The Musk Orchestra

Suncatcher M1 — orbit + contact after Transporter-18

Fact SpaceX Falcon 9 launched Transporter-18 from Vandenberg SLC-4E on Oct 1 (~11:32 a.m. PT); first stage landed LZ-4. Project Suncatcher M1 (Planet Labs bus with Google TPUs) deployed; Google confirmed contact and that the satellite is operating as expected (SpaceX mission page; Google Research blog Oct 1). Coming weeks: in-orbit data on TPU radiation / thermal / spaceflight stress — not a demonstrated orbital AI data center. Judgment Moves competitor-validation from launch-day framing → post-deploy / in-orbit hardware-test framing. Still one-server-class MVP; belief flip remains rad/HBM telemetry quality + 2027 formation, not today’s contact confirmation. Do not invent AI workload results or formation capability.

AI5 / AI6 RAM cut — Optimus volume constraint

Fact Musk (X, Oct 1): Tesla cut AI5 RAM to 72GB LP5 (half) and AI6 to 144GB LP6 (one-third cut) as “the only way to get enough volume for Optimus production,” also cutting cost; expects negligible Optimus performance impact because memory bandwidth (held constant) binds more than capacity. Judgment Moves LABOR/SILICON framing: Optimus rate depends on memory supply, not only chip design bragging rights. Does not flip Optimus unsupervised-hours gate. Unpublished Do not invent wafer starts, Optimus unit counts, or shipping dates from the spec change.

Suncatcher M1 — Transporter-18 launch day

Fact SpaceX targets Transporter-18 today from Vandenberg SLC-4E (window opens ~11:18 a.m. PT; backup Oct 2). Manifest includes Project Suncatcher M1 (Planet Labs) carrying Google TPUs — Alphabet’s first in-orbit Suncatcher pathfinder / AI-compute radiation+thermal test (CNBC Oct 1; SpaceX mission page). Judgment Moves the Suncatcher watch from calendar NET → launch-day / confirmed-payload framing on the competitor-validation thesis. Still a one-server-class MVP (~4 TPUs / ~1 kW class per prior brief) — belief flip remains post-deploy telemetry + rad/HBM, then 2027 formation. Do not invent orbit success or chip results before they exist.

Starship Flight 14 — first orbit + operational Starlink V3

Fact Flight 14 lifted from Starbase Pad 2 (~8:48–8:49 a.m. EDT Sep 28). Ship reached low Earth orbit for the first time (one Ship Raptor shut down early on ascent; controllers still cleared a single-Raptor orbit-insertion burn). Deployed 26 operational Starlink V3 satellites; mission truncated after ~1 orbit (plan had been ~6); controlled splashdown north of Hawaii (~11:57 a.m. EDT); Ship tipped and exploded on ocean impact as typical for splashdowns. Booster executed controlled Gulf splashdown (no tower catch this flight). Judgment Moves Gate A / LIFT: first-orbit + operational payload success logged. Remaining belief flip is whether cadence becomes boring (reuse, catch, iteration) — one spectacular flight does not flip the gate alone. Unpublished Do not invent future flight counts, catch timelines, or tanker/refuel schedules from this result.

Starship Flight 14 — FAA license; orbital attempt day

Fact FAA issued modified Starship–Super Heavy Flight 14 launch and reentry authorization (FAA statement cited Sep 26 issuance; coverage Sep 27). Sep 28 ~75-min window from Starbase Pad 2 targets first sustained-orbit profile plus first operational Starlink V3 deploy (26 sats, Group 31-1; Ship 41 / Booster 21; no recovery this flight; RTLS catch retargeted toward F15). Judgment Removes Gate A “pending FAA” blocker; belief flip remains first-orbit success + whether cadence becomes boring — license alone does not flip the gate. Unpublished As of this refresh, no confirmed liftoff/flight outcome — do not treat attempt-day as success.

Tesla Semi — volume production claim at Sparks

Fact Public coverage Sep 25 reports Tesla launched high-volume Semi production at the Sparks, Nevada factory and said deliveries “start now,” with redesign details (4680 cells, electric steering as autonomy groundwork) and Megacharger corridor framing. Judgment Moves Gate G from inauguration signal toward a production claim; belief flip remains sustained rate + customer fleet hours that matter for trucking economics — not the ribbon-cut. Unpublished Prefer disclosed production/fleet evidence over invented Semi unit counts or plant-capacity claims as proven rate.

Starship Flight 14 — WDR complete; FAA still open

Fact SpaceX completed Flight 14 wet dress rehearsal Sep 24 on Booster 21 / Ship 41 (fueled countdown short of ignition). Target remains Monday Sep 28 ~7:15 a.m. CT for first orbital attempt (~six orbits ~275 km, 26 Starlink V3, Pacific splashdown), pending FAA launch license. Judgment Moves Gate A readiness (technical path clear); belief flip is still first-orbit success + whether cadence becomes boring — WDR alone does not flip the gate. Regulatory open is the remaining near-term choke.

Google Project Suncatcher — competitor-validation, not disruption

Fact Google Project Suncatcher: MVP ~Oct 1 Falcon 9 Vandenberg — 1 sat, 4 TPUs, ~1 kW (≈ one DC server), Planet Labs bus; Gemini-from-orbit ~1 yr (pulled from early 2027); next 2 sats 2027; long-term 80+ clusters. Not a first — Starcloud H100 already in orbit; SpaceX talks GW-scale. Terrestrial Google↔SpaceX deal ~$920M/mo, cancellable ~90d. Judgment Tweet hype ≫ substance. Close to neutral / slightly positive for SPCX — not thesis-breaking. Bear (mod conf): Google is a credible orbital-compute competitor (owns chips; earlier rad/hardware data before SpaceX Q4 2027 compute launches). Why not: Google pays SpaceX and needs ~$200/kg mid-2030s → Starship; Google large SPCX shareholder (>$100B post-IPO reported); 1 kW proves almost none of the hard parts — 2027 formation is the real test; Starlink sats/lasers = SpaceX edge; concept validation supports the ODC valuation leg more than it competes now. Elevate Terafab (chip-cost answer to Nvidia margin / Google TPU). Watches: Oct 1 telemetry; rad/HBM (~2 krad HBM ground issues); 2027 formation; Google books Starship (bull) vs Blue/ULA/RL (bear); terrestrial renew/cancel. Position: no change to SPCX shares or 12/2028 LEAPs; doesn’t move SpaceX Q4 2027 timeline; dated risk still lock-up ~Dec 9. Semi Sparks inauguration today stays Gate G secondary.

Semi Sparks → Gate G; CernBasher calendar kept

Fact CernBasher two-week chart (Sep 21–Oct 2): Tesla Semi Sparks factory inauguration Sep 24; Roadster next-gen reveal Waco Oct 1 8:30pm ET; Q3 P&D expected ~Oct 2. SpaceX: F9 USSF-385 Vandenberg Sep 26; Starship F14 first orbital attempt Starbase NET Sep 28 pending approval (Gate A); F9 Starlink 15-25 Sep 30; Crew-13 NET Oct 1 (Dragon valve-leak delay). Judgment Semi is elevated to Gate G (production/fleet belief flip — not calendar-only). F14 remains Gate A / lift. Roadster + Crew-13 stay founder-attention / cadence context. Unverified Same post teased an item that “starts with M and ends with r” — leave as footnote only; do not elevate to a gate.

F14 slip to Sep 28; Terafab trademark suit

Fact SpaceX said Flight 14 now targets as early as Monday Sep 28 (pending approval) — ~6-day slip from NET Sep 22; still first planned orbital attempt with Starlink V3 (26) in plan. Same window: Tesla + SpaceX (+ SpaceXAI) filed in W.D. Tex. Austin vs TERA-print LLC over the “Terafab” name (declaratory judgment after C&D). Judgment Moves Gate A / lift timing ~6 days; Gate D / Risk gains a second Terafab legal front beside the Grimes County records fight — disclosure/brand friction, not capacity proof.

Starship Flight 14 targets first orbital insertion

Fact SpaceX now targets Flight 14 as early as Sep 28 (pending regulatory approval; was NET Sep 22): first planned Starship Earth-orbit attempt (~275 km, ~6 orbits over ~10 hours) plus deploy of 26 Starlink V3. Judgment Moves Gate A / lift Dep — orbital payload path is the next belief flip after suborbital learning; still not “boring cadence,” and thermal/power for Starmind remains a separate falsifier.

X/SpaceXAI dismiss Apple claims; All-In structure non-denial

Fact X Corp and SpaceXAI moved to dismiss Apple claims in their Texas antitrust suit (OpenAI claims continue; terms unpublished). At All-In with Shotwell, Musk declined to rule out TSLA–SPCX structural action given collaboration depth — no deal announced. Judgment Moves Risk (Apple-distribution overhang) and Focus/governance watch; does not change LIFT · SILICON · LABOR binding constraints.

Terafab records suit; SPCX Day-90 / affiliate lock-up eligibility

Fact SpaceX sued in Travis County to block Grimes County release of Terafab tax-abatement application and negotiating records (AG had ordered most withheld claims rejected). SPCX Day-90 lock-up tranche became eligible Sep 9; Rule 144 affiliate block eligible Sep 10. Judgment Moves Dep/Gate D/Risk watches on JV disclosure boundaries and float absorption — not wafer starts or forced selling.

Board grounded on value routing

Judgment Anchor is the SPCX↔TSLA map (Starship, Starmind, Optimus, Terafab seam; Autonomy adjacent). Constraint spine: LIFT · SILICON · LABOR · FOCUS (Focus = scarce founder bandwidth). Downstream sections are Grok barriers and belief thresholds — mechanisms first, not a reconstructed ops ledger.

Is Starmind lift-bound or physics-bound?

Judgment watch If Starship succeeds and Starmind still stalls, the bottleneck was never mass-to-orbit — it was power/thermal/radiation/networking. That split decides whether “more flights” is the right KPI.

How to check Track whether post-success language migrates to radiator mass, power budget, and bit-error rates. If it does, physics was the bet all along.

Does Optimus prove itself inside the house first?

Judgment watch External humanoid revenue chatter is noise until Tesla/SpaceX factories quietly replace shifts. Watch for boring internal utilization, not keynotes.

How to check Prefer qualitative ops trust (“we’d rather have the robot than the contractor”) over unit shipments. Absence after repeated demos is a negative update.

Terafab: one roof, two gods

Judgment watch Track whether D3 rad-hard and AI6-class efforts get ring-fenced capacity and separate success metrics. Shared slogans without capacity walls = future conflict.

How to check Look for kill criteria and capacity walls in any serious JV description — not wafer-start theater.

Who pays for which dream?

Judgment watch When capital is tight, does public TSLA fund space-adjacent optionality, or do space milestones have to stand alone? The answer is a governance fact, not a vibe — and it is mostly unpublished.

How to check Capital actions and disclosure boundaries under stress — ambiguity dissolving after a miss is the signal.

Focus as the hidden critical path

Judgment watch Correlated slips across Starship, Optimus, Terafab, and xAI after a political or platform fire drill would confirm Focus (scarce founder bandwidth) as the binding constraint — not any single engineering KPI.

Vehicles & Energy stay off-page · Autonomy is live

Scope Deliberate. This board is the orchestra’s routing spine (lift, orbital compute, humanoid rate, fab seam, founder focus) — not a full Tesla fundamental model. Autonomy stays adjacent: Autonomy Race · live board.

Watermark — constraint spine + live Autonomy link

Meta Standalone Grok brief: sticky Exec brief with LIFT · SILICON · LABOR · FOCUS; Dependencies / Gates / Risk / Log / Barriers cooked from first principles with Fact / Judgment / Unpublished labels. Autonomy Race linked live.