SpaceX is pursuing full reusability with Starship — tower-catching both stages and targeting in-orbit propellant transfer in 2026 — but the stainless second stage has yet to demonstrate routine reentry-and-reflight at low refurbishment cost. The cadence and depot architecture, not a single landing, are the real test.

Stoke Space's Nova rings its second-stage engine with a regeneratively cooled metallic heatshield, eliminating fragile thermal tiles entirely; the company finished first-stage proto-qualification in June 2026 and targets a late-2026 debut. The bet is that an actively cooled shield is more durable and turnaround-friendly than ablatives or ceramics.

Rocket Lab's Neutron and Relativity's Terran R recover the first stage while expending the upper stage, accepting a higher floor on cost in exchange for near-term medium-lift capacity. This captures most of the savings while sidestepping the unsolved reusable-upper-stage problem — a pragmatic hedge if full reuse proves harder than hoped.

Starfish Space is developing its Otter servicing vehicle to rendezvous with and dock to satellites that lack any docking aid, using electrostatic capture and autonomy; NASA has contracted it for servicing and de-orbit demonstrations. Docking with a non-cooperative target is the central unsolved capability that gates the whole servicing market.

SpaceX's planned 2026 ship-to-ship transfer between two Starships is the largest-scale attempt at moving cryogens in orbit, the enabling step for refuel-and-go architectures beyond LEO. Managing boil-off and ullage in microgravity at scale remains unproven outside small NASA experiments.

Impulse Space (Mira and the Helios kick stage) and Loft Orbital's hosted-payload buses move satellites to precise orbits and extend mission flexibility without full servicing. These tugs solve the placement-and-relocation slice of the problem today, building the rendezvous and maneuver heritage that fuller servicing depends on.

LeoLabs operates a global phased-array radar network for low-latency LEO tracking, while Slingshot Aerospace and Kayhan Space build the software for conjunction assessment and autonomous collision avoidance. Better catalogs and prediction reduce needless maneuvers, but tracking small fragments and fusing commercial and government data remains incomplete.

Astroscale is flying rendezvous-and-capture missions (its ADRAS-J inspected a real rocket upper stage; ADRAS-J2 with a robotic arm targets removal in FY2027) to prove that a non-cooperative derelict can be approached and de-orbited. The economics — who pays to clean up someone else's junk — are as unsolved as the engineering.

True Anomaly and Turion Space build maneuverable vehicles that inspect and characterize objects in orbit, blending debris awareness with national-security space-domain awareness. This dual-use approach attracts the defense funding that pure debris cleanup struggles to find, accelerating the underlying rendezvous and tracking technology.

Varda Space flies autonomous W-series capsules that process small-molecule drugs in microgravity and reenter for terrestrial recovery; it reentered W-5 and W-6 in early 2026 and is partnered with United Therapeutics on a pulmonary-disease formulation. The open question is moving from successful reentries to repeatable, certifiable pharmaceutical yield at scale.

Vast (Haven-1), Axiom, and Voyager's Starlab are building stations explicitly marketed as microgravity research and manufacturing platforms, betting that a crewed, serviceable facility beats one-shot capsules for iterative R&D. The risk is that demand for orbital lab time outside subsidized government seats stays thin.

Sierra Space's inflatable habitats and the Dream Chaser spaceplane, alongside Varda's in-house C-PICA heatshields, target the logistics layer — getting payloads up and products down affordably. Cheap, reusable return capacity is the hidden gate; without it, even high-yield processing cannot close the business case.

AST SpaceMobile flies the largest commercial communications arrays in LEO (its BlueBird arrays span thousands of square feet) to close a broadband link to standard handsets; it holds FCC commercial authorization for 248 satellites. The bet is that brute-force aperture, not handset changes, is the way to broadband — at the cost of huge, expensive spacecraft.

Lynk Global fields smaller satellites delivering text and narrowband connectivity to unmodified phones, with regulatory approvals across multiple countries. Starting with low-bandwidth services sidesteps the hardest part of the link budget and builds carrier relationships, but risks being leapfrogged by broadband-capable rivals.

The FCC's 2026 Supplemental Coverage from Space framework requires satellite operators to coordinate with terrestrial licensees (AT&T, Verizon, FirstNet for AST), and a turf war over interference is already underway. The regulatory link budget — who may transmit where without harming terrestrial networks — is as binding as the physical one.

Intuitive Machines, Firefly Aerospace (Blue Ghost), Astrobotic (Griffin), and ispace are flying NASA-backed landers to build reliable surface access — Firefly's Blue Ghost achieved a clean upright landing in 2025 while several peers tipped or crashed. Repeatable soft landing is the gate before any resource work can begin.

Interlune is prototyping a regolith excavator to harvest lunar helium-3, with a 2027 prospecting mission and 2029 pilot plant, plus offtake commitments from the DOE and a quantum-computing buyer. It is the most concrete commercial ISRU plan, but extraction has never been demonstrated on the lunar surface.

Lunar power networks, communications relays, and rovers (Astrobotic's CubeRover and FLIP rover deliveries, Firefly's Elytra orbital relay) are being built to support operations through the lunar night and across the surface. Power and thermal survival through 14-day darkness remains a fundamental unsolved constraint on any permanent presence.

K2 Space's Mega-class satellite delivers roughly 10x the power of comparable buses (its GRAVITAS demo flew in 2026), giving onboard processors the energy budget that capable compute demands. Abundant power is a precondition for running modern, hungry electronics in orbit even before the radiation problem is fully solved.

EO and SAR operators like Planet Labs, ICEYE, Umbra, and Pixxel push toward more onboard processing to cut downlink and deliver faster analytics, leaning on redundancy, watchdogs, and software error-correction to fly nearer-to-commercial parts. The trade-off between performance and upset tolerance is managed, not eliminated.

Loft Orbital and Apex Space sell configurable, productized satellite buses that let payload developers add processing without redesigning the spacecraft, spreading the cost of qualifying resilient avionics across many customers. Standardization lowers the barrier to flying better compute, but the underlying rad-hardness ceiling still applies.