If we explicitly model competing Wright’s-law curves for (a) terrestrial fabs/data centers and (b) orbital manufacturing and compute under cheap launch, in which concrete product categories (e.g., specialty fibers, radiation-hardened components, ultra-secure enclave compute, high-rad testing) does orbit ever reach durable cost crossover, and where do Earth cost curves outrun any microgravity or vantage-point advantage even on optimistic space assumptions?

Answer

Orbit reaches durable crossover only in a few thin, high-value niches; in most product categories terrestrial fabs and data centers stay cheaper once their own Wright’s-law curves are included, even with very low launch cost.

Likely durable orbital crossover niches (under optimistic but not utopian space assumptions)

• High-rad / extreme-environment testing as a service

  • Products: lifetime and soft-error testing for advanced nodes, avionics, nuclear/defense electronics.
  • Why orbit crosses: value is information, not mass; Earth analogs (accelerators, reactors) have high capex and limited realism. Launch-cost collapse plus reusable orbital testbeds can keep per-test cost falling with flight rate. Earth facilities have slower learning and siting constraints.

• Very high-value, low-throughput microgravity materials

  • Products: a few specialty optical fibers (e.g., ZBLAN-like where loss really matters), niche protein crystals, small-batch exotic alloys or composites where structure from microgravity is uniquely useful.
  • Why orbit can win: revenue/kg is enormous, process tools miniaturize, and orbital hardware can reuse common platforms and robotics. Terrestrial fabs also learn, but some processes may be impossible to replicate on Earth even with advanced control.

• Sovereign / enclave orbital compute and key management

  • Products: small secure enclaves for very sensitive workloads, key custody, or special-purpose crypto/AI models.
  • Why orbit may reach parity or modest advantage: customers pay a security/jurisdiction premium; required capacity is small; orbital platforms can amortize cost over many tenants. Earth data centers improve fast but cannot fully match some threat models (physical isolation, multi-jurisdiction control).

Borderline or non-crossing categories (Earth curves dominate)

• General-purpose cloud compute and bulk storage

  • Terrestrial hyperscale centers ride very steep learning curves and benefit from cheap power, cooling, and massive volume. Even with cheap launch and standardized platforms, orbital $/TFlop or $/TB is unlikely to beat Earth except for tiny, ultra-secure enclaves.

• Mainstream semiconductors and most radiation-hardened components

  • Modern fabs are extreme Wright’s-law machines with huge volume; radiation-hardening and reliability features are mostly design/process choices inside Earth fabs. Orbital fabs suffer small volumes and high fixed complexity, so per-die cost stays higher.

• Bulk microgravity manufacturing (commodity fibers, bulk pharma, mass-market alloys)

  • As Earth automation improves, most mass-market lines get cheaper faster than an orbital factory with limited volume. Launch cost still adds a hard floor, and many microgravity benefits can be engineered around with process tricks and advanced equipment on Earth.

• Large-scale high-rad computing farms

  • For most high-rad workloads (e.g., defense, nuclear), hardening terrestrial facilities and using specialized chips is cheaper than placing large compute masses in orbit and maintaining them.

Overall pattern

• Orbit can win where:
  • Value density (value per kg launched/returned) is extremely high.
  • Volume stays low enough that terrestrial megafabs never fully industrialize the niche.
  • The orbital advantage (true microgravity, natural radiation, vantage point, jurisdictional separation) cannot be convincingly mimicked on Earth.
• Earth wins where:
  • Learning rates are fast and volumes large (mainstream chips, general cloud, most materials).
  • Orbital advantages are incremental rather than categorical.

So under explicit, competing Wright’s-law curves, the stable orbital economy looks like a thin layer of specialized test, security, and microgravity products on top of an overwhelmingly Earth-based manufacturing and compute base, not a broad replacement for terrestrial fabs and data centers.