Orbital

When you need a proven, enterprise-ready platform today

• Orbital is pre-seed as of June 9, 2026 and does not publicly document common enterprise security/compliance artifacts.

When launch economics and service continuity are unacceptable risks

• The business case is described in press as not closing under current launch economics; mission failure risk and schedule uncertainty are structural.

When your workload is bandwidth-heavy or needs tight interaction with terrestrial systems

• Many inference systems rely on constant data exchange, monitoring, and rapid iteration; orbital links may impose cost/latency constraints.

When a terrestrial GPU cloud is sufficient

• If the objective is to minimize cost or accelerate provisioning, established GPU cloud providers are likely to outperform on time-to-value.

When you cannot tolerate vendor/market uncertainty

• The orbital data center category is nascent; vendor survivability, standards, and interoperability are not mature.

Vendor-acknowledged limitations

• Public press coverage indicates the company acknowledges that current launch economics are a major blocker and that near-term feasibility depends on lower-cost launch.

Roadmap signals (publicly discussed)

• A 2027 prototype/test mission is referenced in coverage as a validation step (timing and launch partners are subject to change).

Scale limitations (what breaks first?)

• Based on independent technical analysis of orbital data centers generally (not Orbital-specific):
  • Communications throughput/cost
  • Thermal/radiator sizing vs. mass budget
  • Power generation and storage across orbital conditions
  • Reliability/replacement cadence and debris risk

Security/compliance gaps

• No publicly documented security certifications, audit reports, or enterprise identity controls were found for Orbital (orbital.inc).

Potential differentiators (mostly vendor/press-stated; not yet validated by customer evidence)

• Use of the space environment for:
  • Solar power availability
  • Radiative heat rejection
• Targeting AI inference as a plausible initial workload class rather than training.

Differentiation that is not yet proven

• Claims of lower effective cost per inference compared with Earth-based options have not been demonstrated publicly.
• Framing of effectively unlimited power should be treated cautiously: power generation, storage (eclipse), thermal constraints, and spacecraft mass/power tradeoffs are central engineering limiters per technical analyses.

Areas where competitors may have advantages

• Starcloud appears to have greater public visibility and press narratives about in-orbit GPU efforts.
• Marketplace/integrator models (ODC.space) may reduce single-vendor risk for early adopters.

• AI application providers needing sustained inference capacity
• Satellite/space companies that would benefit from processing data closer to collection points
• Government/defense users exploring resilient/off-grid compute