2026-07-11 Full-Cycle Decision

Rocket Lab remains the best execution-led space platform, but USD 46.9 billion market value and about 69 times corrected TTM sales require many successes at once. The Iridium transaction adds financing, approval, integration, and management-bandwidth risk. Peter Beck's trust sold 3.276 million shares under a pre-Iridium 10b5-1 plan; that is not a thesis break, but it is a large ownership-overhang signal. At 33.95% of the account, the existing position already captures upside.

Direct-to-device broadband still has extraordinary right-tail potential and BlueBird 8–10 launch evidence remains constructive. However, USD 21.3 billion market value is supported by less than USD 5 million TTM revenue; on-orbit performance, continuous coverage, MNO economics, regulatory approval, later satellite financing, and Starlink competition remain binary. An approximately 19.88% account weight is already sufficient.

Credo has excellent AI interconnect growth and margins, but about 36.0 times corrected TTM sales and roughly 90% revenue concentration in the top ten customers still leave little margin for error. The demanding CEO performance award aligns incentives better than a time-only grant but can still dilute shareholders. Quality is high; expected entry return is not.

Astera's PCIe, CXL, Ethernet, and rack-scale connectivity platform is strategically relevant, but about 70.7 times corrected TTM sales and hyperscaler/design-win concentration price in exceptional execution. A good company can still be a poor current entry.

Data-center power and liquid-cooling demand is real, orders are strong, and Vertiv has scaled execution. At roughly USD 122.5 billion market value and a high earnings multiple, the company is too mature and too fully priced for the satellite account's extreme-upside mission without a major dislocation.

Nebius has real GPU, power, and AI-factory exposure, and recent acquisition completion reduces one transaction uncertainty. About USD 53.3 billion market value on roughly USD 530 million TTM revenue, capex needs, financing, utilization, receivables, and integration risk make current equity returns too dependent on perfect scale-up.

Micron is the strongest direct HBM and advanced-memory bottleneck, but the roughly USD 1.1 trillion scale and unusually strong memory-pricing cycle constrain multi-decade asymmetric upside. Commodity pricing, capacity response, China exposure, and cyclicality can reverse earnings quickly.

CoreWeave has direct scarce-GPU capacity and revenue growth, but about USD 24.9 billion debt, expensive financing, customer concentration, GPU depreciation, and utilization risk can transfer enterprise growth away from common equity. July planned insider sales reinforce entry caution.

Circle is the clearest public stablecoin toll-point candidate, and final OCC approval is positive. Corrected gross price-to-sales is about 5.7, but price-to-revenue-less-distribution-costs is about 14.4; reserve income is rate-sensitive, distributor costs consume much of the economics, USDC circulation declined about 5% from quarter-end to July 9, identified dilution is meaningful, and ARC token value capture for common shareholders is unclear.

Lumentum has real AI photonics, laser, and optical-switch relevance, but revenue weakness, customer concentration, debt, convertible notes, and NVIDIA-linked preferred capital complicate the common-equity payoff. At about 37.9 times corrected TTM sales, the entry does not compensate.

Cerebras offers a differentiated wafer-scale compute system, but two related customers represented about 74% of Q1 revenue and FMR's 25.7% converted ownership underscores a concentrated capital structure. A short public record and roughly 70 times sales block buying.

Power sites and AI-cloud conversion offer real optionality, but bitcoin exposure, customer/utilization proof, construction financing, USD 3 billion convertible debt, and 18.2 million time-based joint-CEO RSUs create material capital and governance risk. The long locks do not remove dilution.

Advanced nuclear could become a durable power bottleneck, but Oklo is pre-commercial and depends on NRC licensing, fuel, project execution, customer contract quality, and financing. Roughly USD 8.5 billion market value plus a USD 1 billion ATM prices too much success before deployment.

Onsite power demand is real, but the July short-report dispute raises accounting, terminal-customer, backlog, and scandium-supply questions. Bloom's 8-K denies the claims without quantitative reconciliation; its Q1 10-Q independently confirms two counterparties at 50% and 12% of revenue. Roughly 28.4 times corrected TTM sales already failed the entry gate.

The DOE award confirms strategic HALEU scarcity, and ACO title to completed capacity strengthens shareholder value capture. The full firm-fixed-price contract, milestones, termination rights, cost sharing, appropriations, TENEX supply, expansion financing, and per-share economics remain decisive, so any current action must be minimal and short-dated.

Globalstar owns scarce spectrum and has a meaningful partner relationship, but approximately 38 times sales, partner concentration, network capex, and governance mean the network's value may not accrue proportionately to common shareholders. It is a weaker direct-to-device expression than the existing ASTS position.

Kratos has real defense-drone, propulsion, and space exposure, but project timing, contractor margins, authorized-share dilution capacity, and an extreme earnings multiple reduce the asymmetric setup. MDA and GILT now improve the same-lane opportunity-cost comparison.

IonQ has cash, enterprise relationships, and a plausible quantum platform, but roughly USD 16 billion market value, about 85.5 times corrected TTM sales, uncertain technical roadmaps, and SkyWater integration leave no valuation cushion. New IQMX, QBTS, QUBT, and HQ peers strengthen the conclusion that the whole public quantum basket is overheated.

Intuitive Machines offers relatively small-cap lunar and government mission optionality, but negative gross margin, mission failure risk, customer concentration, acquisitions, and ATM dilution can impair the equity before the upside arrives. The existing space concentration eliminates the need for another binary name.

Redwire improved liquidity by expanding its revolver and paying down term debt, but a USD 350 million ATM, preferred conversion, Edge Autonomy integration, internal-control risk, and weak cash generation remain. MDA is a cleaner operating-quality space-systems comparator.

Firefly's launch and spacecraft exposure is direct, but losses, cash runway, offerings, selling-holder pressure, mission risk, and a new USD 22 million stock-financed Space-ng acquisition with undisclosed dilution make survival and capital structure more important than the theme.

York has real defense-space revenue and backlog and more upside scale than MDA, but customer concentration, contract accounting, material weakness, acquisition integration, and 2.289 million newly issued All.Space shares keep the short public record too risky.

Voyager has defense-space systems and Starlab optionality, but negative gross margin, losses, convertible debt, controlled governance, and Starlab funding needs dominate. The expanded revolver increases potential liquidity, not cash or intrinsic value, and adds covenant/leverage relevance.

Xanadu is a real photonic quantum option, but revenue is tiny, customer concentration is high, corrected share count implies a demanding valuation, and Yorkville/S-8 dilution can arrive before commercial validation. IQMX, QBTS, QUBT, and HQ increase its opportunity cost.

Fabrinet has real AI optical manufacturing exposure and a comparatively reasonable roughly four-times-sales valuation, but contract manufacturing has lower margins, customer concentration, and limited pricing power. It is useful infrastructure, not a strong enough bottleneck owner for the satellite mission.

Eaton's electrical and liquid-cooling exposure is high quality, and the Mobility/Dana separation should sharpen focus and provide about USD 1.1 billion cash. At roughly USD 158 billion market value, acquisition and separation execution plus mature scale sharply limit hundredfold upside.

Quanta directly benefits from grid, transmission, substation, and large-load construction scarcity, but contractor economics, labor, project estimates, acquisitions, and roughly USD 99 billion scale limit the shareholder return path despite strong demand.

SpaceX has exceptional launch, Starlink, direct-to-device, and AI-platform assets, but a trillion-dollar-scale starting valuation, controlled governance, xAI/X losses, USD 25 billion notes, stock-funded transactions, and a very short public trading record conflict with the satellite account's need for early asymmetric upside.

MDA is the best new operating-quality space peer: Q1 revenue grew 32.2%, adjusted EBITDA was CAD 90.6 million, backlog CAD 3.69 billion, and net cash CAD 299.3 million. CLS and Blue Canyon broaden recurring analytics and U.S. defense exposure, but the 23 million-share offering, final capitalization, two integrations, and existing account space concentration keep entry trigger-only.

Gilat is the cheapest real new operating peer at about 1.96 times sales, with USD 451.7 million annual revenue, USD 20.7 million net income, and USD 169 million cash. Comtech and USD 11 million of new U.S. defense orders justify promotion, but hardware/contractor economics, transaction completion, integration, and existing space concentration prevent a current order.

IQM has delivered about 18 systems, recorded roughly EUR 31.3 million 2025 revenue and more than EUR 67 million backlog, making its commercial evidence stronger than most quantum peers. A roughly USD 2.2–2.4 billion pro-forma value, losses, cash burn, financing, and post-SPAC structure imply about 60 times sales.

Nano Nuclear has substantial liquidity and acquired nuclear-material logistics, but no material reactor revenue, a USD 300–350 million prototype estimate, licensing years, a USD 900 million shelf, and about 25% six-month share growth put dilution ahead of commercial proof. LEU is the superior direct fuel-bottleneck expression.

Horizon provides a real quantum software/compiler angle and has liquidity, but 2025 revenue was only about USD 40 thousand, losses are material, and resale/warrant overhang is large. There is no evidence that customers will pay at the scale implied by the public valuation.

Aurora resources and a debt-light balance sheet are tangible, but the company is pre-revenue, still drilling and permitting, targets a PFS only in late 2027, and carries layered warrant, preferred, resale, and S-8 dilution. Small market value alone is not a thesis.

D-Wave has real USD 33.4 million bookings, systems, and enterprise services, but Q1 revenue was only USD 2.9 million against about USD 7.37 billion market value. Even annualizing the quarter leaves an extreme multiple with continuing losses.

Quantum Computing has nearly USD 1.4 billion cash and investments and Q1 revenue improved to USD 3.7 million, but gross profit was negative and growth was acquisition-driven. The equity is primarily a capital-allocation bet before operating value creation is proved.

No policy-eligible directly listed U.S. common stock exists. Private shares, proxy securities, or complex vehicles are outside policy v1.1; this is objective instrument ineligibility, not a research gap.

No policy-eligible directly listed U.S. common stock exists. Private exposure would add valuation, liquidity, governance, and policy risks that the account does not permit.