidance (25%), cash/balance sheet (20%), and expectations/risk (10%). Prices are the cross-checked 2026-08-13 U.S. close. “Base FV / 12M target” is the rounded, self-derived fair-value anchor under a one-year convergence assumption; Street targets receive zero weight. Bear and Bull are scenario anchors, not a statistical interval. “Pending” names remain capped until the next release.

Tier	Company	Status	Score	Close	Bear / Base FV / Bull	Upside to Base	Selection view
A	Microsoft (MSFT) · external	Q2 reported	8.8	$496.88	$410 / $525 / $675	+5.7%	Top Pick — best positive-gap evidence and cash quality
A	TSMC (TSM)	Q2 reported	8.6	$430.49	$230 / $320 / $395	-25.7%	operating quality is exceptional; price exceeds Bull
A-	ASML (ASML) · external	Q2 reported	8.2	$1,847.90	$1,100 / $1,504 / $2,050	-18.6%	monopoly quality; cash collection and multiple risk
A-	GE Vernova (GEV)	Q2 reported	8.0	$1,049.42	$600 / $885 / $1,250	-15.7%	orders are strong; normalize advance-driven FCF
B+	AMD (AMD) · external	Q2 reported	7.9	$483.01	$250 / $430 / $700	-11.0%	Helios upside needs the modeled margin ramp
B+	NVIDIA (NVDA)	Q2 pending	7.9	$225.30	$125 / $221 / $320	-1.9%	close to FV; August 26 settles the next mark
B+	Palantir (PLTR)	Q2 reported	7.8	$179.01	$70 / $130 / $235	-27.4%	elite execution; SBC-adjusted expectations remain extreme
B+	Corning (GLW)	Q2 reported	7.7	$158.54	$105 / $148 / $205	-6.6%	optical breadth; exclude customer advances from FCF
B+	Meta (META) · external	Q2 reported	7.7	$594.97	$440 / $635 / $825	+6.7%	highest liquid positive gap; lower confidence than MSFT
B	Micron (MU)	June quarter pending	7.6	$949.83	$310 / $474 / $650	-50.1%	mid-80s gross margin cannot be terminalized
B	Bloom Energy (BE)	Q2 reported	7.5	$236.22	$100 / $167 / $275	-29.3%	inflection is real; deposits, debt and dilution matter
B	Alphabet (GOOG)	Q2 reported	7.5	$343.94	$225 / $325 / $450	-5.5%	cloud strength versus long AI-capex payback
B	Coherent (COHR)	Q2 reported	7.4	$327.23	$210 / $290 / $400	-11.4%	optics conversion; negative FCF and dilution cap value
B	Amazon (AMZN)	Q2 reported	7.4	$265.13	$130 / $196 / $240	-26.1%	AWS reaccelerated; $220B capex delays convergence
B	Broadcom (AVGO)	June quarter pending	7.4	$417.82	$280 / $425 / $620	+1.7%	positive gap is too small before the pending print
B	Vicor (VICR)	Q2 reported	7.3	$226.64	$135 / $194 / $285	-14.4%	normalize royalty recognition and the tax benefit
B-	Sandisk (SNDK)	Q2 reported	7.2	$1,528.11	$380 / $719 / $1,450	-52.9%	current price exceeds even the normalized Bull anchor
B-	SpaceX (SPCX)	Q2 reported	7.1	$141.29	$55 / $105 / $195	-25.7%	re-underwritten post-IPO; public history remains one quarter
B-	Marvell (MRVL)	June quarter pending	7.1	$222.18	$105 / $185 / $300	-16.7%	FY28 target dependence and customer concentration
C+	Apple (AAPL)	Q2 reported	6.8	$305.26	$205 / $265 / $335	-13.2%	refund-normalized margin and memory costs constrain value
C+	Redwire (RDW)	Q2 reported	6.6	$13.20	$8 / $14 / $22	+6.1%	positive gap, but dilution and negative FCF lower confidence
C+	Rocket Lab (RKLB)	Q2 reported	6.6	$80.10	$22 / $45 / $82	-43.8%	current price already approximates the Bull case
C	Planet Labs (PL)	July quarter pending	6.4	$24.75	$7 / $14 / $27	-43.4%	September 3 must validate FY28 revenue assumptions
C	CoreWeave (CRWV) · external	Q2 reported	6.3	$106.29	$45 / $105 / $190	-1.2%	near Base, but leverage and capex keep confidence low
C-	Arm (ARM) · external	Q2 reported	6.1	$278.65	$70 / $150 / $300	-46.2%	licensing moat cannot offset the option-heavy price
D	Tesla (TSLA)	Q2 reported	5.1	$339.96	$85 / $200 / $480	-41.2%	Base does not capitalize unproven autonomy at full value

What to Watch Now

The next five releases are portfolio-level events, not routine updates: NVIDIA on August 26 and Planet Labs on September 3 are confirmed; Marvell around August 27, Broadcom around September 3, and Micron in late September remain estimates. Treat the first four semiconductor prints as an AI-demand breadth test and Planet as the missing space-data quarter. The decisive evidence is next-quarter revenue and margin guidance, not headline EPS.

What to Watch Next (Technology/Structure)

By the next full reporting cycle, track whether AI capex converts into cloud revenue and cash, whether memory gross margins normalize without collapsing, and whether power/optics backlogs convert without deposit reversals. In space, distinguish launch and constellation execution from long-dated addressable-market claims. Across the whole book, supplier concentration and financing terms matter more than aggregate demand headlines.

Allocation & Selection Notes

MSFT remains the selection because it is the only high-confidence, top-tier candidate combining a completed June quarter, accelerating core guidance, broad backlog, positive FCF, and a positive Base gap. META has a slightly larger gap but lower confidence because capex, legal exposure and Reality Labs widen the range; AVGO is report-pending and RDW is dilution-sensitive. These constituent marks are comparison evidence, not a portfolio-level Fair Value, so the root forecast remains the frozen 20-name basket.

Catalysts, Leading Indicators & Risks

Window	Leading indicator / catalyst	Bull confirmation	Bear confirmation
Aug 26–late Sep	NVDA, MRVL, AVGO, PL, MU releases	three or more constructive demand guides	two or more guide cuts
Monthly	20-name breadth vs SPY	at least 14/20 positive and >3pp excess return	8/20 or fewer positive
Quarterly	Azure, TSM, memory and optical conversion	cloud/foundry growth with stable FCF	growth holds but FCF deteriorates broadly
6–12M	AI-capex return and space milestones	backlog becomes revenue and cash	capacity, financing, or launch slippage

Falsification & Review Rules

Aggregate Base thesis fails if the normalized basket resolves at or below 88 at 1M, or if two of the five pending reporters cut demand guidance.
The MSFT selection is re-opened if Azure misses 40% growth, non-frontier backlog breadth reverses, or FCF turns negative without a bounded capacity explanation.
Move to Bear if at least eight of the 20 portfolio companies show worsening cash conversion for two consecutive reported quarters.
Review probabilities after each of the five pending reports, then monthly for breadth and quarterly for operating evidence; never revise probabilities from price alone.

Portfolio Q2'26 Earnings Review — 20 Holdings, 6 External Peers, One Selection (Portfolio Q2 Earnings Sector)