Orbital Computing // Space Infrastructure // AI
Starcloud 
Private U.S. developer of GPU-equipped orbital data centers and space-based cloud infrastructure
PRIVATE
No public ticker
Starcloud has no listed security. The latest verified valuation is a dated financing mark, not a live market value.
Starcloud announced a $250M Series A extension on 21 August 2026 at a $2.3B post-money valuation, taking company-stated total capital raised to $450M. That total should not be read as cash on hand. Public round-by-round disclosures reviewed by Defense Briefing do not fully reconcile to $450M, so no undisclosed amount has been reverse-engineered.
Starcloud publicly operates as Starcloud, Inc. and its website carries that name. However, the January 2025 SEC Form D and a 25 August 2026 FCC experimental filing identify the applicant as Lumen Orbit, Inc. Defense Briefing therefore treats Starcloud as the operating name and flags the current legal-name reconciliation as unresolved in public records rather than assuming a completed legal rename.
Starcloud designs satellites that carry data-center-class graphics processing units, storage, networking, solar-power hardware and thermal-control systems so computing can occur in orbit rather than sending every raw data set to Earth first. The near-term business case is edge processing, storage and cloud workloads for spacecraft operators and terrestrial users. The long-term company thesis is much larger: distributed orbital data centers that use high-availability solar power and radiators to dissipate waste heat.
The distinction matters. Starcloud has demonstrated a single H100-class GPU in orbit on Starcloud-1. Its first commercial platform, Starcloud-2, remains a planned mission. Its proposed 88,000-satellite orbital data-center system is an FCC application, not an authorization or deployed constellation.
Launched in November 2025, Starcloud-1 carried an NVIDIA H100 GPU into low Earth orbit. Starcloud says the spacecraft later ran a version of Google Gemini and trained the nanoGPT language model in orbit. The mission demonstrates that a terrestrial-class accelerator can operate in the orbital environment, but it is a technology demonstration rather than a commercial hyperscale data center.
Starcloud describes Starcloud-2 as its first commercial mission, with a GPU cluster, persistent storage, continuous customer access and proprietary power and thermal systems in a smallsat form factor. The company says it is intended to process space-generated data in orbit and provide storage and cloud services to terrestrial users. Starcloud currently says the spacecraft will be fully operational in sun-synchronous orbit by 2027.
Starcloud has publicly named Starcloud-3 but has not published a complete primary-source specification set on its product page. The 21 August 2026 financing release says Starcloud-3 development production lines are underway at a new 100,000-square-foot manufacturing facility in Woodinville, Washington. Claims about eventual mass production and heavy-lift deployment remain forward-looking until hardware and cadence are demonstrated.
Starcloud filed FCC application SAT-LOA-20260202-00073 for a non-geostationary system of up to 88,000 satellites. The application has drawn public comments over orbital safety, astronomy and environmental effects. The proposed system should be treated as a regulatory application and long-range architecture, not authorized capacity.
| Date | Counterparty / Body | Program / Scope | Value | Status |
|---|---|---|---|---|
| 26-MAY-2026 | SpaceX / Starlink | Contract for 50+ Starlink Mini Laser terminals across 25+ Starcloud satellites, with two optical terminals planned per spacecraft | Not publicly disclosed | Commercial contract signed; first hardware planned on orbit within one year of announcement |
| 22-OCT-2025 | Crusoe | Strategic partnership to deploy Crusoe Cloud on a Starcloud satellite and offer limited orbital GPU capacity | Not publicly disclosed | Commercial deployment planned; service not yet operational |
| 16-MAR-2026 | NVIDIA | Collaboration around NVIDIA accelerated computing for space, including the Space-1 Vera Rubin Module | Not publicly disclosed | Technical collaboration announced |
| 2025-2026 | IQT | IQT portfolio investment supporting orbital compute technology | Not publicly disclosed | Strategic investment; not a government procurement award |
Defense Briefing did not identify a public Department of Defense, NASA or Space Force prime-contract award to Starcloud in the primary sources reviewed through 2 September 2026. IQT backing is evidence of national-security interest but should not be described as a federal procurement contract.
Starcloud does not disclose revenue, margins, bookings or backlog. Its near-term monetization path depends on converting Starcloud-2 from a planned spacecraft into an operating commercial compute service, beginning with workloads such as satellite-data processing, storage and limited cloud capacity. Crusoe is the clearest publicly named early commercial customer.
Three growth inputs are visible even without financial statements. First, the company has raised substantial venture capital, including $420M across the announced March and August 2026 Series A tranches. Second, it is adding manufacturing capacity through a new 100,000-square-foot Woodinville facility. Third, it has contracted for optical intersatellite connectivity through Starlink, which could reduce dependence on direct ground-station downlinks. None of those inputs is equivalent to recognized revenue or delivered production.
Longer term, Starcloud's economics depend heavily on launch cost, reliable heavy-lift availability, thermal performance, radiation tolerance and the ability to sell enough high-value compute to offset the capital cost of continuously replacing spacecraft.
Manhattan West led the extension with participation from NVIDIA, Cisco Investments and existing investors. Starcloud said the transaction lifted total capital raised since 2024 to $450M and would support manufacturing, NVIDIA engineering work and launch allocation.
Starcloud signed for more than 50 Starlink Mini Laser terminals across at least 25 satellites. The company plans two terminals per spacecraft, making Starlink's optical mesh an important element of its planned distributed architecture.
NVIDIA named Starcloud among companies using its accelerated computing platforms for space missions and introduced the Space-1 Vera Rubin Module for orbital workloads.
Starcloud filed for authority to launch and operate an orbital data-center system of up to 88,000 non-geostationary satellites. The filing remains a proposal subject to FCC review and public-interest scrutiny.
The technology demonstrator carried an H100 GPU into orbit. Starcloud subsequently reported in-orbit model inference and training demonstrations.
Crusoe announced a strategic partnership to place Crusoe Cloud on Starcloud hardware, with limited GPU capacity planned after spacecraft deployment.
Starcloud has moved beyond a slide-deck-only orbital-compute thesis because it has flown an NVIDIA H100 and attracted major strategic capital. The decisive test now changes from technical novelty to repeatable commercial execution: Starcloud-2 must launch, remain thermally stable, deliver useful workloads and show customers will pay for compute in orbit.
For national security, the near-term relevance is not an 88,000-satellite AI supercluster. It is the possibility of processing intelligence, surveillance and reconnaissance data closer to the sensor, reducing raw-data downlink requirements and supporting resilient storage or compute outside terrestrial infrastructure. Those advantages are potentially valuable in contested communications environments, but Starcloud has not publicly disclosed a Defense Department production contract.
The capital question is equally important. Orbital compute is unusually launch-intensive and hardware must be replaced over time. Starcloud's latest financing gives it more room to fund spacecraft, manufacturing and launch reservations, but its economic model remains exposed to launch supply and costs that the company does not control.
| Competitor | Overlap | Relative Position |
|---|---|---|
| SpaceX | Dedicated large-scale orbital data centers, launch and optical networking | Far greater launch and constellation integration; orbital-data-center system remains proposed |
| Blue Origin | Large proposed orbital data-center constellation and heavy-lift launch | Potential vertical-integration advantage if launch cadence and data-center architecture mature |
| Cowboy Space | Dedicated orbital power and compute infrastructure | Direct startup competition for capital, launch access and early customers |
| Axiom Space | On-orbit compute and data-center nodes tied to commercial space-station infrastructure | Different architecture with nearer-term station-based compute use cases |
| Sophia Space | Modular orbital compute and edge AI | More focused on hosted/modular processing than Starcloud's long-range hyperscale constellation thesis |
Starcloud's strongest present advantage is demonstrated integration experience. It has already operated a high-end NVIDIA data-center GPU in orbit, which reduces one category of technical uncertainty that many proposed orbital-data-center projects have not yet crossed. Its founding team also combines satellite structures, large cloud/GPU systems and Starlink software experience.
The second potential advantage is strategic supplier alignment. NVIDIA is both a technology collaborator and an investor, while the Starlink laser contract gives Starcloud a path to high-bandwidth intersatellite connectivity without building a global optical relay network from scratch. That can accelerate deployment, but it also creates supplier dependence.
This is not yet a durable moat. SpaceX has launch, satellite manufacturing, optical networking and constellation operations under one corporate roof. Blue Origin can pair orbital infrastructure ambitions with New Glenn. Starcloud must prove that focused execution, capital and early technical lead can offset competitors with much deeper vertical integration.
Starcloud-1 is a demonstrator. The company has not yet publicly demonstrated sustained commercial cloud revenue from orbit.
The business model requires frequent, affordable launch. Capacity constraints or delays in heavy-lift systems could materially slow scaling.
High-power GPUs create substantial waste heat and terrestrial-class electronics face radiation effects. Starcloud must prove these systems at larger scale and over long durations.
The 88,000-satellite architecture requires FCC authorization and faces public scrutiny involving debris, astronomy and environmental effects.
Revenue, burn rate, cash and backlog are undisclosed. Large-scale constellation deployment would require substantial ongoing capital beyond today's financing.
The current architecture depends materially on third parties including launch providers, NVIDIA compute hardware and Starlink optical networking.
Crusoe is the clearest named early customer. Broader customer concentration and contracted revenue are not publicly disclosed.
A new 100,000-square-foot facility is evidence of planned capacity, not evidence of qualified satellite output at production scale.
Starcloud does not publish a current complete board roster. Its corrected March 2026 Series A announcement says Benchmark General Partner Chetan Puttagunta joined the board. The January 2025 SEC Form D had identified Philip Johnston, Ezra Feilden, Adi Oltean and NFX's Morgan Beller as directors, but Defense Briefing did not find a later primary filing confirming that every 2025 director still serves. The August 2026 financing announcement identified Lauren Selig as a board observer, not a director.
| Filed / Published | Document | Registry / Authority | Link |
|---|---|---|---|
| 25-AUG-2026 | Experimental Special Temporary Authority application 1702-EX-ST-2026, call sign WP2XJD | FCC Experimental Licensing System; applicant shown as Lumen Orbit, Inc. | Public filing index → |
| 04-FEB-2026 | SAT-LOA-20260202-00073, Starcloud orbital datacenter system | Federal Communications Commission, International Communications Filing System | FCC application → |
| 12-JAN-2025 | Form D, Rule 506(b) exempt offering | U.S. Securities and Exchange Commission; issuer shown as Lumen Orbit, Inc. | SEC filing → |
Private companies do not file 10-K, 10-Q or DEF 14A reports unless another legal obligation applies. No audited public financial statements were located for Starcloud. The FCC.report link above is an independent public index that reproduces FCC Experimental Licensing System data and is used because the FCC's current dynamic interface is difficult to deep-link reliably.
N/A: no Section 16 reporting. Starcloud has no registered public equity class, so Forms 3, 4 and 5 do not provide an insider-trading record.
| Date | Round / Filing | Lead / Notable Investors | Amount | Post-Money |
|---|---|---|---|---|
| 21-AUG-2026 | Series A extension | Manhattan West; NVIDIA, Cisco Investments, Benchmark, EQT, Soma, NFX, 776 and others | $250M | $2.3B |
| 30-MAR-2026 | Series A | Benchmark and EQT led; additional investors publicly reported by company and participants | $170M | $1.1B |
| 12-JAN-2025 filing | SEC Form D, Rule 506(b) | 65 investors in filing; related persons included founders and Morgan Beller | $14.137M sold under notice | Not disclosed |
Starcloud states it has raised $450M since its 2024 founding. The publicly itemized amounts above do not fully reconstruct that total because the SEC Form D is a filing for one exempt offering and does not necessarily map one-for-one to every company-announced financing tranche. Defense Briefing therefore uses the company's $450M aggregate without inventing an undisclosed bridge amount.
The business now branded Starcloud emerged from Lumen Orbit. The January 2025 SEC Form D says Lumen Orbit, Inc. was incorporated in Delaware in 2023, while Starcloud's current public materials describe the company as founded in 2024. The founders are Philip Johnston, Ezra Feilden and Adi Oltean. The company participated in Y Combinator's 2024 cohort and initially raised venture capital around the thesis that falling launch costs could make orbital computing economically useful.
The first major technical milestone came in November 2025 when Starcloud-1 reached orbit carrying an NVIDIA H100 GPU. In October 2025, Crusoe had already announced plans to deploy its cloud software on a later Starcloud spacecraft. During 2026 the company moved rapidly from demonstration toward scale: it announced a $170M Series A in March, joined NVIDIA's Space-1 orbital-computing ecosystem, contracted for Starlink Mini Laser terminals in May and filed for an 88,000-satellite orbital data-center system with the FCC.
On 21 August 2026, Starcloud announced another $250M of Series A financing at a $2.3B post-money valuation and said it was building out a 100,000-square-foot Woodinville manufacturing facility. That capital and infrastructure do not resolve the central execution question. The company still has to convert a one-satellite technical achievement into reliable commercial service and then into a repeatable production and launch system.