SPCX · SpaceX (Space Exploration Technologies Corp.)

Prices below are public closes from the split/dividend-adjusted daily series; the reference close is 2026-09-02 at $140.71.

SPCX — Space Exploration Technologies Corp.

Current Thesis

Three pieces of positive news landed on 2026-09-02 and the stock closed lower on all of them. Oppenheimer's Timothy Horan raised his target to $280 from $250 and kept Outperform, lifting his long-term revenue expectation by roughly 10% on better compute economics. Bernstein SocGen reiterated Outperform at $248 the same day, citing launch services, orbital data centres and Starlink, and describing the Louisiana site as ten or more pads capable of 30 launches per day from 2028. Musk said publicly that Google and Anthropic came to SpaceX for compute because grid power is tight and that SpaceX built its own generation to move faster. The close was $140.71 against $142.23 on 2026-09-01, with RSI(14) at 49.2 versus 45.1 — momentum flat while price gave back ground.

The 2026-08-12 close of $146.15 is now fifteen sessions old and remains the only level in the August tape not taken back. Underneath, the floor has held: every close since 2026-08-21 has printed at or above the $135 IPO price, and $134.00 on 2026-08-20 is still the low close of the post-tranche window.

One genuinely new dated item entered the calendar. Reporting on 2026-09-02 put a 2026-09-30 capacity deadline on Google's contract — $920M per month running October 2026 to June 2029 — and said SpaceX reshuffled its data-centre leadership, moving rocket executives in, after reliability problems at Tennessee and Mississippi sites. That places an execution test three weeks out, inside the same window as two lock-up tranches and the filed Starship Flight 14 date.

The narrative is maturing — the space-plus-AI-power frame has been mainstream since the June listing, and the flow behind it has thinned visibly since 2026-08-12, with the 2026-09-02 stack of two target actions and a Musk compute headline producing a lower close rather than a test of $146.15.

Bullish and bearish views on SpaceX (Space Exploration Technologies Corp.)

The model's bull view on SpaceX (Space Exploration Technologies Corp.) (SPCX), in brief: The compute book is contracted and monthly, not a slide. The bear view: The headline stack did not move the ceiling. Both cases follow in full.

Bull Case

• The compute book is contracted and monthly, not a slide. Per 2026-06-27 reporting: Anthropic pays $1.25B/month from May 2026 through 2029 (over $40B, over 300MW of capacity), Google $920M/month from October 2026 to June 2029 (over $30B), Reflection $150M/month from July 2026 ($6.3B). Google's payments begin next month.
• Two firms marked the name up or reaffirmed above $240 on 2026-09-02 — Oppenheimer to $280 from $250, Bernstein SocGen at $248 — into a session the stock closed down.
• Q2 2026, reported 2026-08-04 and still the only print on the record: group revenue $7.81B (+92% YoY), adjusted EBITDA $3.54B (+191%), AI segment revenue $2.56B (+247%), Connectivity $4.3B on 1.7M net subscriber adds, capex $18.37B.
• Larger supply has already cleared. The 2026-08-06 tranche released roughly 911.5M shares and the stock closed up 6.1% that day (CNBC, 2026-08-06). The ~2026-09-09 release is listed at up to 319.0M, roughly a third of that.
• The build-out is dated and physical. The Louisiana Starbase commits $100B across 125,000 acres in Vermilion Parish with construction from 2027; the Bastrop, Texas blades-and-vanes foundry was confirmed 2026-08-30 on roughly 830 acres, with Musk saying in-house casting could pull forward natural-gas turbine availability by up to 18 months.
• Flight 14 is filed, not aspirational. SpaceX's FCC request targets 2026-09-15 for the first orbital-class attempt with roughly 20 operational Starlink V3 satellites; Booster 21 completed a 33-engine static fire on 2026-08-28 and Ship 41 cleared a six-engine test the week prior.

Bear Case

• The headline stack did not move the ceiling. $146.15 has stood since 2026-08-12 through a $100B spaceport announcement (late August), a turbine foundry (2026-08-30), and two target actions (2026-09-02).
• The compute contracts are cancellable. All three carry 90-day termination rights typically effective at the start of 2027 (2026-06-27 reporting), so the roughly $76B in aggregate headline value is not booked backlog.
• Execution cracks are now on the record. The data-centre leadership reshuffle followed reliability problems in Tennessee and Mississippi, reported 2026-09-01 and 2026-09-02, three weeks ahead of the 2026-09-30 Google capacity date.
• The supply calendar is front-loaded into this quarter. Trackers put roughly 700M shares across September (~2026-09-09 and ~2026-09-24), then about 28% of the 180-day block — on the order of 1.3 billion shares — around Q3 results near 2026-11-04, then the 2026-12-08 backstop for the balance of the ~4.6 billion-share block.
• The AI segment loses money at scale. Q2 showed a $1.257B segment operating loss against $2.56B of segment revenue, with Cursor consolidating only from the 2026-08-14 close.
• Valuation dispersion is unresolved. Published views run from Morningstar's roughly $62 fair value and a $75 street low to a $800 high, with the aggregate average near $220 across 41 analysts as read 2026-09-01. ARK called the Louisiana spaceport massively overbuilt on 2026-09-01.
• Attention is broad and retrospective. SPCX appeared among Benzinga Pro's twelve most-searched tickers for August (published 2026-09-02), alongside Nebius, and the 2026-09-01 coverage crop was headlined on the stock's 32% August gain.

Setup & Price Structure

The range is three weeks old and roughly 9% wide: $134.00 (close, 2026-08-20) to $146.15 (close, 2026-08-12), with 2026-09-02's $140.71 sitting near the middle. The closing sequence since the low is $136.97, $135.00, $137.95, $140.87, $141.50 on 2026-08-28, $143.69, $142.23, $140.71 — a run of higher lows into 2026-08-31 followed by two down closes that stalled the approach around $143–144.

RSI(14) at 49.2 is mid-range and carries no directional information here; the more useful reading is that it rose from 45.1 while price fell, which describes a market that is neither distributing hard nor bidding.

The $135 IPO price is the structural line. It has been defended on every close since 2026-08-21, including on tranche days, which is the single piece of evidence that supply is being met rather than merely paused. A daily close above $146.15 — particularly on or immediately after a release date — would date a bid that widened with the float. Absent that, the tape is a supply-capped range with two more releases scheduled inside it.

Crowding observables, stated as observables: the name ranked inside a retail-platform top-twelve search list for August; two sell-side actions arrived on the same day after a 32% monthly gain; roughly 219.3M shares with days-to-cover near one at the last read, meaning the short leg can be retired inside a session and is not a persistent source of demand and free float was 11.8% of 13.18 billion shares after the August tranches, with two more scheduled this month.

Catalyst Calendar (next 30 days)

• 2026-09-15 — Starship Flight 14, per SpaceX's FCC filing: first orbital-class trajectory, roughly 20 operational Starlink V3 satellites, no ship-tower catch. One tracker carries 2026-09-30 instead.
• ~2026-09-24 (est.) — Fourth 7% milestone tranche.
• 2026-09-30 — Reported capacity deadline tied to Google's $920M/month compute contract, whose payments are scheduled to begin in October 2026.
• ~2026-10-09 (est.) — Fifth 7% tranche, immediately beyond the 30-day window, with a further release near ~2026-10-24.

Elapsed catalysts

• ~2026-09-09 (est.) — Third 7% milestone lock-up tranche, up to 319.0M shares per third-party trackers. (passed 2d ago)

What Would Change Our Mind

The whole structure rests on the $135 IPO-price shelf holding through a supply calendar that has two more releases in it this month. Losing that shelf is the break: a weekly close below $133 takes out the $134.00 low close of 2026-08-20 and reduces the defence of $135 since 2026-08-21 to an excursion rather than a floor.

The second condition is time-based and equally gradeable. If the ~2026-09-09 tranche, the 2026-09-15 Flight 14 window, the ~2026-09-24 tranche and the 2026-09-30 Google capacity date all pass with no close above $146.15, the range resolves as supply-capped and the maturing read hardens toward a saturated one — mainstream coverage and a retail search ranking already present, with no widening bid behind them.

On the fundamental side, a 90-day termination notice from any of Anthropic, Google or Reflection would remove the contracted-revenue leg that the $248–$280 targets are built on, since those rights become effective around the start of 2027. A disclosed slip past the 2026-09-30 Google capacity date, or a Q3 AI operating loss wider than Q2's $1.257B while nameplate compute rises, would do similar damage to the AI-infrastructure frame.

The upside confirmation is symmetric and specific: a daily close above $146.15 on or immediately after a tranche date would show the bid widening with the float, which is the one thing the August tape has not produced.

Correlation Notes

• Counterparty concentration, not index beta. The compute revenue line depends on three named payers — Anthropic, Google and Reflection — none of which trades as an offsetting hedge for the first two (Anthropic is private; Alphabet is the buyer, not a proxy).
• Nvidia supply linkage. Musk has stated SpaceX will use Nvidia chips exclusively for its data centres, so GPU allocation news is a direct input to the compute build rather than a sector read-across.
• Turbine and power complex. The 2026-08-30 Bastrop foundry puts SPCX in the same conversation as GE Vernova's turbine backlog; grid-constraint headlines that lift independent-power names now cut both ways here, since Musk's 2026-09-02 argument was that tight power is why customers came.
• Retail attention clusters with the neocloud group. SPCX and Nebius both appeared on the same August most-searched list published 2026-09-02, so flow into that basket tends to arrive and leave together.
• The lock-up schedule is idiosyncratic. September and December supply steps are company-specific and will not correlate with space, AI or power sector moves in either direction.