SPCX · SpaceX (Space Exploration Technologies Corp.) · Stock research

Prices below are public closes from the split/dividend-adjusted daily series; the reference close is 2026-08-24 at $135.00.

SPCX — Space Exploration Technologies Corp.

Current Thesis

Two things moved since the last update, and they pull in opposite directions.

The orbital-compute line acquired a named architecture and a date. NVIDIA's newsroom release of 2026-08-24 states that SpaceXAI is adopting the Vera CPU for agentic AI and that the first-generation Starmind satellite will be built on an optimized NVIDIA Vera Rubin NVL72 rack-scale system. Musk's same-day X post dates a space-optimized Vera Rubin NVL72 for launch to orbit in Q4 of next year — Q4 2027 — with "significant scale" in 2028. Until 2026-08-24 the orbital data-centre revenue line in the high-end sell-side models rested on architecture with no vendor commitment and no flight date attached. It now has both, and neither produces revenue inside any 2026 or 2027 quarter.

The nearer test got smaller. Flight 14 was previously framed around the first attempt to catch the Starship upper stage with the tower. That attempt is deferred: Musk said on 2026-08-20 that SpaceX would "probably catch the ship with the tower in a few months," with FAA landing-zone approval described in third-party coverage as the limiter rather than vehicle readiness. What remains is still material — Flight 14 targets the vehicle's first orbital-class trajectory and deploys roughly 20 Starlink V3 satellites — but it no longer resolves ship reuse, the assumption underneath Bernstein's $248 target. An FCC filing shows the window opening 2026-08-28; third-party tracking now points to an early-to-mid September attempt.

Price reflects the standoff. The 2026-08-24 close of $135.00 sits exactly at the June IPO price. RSI(14) reads 55.1, down from 60.8 three sessions earlier. The 2026-08-21 reclaim held but did not extend.

The narrative is maturing, close to the edge. The story still works measured against the 2026-08-05 area low near $105, and the 2026-08-24 NVIDIA release is genuinely fresh news. But the coverage mix has turned late: retail roundup formats on 2026-08-21 ("5 Stocks Investors Couldn't Stop Buzzing About This Week," alongside MRNA and MSTR), a 2026-08-23/24 commentary cycle using SPCX's IPO as the cautionary comparison for Anthropic's $2 trillion valuation, and a 2026-08-25 headline on a disclosure that Trump purchased SpaceX shares 11 days after the IPO. Celebrity-ownership and warning-example coverage is not new marginal demand. Eight sessions have passed with no close above the 2026-08-12 print of $146.15. A second lower high into the ~2026-09-09 tranche would date the flip to saturated.

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

The model's bull view on SpaceX (Space Exploration Technologies Corp.) (SPCX), in brief: NVIDIA newsroom, 2026-08-24: SpaceXAI adopts the Vera CPU — 88 NVIDIA-designed Olympus cores, up to 1.2 TB/s of bandwidth, and up to 1.8x faster task completion versus x86 CPUs per NVIDIA's own claims — and the first-generation Starmind AI satellite will be based on an optimized… The bear view: No close above $146.15 since 2026-08-12. Both cases follow in full.

Bull Case

• NVIDIA newsroom, 2026-08-24: SpaceXAI adopts the Vera CPU — 88 NVIDIA-designed Olympus cores, up to 1.2 TB/s of bandwidth, and up to 1.8x faster task completion versus x86 CPUs per NVIDIA's own claims — and the first-generation Starmind AI satellite will be based on an optimized Vera Rubin NVL72. StorageReview's 2026-08-24 write-up describes NVL72-equivalent hardware per satellite: 36 Vera CPUs and 72 Rubin GPUs on an NVLink 6 fabric at 260 TB/s all-to-all. The per-satellite configuration is third-party reporting; the vendor release itself does not quantify it.
• Q2 2026 print, 2026-08-04: group revenue $7.81B (+92% YoY), adjusted EBITDA $3.54B (+191%), AI segment revenue $2.56B (+247%), Connectivity revenue $4.3B on 1.7 million net subscriber adds. Capex ran $18.37B in the quarter.
• Both completed lock-up tranches were absorbed within one session. The 2026-08-06 release of roughly 911.5 million shares — which lifted free float from 4.9% to 11.8% of 13.18 billion shares outstanding — closed up 6.1%. The 2026-08-20 release of 319.0 million shares (about 2.44% of the basic share count) closed at $134.00, then $136.97 on 2026-08-21 with the session low of $131.22 bought back to a $137.35 high.
• NSPM-17, signed 2026-08-20 — the first National Space Transportation Policy update since 2013 — sets a goal of over 1,000 US launches and reentries per year by 2030 against just under 200 US-vehicle launches in 2025 (SpaceNews, 2026-08-20), and directs agencies to expedite permitting, publish range schedules and co-develop launch infrastructure with private partners.
• Flight 14 carries operational payload for the first time. Roughly 20 Starlink V3 satellites — flat spacecraft with laser inter-satellite links intended to replace the current constellation — deployed from an orbital-class trajectory would connect Starship cadence directly to the Connectivity segment that generated $4.3B in Q2.
• Grok distribution widened across three sessions: Amazon Bedrock on 2026-08-19, Google Cloud Vertex AI on 2026-08-21, with bot availability extended to SuperGrok Plus, Cursor Pro+ and Cursor Teams the same day.

Bear Case

• No close above $146.15 since 2026-08-12. The sequence runs $141.29 (2026-08-13), $140.00 (2026-08-14), $139.65 (2026-08-19), $134.00 (2026-08-20), $136.97 (2026-08-21), $135.00 (2026-08-24). The 2026-08-24 NVIDIA announcement did not produce an up close.
• The dated ship-reuse test moved out. With the catch deferred to "a few months" (Musk, 2026-08-20), no scheduled event before Q4 validates the reuse assumption in Bernstein's model of roughly 3,500 Starship launches supporting about $600B of 2031 revenue.
• Starmind revenue is dated Q4 2027 at the earliest, per Musk's 2026-08-24 post. The AI segment posted a $1.257B operating loss in Q2 2026 against $18.37B of group capex in the same quarter.
• The supply calendar is not exhausted. Trackers put the third milestone tranche of roughly 319 million shares at ~2026-09-09 or ~2026-09-10, a fourth at ~2026-09-24, and further releases at ~2026-10-09 and ~2026-10-24. The Q3 print (~2026-11-04, est.) is modelled to release roughly 28% of the 180-day block — on the order of 1.3 billion shares — and the 2026-12-08 backstop covers the balance of the roughly 4.6 billion-share block, including 455.8 million conditional shares that failed their $175.50 August trigger.
• Published valuations do not agree on the order of magnitude. Raymond James $800 and Morgan Stanley $300 at the top; Bernstein $248 (raised from $239 on 2026-08-18); BofA $235; Argus $160; a $75 low in a 16-analyst poll; Morningstar fair value near $62. Tracker averages read $220.20 (MarketBeat), $232.35 (TipRanks) and $213.50 (16-analyst poll), retrieved 2026-08-24, on a consensus rating of Buy with 28 buy versus 2 sell.
• The Anysphere/Cursor $60B all-stock acquisition, expected to close in Q3 2026, sets its exchange ratio on a seven-day average share price at closing. Shares issued scale inversely with the quote, so a soft tape into the closing window mechanically increases the dilution.
• Texas froze broadband grants on 2026-08-13 pending a probe into alleged Starlink favouritism — a state-level challenge to the subsidy channel feeding the Connectivity segment.

Setup & Price Structure

The 2026-08-24 close of $135.00 lands on the June IPO price, which has flipped roles twice in four sessions: lost on 2026-08-20 at $134.00, reclaimed on 2026-08-21 at $136.97, and now sat on. The 52-week range spans $104.83 to $225.64, so price is nearer the low of the post-IPO span than the high. The level that matters more than the range is $146.15, which has capped every close since 2026-08-12.

Support structure below: the 2026-08-20 session low of $131.22, then the 2026-08-07 to 2026-08-11 consolidation shelf, then the 2026-08-05 area low near $105. RSI(14) at 55.1 is mid-range and carries no directional information on its own.

Crowding and positioning observables, stated without a verdict attached:

• Free float is 11.8% of 13.18 billion shares outstanding after the 2026-08-06 unlock, plus the 319.0 million shares released 2026-08-20.
• Musk's roughly 6.4 billion shares are locked for 366 days with no early-release provision, so the largest single holding is absent from the entire 2026 supply calendar.
• No Form 4 or Form 144 filings disclosing pre-IPO block sales had appeared through 2026-08-24. Insider distribution reports on a lag, so silence is not evidence of retention.
• Retail-sentiment coverage clustered on 2026-08-21 across two separate roundup formats, and by 2026-08-23/24 SPCX was being cited as the warning example in Anthropic valuation commentary.
• Consensus is one-sided (28 buy, 2 sell) while the published target range spans more than 12x, from Morningstar's ~$62 fair value to Raymond James' $800.

Catalyst Calendar (next 30 days)

• 2026-08-28 — Starship Flight 14 launch window opens per SpaceX's FCC filing. Third-party tracking points to early-to-mid September for the actual attempt. First orbital-class trajectory; roughly 20 Starlink V3 satellites deployed; Super Heavy splashes down in the Gulf without a tower catch, and no ship catch is attempted.
• ~2026-09-09 (est.) — Third milestone lock-up tranche, roughly 319 million shares. Trackers differ by a day between 2026-09-09 and 2026-09-10.
• ~2026-09-24 (est.) — Fourth milestone tranche in the two-to-four-week staggered cadence, with further releases at ~2026-10-09 and ~2026-10-24.
• ~2026-09-30 (est.) — Anysphere/Cursor $60B all-stock acquisition expected to close in Q3 2026; exchange ratio set on a seven-day average share price at closing.

What Would Change Our Mind

The structural break is the loss of the $135 IPO price as a floor rather than a ceiling. Concretely: a weekly close below $133 puts price back inside the 2026-08-07 to 2026-08-11 shelf and reduces the 2026-08-21 reclaim to a single-session excursion, at which point the descending sequence off $146.15 is the whole of the August tape. The secondary condition is the ~2026-09-09 tranche arriving and passing with another lower high and still no close above $146.15 — the same shape as 2026-08-20, twice, which would date the flip from maturing to saturated.

On the other side, the frame improves on a close above $146.15 that holds for more than one session, particularly if it comes after the ~2026-09-09 release rather than before it, since that would show tranche supply being taken at a higher low rather than merely not causing a break.

On fundamentals rather than tape: a Flight 14 that reaches orbital-class trajectory and deploys the Starlink V3 batch cleanly advances the Connectivity revenue line on a schedule; a Flight 14 that loses the vehicle before deployment removes the near-term cadence argument and pushes the next dated proof point into Q4. Separately, a target cut or downgrade from any firm currently above $200 — Bernstein, BofA, Morgan Stanley, Raymond James — would mark the first crack in a consensus that is 28 buy to 2 sell.

Correlation Notes

• NVIDIA is now an explicit counterparty, not an analogue. The 2026-08-24 Vera CPU adoption and the Starmind Vera Rubin NVL72 design tie SPCX headline flow to NVIDIA product cycles in both directions. Observed relationship, not a measured beta: the AI segment ($2.56B of Q2 2026 revenue) trades on the same capex narrative that prices the accelerator complex.
• AI credit conditions. SPCX has already taken a $25B investment-grade issue, and a 2026-08-24 feature framed AI-sector bond issuance at roughly $220 billion as competing with Treasury supply. Widening spreads in that complex would raise the cost of the gigawatt-scale build the AI segment is funding.
• Musk-complex headline coupling. Tesla has a Cybercab launch event scheduled 2026-09-03, and 2026-08-24 carried Starlink-over-Russia coverage after Zelensky's request. Neither changes SPCX cash flows directly; both compete for the same attention channel that produced the August retail-roundup clustering.
• Space peers. NSPM-17's 1,000-launches-by-2030 goal is sector-wide policy, so launch-services names move on the same document; SPCX-specific reaction has to be separated from the sector print to read as company news.

--

Archetype: Dominant Narrative. The name is not trading on a discrete binary or on a squeeze mechanic — it is trading on whether the largest IPO in history can keep converting engineering milestones into narrative faster than its own lock-up schedule releases shares.