Minerva Morning Update

10/9/2026

Friday Morning Mish-Mash

No single thread. Thursday was an AI-revenue scare that hit the complex hard — Nasdaq −1.25% to 27,193, S&P −0.47% to 7,765, SOX −3.4% — and Friday’s open is a partial unwind plus two discrete catalysts that actually change positioning: SpaceX buying its way into the carrier stack, and Lumentum saying the optical book is sold out through early 2029. Oil is slipping after Trump pushed any Iran strike past the Nov. 3 midterms. Futures: Nasdaq +0.7–0.8%, S&P +0.4%, Dow flat to +0.1%. Indexes were still on track for slim weekly gains into the print. Banks and CPI next week. Don’t force a thesis.

★ MOST ACTIONABLE · $SpaceX(SPCX.US) $AT & T(T.US) $Verizon(VZ.US) $AT & T(T.US)MUS $American Tower(AMT.US) $Crown Castle(CCI.US) $SBA Communications(SBAC.US)

SpaceX Buys the 800 MHz Book — Starlink Mobile as a Fourth Carrier

SpaceX agreed to buy Grain Management’s nationwide 800 MHz portfolio, about $8 billion cash per the WSJ, subject to FCC approval. Up to 14 MHz of paired low-band. The point of the band is indoor penetration — walls and tree cover — which mid-band 2 GHz direct-to-cell does not do. SpaceX’s line: with this plus the Gen2 constellation (FCC cleared 15,000 V2 satellites this week, company claims >100x current bandwidth), Starlink Mobile becomes a major US carrier, satellite plus terrestrial, not a dead-zone add-on. Musk: “complete phone coverage in America.”

Tape: SPCX +3.7% premarket. TMUS and T −6% to −7.5%, VZ −5.3%. Towers the other side — AMT, CCI, SBAC +7–8%. Grain had only just taken this book off T-Mobile in August for cash plus 600 MHz. Verizon’s response is the bear case in one line: spectrum without a terrestrial network is empty airwaves. Capacity in dense urban cores is the open question; New Street-type framing is that a Starlink broadband + mobile bundle (Grok in the package is the xAI kicker being floated) is the real product, not a like-for-like replacement of macro RAN on day one.

Swing read: this is a narrative shock, not a completed network. Incumbent multiple compression can run for days while the FCC clock and build economics get debated. Towers are the hedge — more nodes if Starlink densifies, or a relative-safety bid if the market decides the moat just cracked. EchoStar (ECHO +4%) is the prior spectrum seller sitting in the same complex.

★ MOST ACTIONABLE · ▲ POSITIVE · $Lumentum(LITE.US) $Coherent Corp.(COHR.US) $Applied Optoelectronics(AAOI.US) $Ciena(CIEN.US)

Optics: Sold Out Through Early 2029

Lumentum CEO Michael Hurlston, Tokyo, Friday: optical components “completely sold out” through early 2029. Cannot meet ~70% of demand for some products through next year; still short ~30% of demand on others through 2028. Six months ago the sell-out date he was using was 2028. “We are very, very far behind what customers want… no end in sight.” Nvidia put $2 billion into LITE and $2 billion into Coherent earlier this year. Street expects Lumentum sales to more than double this year.

Premarket: LITE +6%, COHR +5%, with CIEN / AAOI / GLW bid. This is the reversal of Thursday’s optical dump — COHR −9.6%, GLW −6.4%, LITE −5.6% — which was the OpenAI-revenue contagion, not a demand print. FinTwit will treat Hurlston as the clean rebuttal to the “AI capex air pocket” story. It is a supply comment, not a margin or pricing comment. Chase risk is real after a 180%+ YTD on LITE.

OpenAI: The $50B Scare, Then the $70B Walk-Back

What hit the tape Thursday: FT / Bloomberg — annualized revenue ~$50 billion at end of September, vs. the ~$70 billion figure that had been circulating. Enterprise run-rate +107% in Q3, total +77%, per a person familiar cited by CNBC. Anthropic was already through ~$65 billion annualized at end of July, which is why the comparison stung.

What repaired it overnight: Bloomberg — OpenAI told investors it expects to reach or exceed $70 billion annualized by year-end, enterprise-led. NVDA and AMD +~1.7% premarket, AVGO +2.1%. Roundhill DRAM and SOXX +~2%. Thursday damage: NVDA −2.9%, AMD −3.9%, AVGO −4.4%, MU −4.8%, SNDK −4.9%, ORCL −5.5%, INTC −5.3%.

Expert split worth keeping: Bloomberg Intelligence’s Mandeep Singh — revenue-accounting noise, token growth still says infrastructure demand is intact. The bear camp (and Thursday’s price) says a $20 billion hole versus the number the Street was underwriting is exactly how circular-financing anxiety gets a bid. Both can be true for a few sessions. Memory does not get the same overnight absolution optics just got.

▼ NEGATIVE · $Apple(AAPL.US)

Apple Cuts iPhone 18 Pro Orders

Nikkei: Apple told suppliers to cut component orders for iPhone 18 Pro and 18 Pro Max. Memory inflation forced price hikes; premium uptake is softer than the build plan. AAPL −2%+ premarket. This is the other side of the MU / SNDK tape — hyperscaler LTAs win, the handset OEM eats BOM or cuts units. Gross-margin story at the flagship, not a volume story, and the order cut says they are choosing units down over full pass-through.

Offset, not a savior: Bloomberg says a first touchscreen MacBook and a new iPad mini around Oct. 27, plus 14-inch MacBook Pro and iMac on M6, OLED, lighter Pros, Dynamic Island on the Mac. A product event is a headline; it does not fix a Pro-cycle build cut.

★ MOST ACTIONABLE · ▲ $Humana(HUM.US) · ▼ $Alignment Healthcare(ALHC.US)

CMS 2027 Stars — Bonus Money Just Got Allocated

Humana: 95% of Medicare Advantage members in 4.0-star-or-better plans for 2027; 42% in 4.5-star. HUM +15–16% premarket. That is the quality-bonus pool reopening. Alignment Healthcare: a 2027 plan at 3.5 stars, ALHC −20%+. Aetna (CVS): more than 69% of MA members in 4-star-or-better. Clover (CLOV) +10% in sympathy.

Stars are a multi-quarter payment event, not a one-day gap. The trade is the gap versus the bonus math — HUM’s rerating can extend if the 4.5-star cohort is big enough to move 2027 revenue; ALHC is a broken-rating story until management shows a path back to 4.0. Managed care was already a relative hideout while AI de-grossed on Thursday.

▼ NEGATIVE · $Delta Air Lines(DAL.US) · ▲ $Valero Energy Corp New(VLO.US) $Marathon Petroleum(MPC.US)

Delta Misses and Cuts — Fuel Is the P&L

DAL Q3 adj. EPS $1.72 vs. $1.89, revenue $17.59B vs. $17.66B. Full-year adj. EPS cut to $5.10–$5.60 from $6.50–$7.50 in July. Midpoint $5.35 sits under the $5.46 consensus. Annual fuel bill seen ~$6 billion higher than last year. Stock −3%. Demand in the premium cabin is fine; the barrel is not.

Refiners are the long side of the same fact. WSJ this morning: VLO, MPC, PSX set to crush already-strong June-quarter results as wars shrink product supply. Thursday tape already knew — MPC +4.8%, VLO +4.6%. Gulf product exports still estimated around half of pre-war levels. Distillate tightness is the structural piece; the flat price is the tradeable piece.

Macro: Oil Pause, Fed Still Leaning Higher

Trump said the US will not attack Iran before the midterms. WTI ~$90.50 (−1%), Brent ~$103 (−1.4%), off Thursday’s spike (WTI +3.6%, Brent +4.1%, Brent through $104–106 intraday on strike chatter and Hormuz tanker noise). A three-week geopolitical pause is not a supply fix. Hurricane risk still sits on Gulf output.

Fed: policy is 3.75–4.00% after Warsh’s September hike, the first in three years. Musalem Thursday: more firming required, another hike inside six to nine months if you want 2% inside ~18 months. Waller: more hikes likely, but not necessarily consecutive — October pause is open. Minutes: most participants saw another hike by year-end; several say current policy is not restrictive. Futures: October hold the base case (meeting is Oct. 27–28, a week before the election), December still live. 10-year around 5.25%. Today: Michigan sentiment and inflation expectations at 10:00 ET, WASDE at noon, rigs at 1:00 ET.

Analyst Desk

Barclays reshuffled eVTOL overnight: ACHR to Overweight, target $8 from $4.50; EVTL to Equal Weight from Underweight; JOBY to Underweight. A pair trade inside a group the market still treats as a narrative, not a certification calendar.

Morgan Stanley: CBOE to Overweight from Underweight, target $358 from $258 — S&P contract renewal removes the overhang, KPI contracts seen as upside outside the model. GEMI to Underweight, target $4.75 from $5.50 — crypto beta does not fix scale or a path to profitable growth.

JPMorgan: VFC to Neutral from Underweight, target $18 unchanged. KBW cut GNW and VOYA to Market Perform.

Also on the Wire

Samsung is the first tech name to forecast quarterly operating profit above ₩100 trillion (~$74.7 billion). Supportive for the memory-shortage narrative even after US memory was offered on Thursday. Do not confuse a Korean print with a US multiple.

PepsiCo beat on Thursday and was a rare large-cap green print in a red tape. Not a sector call.

RTX: $6.34 billion Standard Missile-3 production award. KOPN: additional $18.6 million US Army contract. Defense bid is idiosyncratic, not a group day.

Canada unemployment 6.5% in September. A small read-through for North American cyclicals, not a US print.

VEEA +40%+ on regained Nasdaq compliance after a reverse split and board cleanup. Micro-cap listing repair, not a signal.

Next week is the real earnings card: big banks Tuesday onward, CPI in the same window. Thursday’s software bounce (CSGP +8%, GDDY and ACN +6%) was a de-grossing of AI beta into “cheap quality,” not a fundamental reset. If the OpenAI walk-back holds through the cash open, that rotation can give some back.

NFA. DYOR. Not investment advice. Catalysts cut both ways; size for the gap, not the headline.