ASTS, IRDM: SpaceX's 100GW Space Datacenter Goal Reprices

SpaceX's S-1 filing on May 20, 2026 made one commitment that received less attention than the $1.75 trillion valuation pitch but matters more for AI infrastructure positioning: "Our goal over time is to launch 100 gigawatts of compute to space each year." That's roughly one-fifth of US annual electricity production. Semianalysis followed up June 3 with a 10,000-word teardown of space datacenter economics — concluding that current Space-Earth cost parity requires "significant engineering work, material science breakthroughs and cost scaling progress" before space-based compute deploys at scale. The window Semianalysis flags for serious commercial deployment is the late 2030s, with cost parity potentially achievable earlier. For investors, the implication is not that space datacenters are imminent — it's that the publicly-listed companies in the satellite communications and launch-adjacent cohort will be repriced based on how investors handicap the timeline. ASTS and IRDM are the two cleanest equity exposures.

The actual challenges Semianalysis flagged

Four widely-repeated bull claims about space datacenters are misleading:

1. "Space provides free 24/7 solar energy." Partially true for Geosynchronous orbit, but Low Earth Orbit (LEO) still has eclipse periods. Net daily solar exposure differs materially between orbit choices.
2. "Cooling is free in space." False. Vacuum eliminates conduction and convection — leaving radiation cooling as the only option. Radiation cooling is dramatically less efficient than terrestrial water/air cooling. Heat dissipation, not power generation, is likely the binding constraint.
3. "Latency is low because it's just light through vacuum." True for LEO (typical 5-30ms one-way), but GEO latency averages 230ms one-way — completely unsuitable for AI inference workloads.
4. "No permitting needed in space." True today, but unlikely to remain true at scale. Spectrum allocation, debris management, and military airspace restrictions will all become regulatory constraints once orbital compute scales.

These constraints don't make space datacenters impossible. They mean the real winners over the next decade will be the companies that build the supporting infrastructure: launch capacity (SpaceX, RKLB), satellite communications backbone (ASTS, IRDM), and orbital servicing/maintenance (still mostly private).

Why ASTS and IRDM are the listed-equity proxies

For public-equity exposure to the space-economy thesis, and as part of the broader AI infrastructure cohort buildout, the cleanest paths are satellite communications players that can both serve as comms backbone for orbital compute and have already raised the capital to deploy meaningful constellations.

ASTS (AST SpaceMobile) is the cellular-to-satellite play. Their BlueBird constellation provides direct-to-mobile satellite connectivity — currently in early commercial deployment with major mobile network operators (T-Mobile, AT&T, Vodafone). At a market cap of $32.2 billion and forward P/S of 86.4×, the equity is priced as a pre-revenue growth optionality bet. FY25 revenue was $14.7 million; consensus forward revenue growth of +338.8% implies a ramp toward $63 million in the next reporting period.

IRDM (Iridium Communications) is the established satellite communications operator — owns the only commercial LEO constellation with global coverage and full crosslinks. FY25 revenue was $872 million, with a 50.4% trailing EBITDA margin and 34.8% trailing FCF margin. Forward P/S of 5.87× makes it the cohort's "value" pick. YTD return +185.4% reflects market enthusiasm that has already partially priced the AI-orbital story.

Data points

drillr terminal snapshot (June 3, 2026):

ASTS tape volatility tells the story: closed June 3 at $107.73 with a -8.83% drop on heavy volume, after May saw a wild trading range from $83.67 (May 15) to $133.09 (May 28). The 1-year return of +369.3% reflects the equity being priced as a binary outcome — either BlueBird succeeds commercially and ASTS becomes a $50-100 billion company, or the cash burn rate (EBITDA margin of -553%, FCF margin of -1527%) consumes equity value before commercial revenue materializes.

IRDM at $48.97 has the more stable thesis. The company has been cash-generating for years (FY25 FCF $300M), pays dividends, and operates the Iridium constellation that competitors have failed to replicate. The forward revenue growth of just +0.8% reflects mature business expectations — but any acceleration from new AI-orbital partnerships would re-rate the equity meaningfully.

{
  "hint": "A clean infographic showing a stylized Earth from space with two satellite constellation patterns overlaid. The first constellation, labeled 'ASTS BlueBird (cellular-to-satellite)', shows a small set of large satellites in low Earth orbit. The second, labeled 'IRDM Iridium NEXT (commercial LEO crosslinks)', shows a more comprehensive mesh covering the entire planet. A small inset shows SpaceX Starlink for comparison. Color palette: deep blue for the Earth and stars, soft orange for ASTS, soft green for IRDM, dim gray for Starlink reference. Editorial science publication aesthetic.",
  "aspect": "16:9",
  "style": "minimalist editorial infographic, clean science publication",
  "alt": "Satellite constellation patterns over Earth showing ASTS BlueBird cellular-to-satellite and IRDM Iridium NEXT commercial LEO for space datacenter communications backbone thesis",
  "caption": "ASTS and IRDM constellations — backbone communications for orbital compute"
}

Analysis: scenarios across the 2026-2035 window

The Semianalysis analysis points to late 2030s as the realistic timeline for commercial space datacenter deployment. The listed-equity implications break into three scenarios.

Scenario A — Pre-commercial space DC announcements through 2027. SpaceX deploys testbed orbital compute by 2027 (not the full 100GW commitment, just demonstration capacity). ASTS and IRDM both benefit as comms backbone candidates. ASTS could re-rate to $150-180 on partnership announcements; IRDM to $60-65 on incremental contract wins.

Scenario B — Commercial space DC moves toward 2030s as Semianalysis suggests. Both ASTS and IRDM remain pre-commercial-deployment for orbital compute. ASTS depends on BlueBird commercial revenue ramp (not space DC). IRDM stays as cash-generating value play.

Scenario C — Space DC concept stalls due to economics. TCO analysis remains negative through 2030+. Capital flows away from space-economy equities. ASTS faces material cash burn pressure; IRDM defends with cash generation.

The asymmetric trade: IRDM at 5.87× forward P/S with dividend protection is the downside-protected position. ASTS is the high-volatility upside option. A combined position weighted 60% IRDM / 40% ASTS captures the asymmetric profile.

What to watch

• SpaceX IPO pricing and follow-on satellite/orbital compute milestones: Real signals from SpaceX validating or moderating the 100GW commitment.
• ASTS BlueBird operator-revenue contracts: T-Mobile, AT&T, Vodafone direct-to-mobile satellite service revenue reports. Watch for Q3 2026 commercial billing numbers.
• IRDM new contract awards: Government and enterprise contracts, particularly any AI/data-related communications wins.
• NVDA / Hyperscaler orbital compute commitments: Watch for any hyperscaler-funded space-DC partnership announcements through 2027.
• Material science announcements (radiation-hardened GPUs, lightweight radiator panels): These are the binding constraints Semianalysis flags. Breakthrough announcements move the thesis timeline.

---