記事一覧に戻る
SpaceX IPO (SPCX) as an AI Stock: $1.75T Listing, Orbital AI Data Centers, and the Anthropic Compute Deal

SpaceX IPO (SPCX) as an AI Stock: $1.75T Listing, Orbital AI Data Centers, and the Anthropic Compute Deal

ZenChAIne·
SpaceXAI InfrastructureIPO

Introduction

SpaceX's 2026 Nasdaq IPO under the ticker SPCX should be read as an AI stock, not simply a space stock. Two facts make that framing necessary. First, the company filed in January 2026 to deploy up to one million satellites carrying around 100 gigawatts of AI compute in orbit. Second, Anthropic just contracted to draw more than 300 megawatts of GPU capacity from SpaceX's Colossus 1 data center.

Press reports point to a target valuation of about $1.75 trillion and a raise of up to $75 billion, which would make SPCX one of the largest IPOs ever proposed. That price tag is no longer explainable by satellite internet revenue alone.

Key Takeaways

  • SpaceX filed its S-1 with the SEC on May 20, 2026 (US Eastern Time), with press reports targeting a Nasdaq listing on or around June 12
  • Reported target valuation is about $1.75T with an offering of up to $75B, which would rank among the largest IPOs ever
  • A January FCC filing proposes up to 1M satellites and 100GW of orbital AI compute (full deployment targeted as early as 2028)
  • A February 2026 merger with xAI brought the Colossus data center (220k+ NVIDIA GPUs) under SpaceX; the Anthropic deal uses Colossus 1 capacity
  • Anthropic secured 300+ MW from Colossus 1 and signaled interest in multi-gigawatt orbital compute
  • Japanese retail investors can likely trade SPCX post-listing through brokers such as SBI / Rakuten / Monex if they cover the ticker; IPO lottery access from Japan is still uncertain

How Large Is the SpaceX IPO?

SpaceX's IPO is on track to be one of the largest equity offerings ever recorded if pricing matches press reports. The company filed its S-1 with the SEC on May 20, 2026 (US Eastern Time). Press reporting points to a roadshow week starting around June 4, pricing as early as June 11, and a Nasdaq listing under the ticker SPCX as early as June 12.

Reported target valuation is roughly $1.75 trillion, with an offering of up to $75 billion. The S-1 itself leaves the official price range and share count blank; those numbers are press estimates rather than confirmed terms.

At $1.75T, SPCX would sit in the multi-trillion-dollar tech tier alongside Apple, Microsoft, NVIDIA, Alphabet, Amazon, and Meta. From day one, it would trade as a mega-cap tech name, not as a niche aerospace stock.

Key financials disclosed in the S-1 include:

Metric2025 valueNote
Revenue$18.7B+33% YoY
Net income-$4.9Bvs +$791M in 2024
Connectivity segment revenue$11.39BLargely Starlink. +49.8% YoY
Starlink subscribers8.9M4.4M (2024) → 8.9M (2025)
Bitcoin held18,712 BTCFair value $1.64B as of 2025-12-31

The notable pattern is that revenue grew 33% while losses expanded. Because the xAI merger closed on February 2, 2026, full-year 2025 results predate the merger, so the loss is driven mainly by upfront spending on Starship development, Starlink V3 satellite production, and orbital data center prototyping. From 2026 onward, the xAI merger consolidates the AI segment (including Colossus) into SpaceX, adding a loss-making compute operation as an additional financial factor.

SpaceX moved from a $791M profit in 2024 to a $4.9B loss in 2025. That can be read as growth-stage reinvestment, but it also creates post-IPO volatility risk.

Why Should SPCX Be Read as an AI Stock?

The core reason to read SPCX as an AI stock is the orbital AI data center proposal SpaceX filed with the FCC on January 30, 2026.

The filing requests authorization for up to one million satellites in low Earth orbit. Launching roughly one million tonnes of satellites per year would carry approximately 100 gigawatts of AI compute, a figure equivalent to about 20 percent of current US electricity consumption, dedicated entirely to AI inference and training. The 100 GW figure is a long-term concept, not deployed capacity.

The technical architecture has several distinct features:

  • Orbit: 500 to 2,000 km, in 30-degree and sun-synchronous inclinations, maximizing solar exposure
  • Connectivity: Optical inter-satellite links, high-bandwidth optical links into Starlink, laser mesh to ground stations
  • Power: Continuous orbital solar generation, freeing the system from grid and cooling constraints on Earth
  • Timing: Per the S-1, full deployment of orbital AI compute satellites is targeted as early as 2028, with Starlink V3 itself expected on Starship around 2027

In February 2026, SpaceX merged with xAI. That deal brought the Colossus data center, with more than 220,000 NVIDIA GPUs, and xAI's training and inference stack inside SpaceX.

The result is that SpaceX now operates large-scale terrestrial AI compute (Colossus), holds a 100 GW orbital AI compute proposal, and connects both through Starlink and its own Starship launch vehicle. That is closer to a vertically integrated AI infrastructure company than a traditional space company.

What Does the Anthropic Compute Deal Signal?

The Anthropic compute deal is the most concrete revenue signal that SpaceX's AI repositioning is real (the contract amount is not publicly disclosed, though SpaceX's S-1 references a May 3, 2026 Anthropic Cloud Services Agreement with monthly fees through May 2029).

According to Anthropic's announcement, the company secured more than 300 megawatts of new capacity, equivalent to more than 220,000 NVIDIA GPUs, from SpaceX's Colossus 1 data center, with the capacity expected to come online within one month of the announcement. That capacity directly supports doubled five-hour rate limits and the removal of peak-hour restrictions for Claude Code Pro and Max users.

Within Anthropic's compute portfolio, SpaceX is a new entrant, but the scale matters and the framing reaches further into orbital compute.

PartnerScaleNature
Amazon (AWS Trainium)Up to 5 GW (~1 GW by end of 2026)Cloud GPU and custom AI silicon
Google / Broadcom5 GW per Anthropic's announcement (Broadcom 8-K confirms ~3.5 GW starting 2027)TPU and dedicated AI silicon
Microsoft / NVIDIA~$30B in Azure capacityCloud GPU
Fluidstack~$50B infrastructure investmentGPU cloud
SpaceX300+ MW from Colossus 1, multi-GW orbital as future intentTerrestrial Colossus + orbital AI compute

The notable line in Anthropic's announcement is its interest in "partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity." That hints at Anthropic potentially becoming an anchor customer for SpaceX's future orbital data center (this is an inference; no anchor-customer arrangement has been formally announced).

SpaceX is positioning itself on two AI axes simultaneously. On the ground, it absorbs xAI's Colossus and starts selling capacity to firms like Anthropic. In orbit, it pursues a 100 GW data center proposal targeted for as early as 2028. SPCX's reported valuation appears to price both axes.

Is a Tesla Merger Coming?

Short answer: not currently planned. In January 2026, Bloomberg and TechCrunch reported that SpaceX was considering a merger with either Tesla or xAI. On February 2, 2026, SpaceX completed an all-stock acquisition of xAI only, leaving Tesla outside the combined entity.

The combined entity at close looked like this:

ItemValue
xAI standalone valuation (at acquisition)~$250B
SpaceX + xAI combined valuation (Feb 2026)~$1.25T
Merger structureAll-stock

The main reason Tesla was excluded is the asymmetry between Tesla (a public company) and SpaceX (private). Public-to-private mergers face fiduciary duty issues, shareholder votes, and appraisal rights from dissenting holders, making a three-way merger impractical.

That said, once SpaceX becomes public via the IPO, a future Tesla–SpaceX combination as a listed-to-listed merger is still on the table. The Motley Fool predicted in March 2026 that Musk will merge Tesla with SpaceX within five years.

Current market-cap comparison (May 2026):

TickerMarket capNote
Tesla (TSLA)~$1.5-1.57T8th largest company in the world
SpaceX (SPCX)Reported target ~$1.75TIf IPO prices match, SPCX would overtake Tesla on day one

The picture is no near-term merger, but if SPCX prices at the reported level, SpaceX would top the Musk empire by market cap from day one. For Tesla shareholders, the strategic question becomes how SpaceX's AI infrastructure expansion (the Anthropic deal, orbital ambitions) interacts with Tesla's own AI roadmap (Optimus, full self-driving, ties to xAI/Grok).

How Can Japanese Investors Buy SPCX?

For Japanese retail investors, the realistic path to SPCX exposure is post-listing US equities trading. IPO lottery access from Japan is still uncertain.

The current options break down as follows:

TimingPathFeasibility
Pre-IPO (until June 2026)Indirect via related ETFs such as Baron First Principles ETF "RONB"High
IPO day (around June 12, 2026)Whether Japan-resident investors can access US retail allocation depends on each broker's announcementUncertain
Post-listingIf brokers cover SPCX, standard US equity trading at SBI, Rakuten, Monex, and othersHigh

Whether Japan-based brokers will participate in the IPO lottery is still pending. Reuters reporting indicates SPCX plans an unusually large retail allocation of up to 30 percent, roughly three times the norm, which could expand international retail access.

Practical considerations:

  • Opening a US equities account and activating US trading can take from a few days to several weeks depending on KYC and broker review, so investors targeting the June listing should start the application early
  • Initial trading days carry high volatility; investors should watch volume and revised guidance rather than chase the first print
  • For an AI-stock thesis, the key drivers post-listing will be Colossus utilization, additional non-hyperscaler compute contracts, and progress on orbital data center milestones

FAQ

Q. Is SpaceX a space stock or an AI stock?

A. It is both. The Connectivity segment (largely Starlink) remains the largest revenue line, but the xAI merger, Anthropic compute deal, and FCC orbital data center proposal make it impossible to ignore the AI infrastructure layer.

Q. Can investors in Japan join the SPCX IPO lottery?

A. It is not yet confirmed. Whether SBI Securities, Rakuten Securities, or Monex Securities will handle the US IPO depends on each broker's own announcement. Post-listing trading is straightforward through standard US equity accounts if each broker covers SPCX.

Q. Is a company with a $4.9B 2025 loss still investable?

A. The loss reflects heavy investment in Starship, Starlink V3, and orbital data center prototyping. Whether that turns into future cash flow depends on growth in Starlink, terrestrial AI compute sales, and orbital data center milestones.

Q. How big is the Anthropic contract relative to SpaceX as a whole?

A. The 300+ MW and 220k+ GPU capacity is meaningful for AI infrastructure, but probably a small share of SpaceX's $18.7B total revenue. The more important signal is that a non-hyperscaler AI company chose SpaceX for compute.

Q. When will orbital AI data centers be operational?

A. Per the S-1, full deployment of orbital AI compute satellites is targeted as early as 2028, with Starlink V3 itself expected on Starship around 2027. Investors should track these milestones and non-hyperscaler compute contract growth rather than expecting near-term revenue impact.

Summary

SpaceX's IPO is, if pricing matches reports, one of the largest equity offerings on record and a tectonic moment for AI compute markets.

The S-1 shows Starlink (within the Connectivity segment) scaling rapidly, the xAI merger brings Colossus inside SpaceX, the Anthropic deal supplies 300+ MW of GPU capacity, and the FCC filing reaches for 100 GW of orbital AI compute. Together they make SPCX impossible to value as a pure rocket or satellite company.

For Japanese investors, IPO lottery participation is still uncertain, but post-listing US equity trading and related ETFs offer realistic exposure. At ZenChAIne, we will continue tracking how space infrastructure and AI infrastructure converge, especially through the lens of how AI labs such as Anthropic source compute.

References