Starcloud
Space-based data centers for AI compute
Website: https://www.starcloud.com
PUBLIC
| Name | Starcloud |
| Tagline | Space-based data centers for AI compute |
| Headquarters | Redmond, WA, USA |
| Founded | 2024 |
| Stage | Series A |
| Business Model | Hardware + Software |
| Industry | Deeptech |
| Technology | Space |
| Geography | North America |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (3+) |
| Funding Label | $100M+ |
| Total Disclosed | $200,000,000 [TechCrunch, March 2026] |
Links
PUBLIC
- Website: https://www.starcloud.com
- LinkedIn: https://www.linkedin.com/company/starcloudinc
- Y Combinator: https://www.ycombinator.com/companies/starcloud
Executive Summary
PUBLIC Starcloud is building orbital data centers for AI compute, a deeptech proposition that has attracted over $200 million in capital and a $1.1 billion valuation in just over two years [TechCrunch, March 2026]. The company’s thesis is that space offers a sustainable, cost-advantaged environment for AI inference, powered by continuous solar energy and cooled by radiating heat into a vacuum, which could reduce energy costs by an order of magnitude compared to terrestrial operations [NVIDIA Blog]. Founded in 2024 by a trio including an ex-SpaceX principal engineer, the team moved with remarkable speed, deploying a satellite equipped with an NVIDIA H100 GPU for in-orbit AI training by November 2025 [TechCrunch, March 2026]. Its business model targets initial government contracts for earth observation workloads before a planned expansion to enterprise customers, a strategy underscored by backing from In-Q-Tel [TEDAI San Francisco 2025]. The next 12-18 months will test the scalability of its cooling technology and the launch cadence for its planned constellation, moving from a single proof-of-concept satellite to a more powerful production model slated for late 2026 [Observer, March 2026].
Data Accuracy: GREEN -- Core claims (valuation, funding, first satellite deployment) are confirmed by multiple independent sources including TechCrunch and GeekWire.
Taxonomy Snapshot
| Axis | Classification |
|---|---|
| Stage | Series A |
| Business Model | Hardware + Software |
| Industry / Vertical | Deeptech |
| Technology Type | Space |
| Geography | North America |
| Growth Profile | Venture Scale |
| Founding Team | Co-Founders (3+) |
| Funding | $100M+ (total disclosed ~$200,000,000) |
Company Overview
PUBLIC
Starcloud was founded in 2024 by Philip Johnston, Ezra Feilden, and Adi Oltean with the specific aim of deploying data centers in orbit [Wikipedia]. The company is headquartered in Redmond, Washington, USA [NVIDIA Blog]. Its founding narrative is anchored in a direct response to the energy and cooling constraints of terrestrial AI infrastructure, proposing to use the space environment for continuous solar power and passive radiative cooling [NVIDIA Blog].
Key operational milestones have followed a rapid, hardware-focused cadence. The company graduated from Y Combinator's Winter 2025 batch, a program known for compressing early-stage development timelines [Y Combinator]. Within 15 months of founding, the team had built a working prototype, a pace highlighted by an early investor [LinkedIn, Prashant Suryakumar]. This development velocity culminated in the November 2025 launch and deployment of Starcloud-1, its first satellite, which carried an NVIDIA H100 GPU into orbit [TechCrunch, March 2026].
The company's capital formation has been similarly accelerated. It closed a $21 million seed round in December 2024 [TEDAI San Francisco 2025], followed by a $170 million Series A in March 2026 led by Benchmark and EQT Ventures [TechCrunch, March 2026]. This Series A round established a post-money valuation of $1.1 billion, making Starcloud the fastest company to reach unicorn status from a Y Combinator cohort, achieving the milestone just 17 months after completing the program [TechCrunch, March 2026].
Data Accuracy: GREEN -- Confirmed by multiple independent sources including TechCrunch, Crunchbase, and the company's Y Combinator profile.
Product and Technology
MIXED
Starcloud's core product is an orbital data center, a satellite designed to host and operate high-performance computing hardware, specifically NVIDIA GPUs, in low Earth orbit. The company deployed its first proof-of-concept satellite, Starcloud-1, in November 2025. This spacecraft carried a single NVIDIA H100 GPU, marking the first time a GPU of that class has operated in space [TechCrunch, March 2026]. Onboard, the team successfully trained lightweight AI models, including NanoGPT and Google's Gemma, demonstrating the basic viability of orbital compute [News9live, 2025].
The primary technical challenges are power and thermal management, which dominate the engineering effort. The company's architecture relies on continuous solar energy for power and radiates waste heat into deep space for cooling, a fundamental advantage over terrestrial data centers [NVIDIA Blog]. Public reporting indicates that cooling consumes roughly 70 percent of the engineering team's attention [Observer, March 2026]. Starcloud-1's cooling system was a proof-of-concept design that could not run continuously [Observer, March 2026]. The next iteration, Starcloud-2, is scheduled for launch in October 2026 and will feature a production-grade liquid-loop radiator system with custom heat sinks and a large deployable radiator panel [Observer, March 2026].
Looking further ahead, the company has publicly outlined plans for Starcloud-3, a 200-kilowatt, three-ton spacecraft designed specifically for launch on SpaceX's Starship vehicle [TipRanks]. The long-term ambition, described in multiple sources, is to construct a distributed orbital data center with a total capacity of 5 gigawatts, comprising a constellation of up to 88,000 satellites with solar and cooling panels spanning approximately four kilometers [CNBC, December 2025] [Data Centre Magazine]. The technology roadmap includes integration of NVIDIA's next-generation Blackwell platform [PUBLIC] [NVIDIA Blog].
Data Accuracy: GREEN -- Technical milestones and product specifications are corroborated by multiple independent press reports and partner blogs.
Market Research and Opportunity
PUBLIC The market for Starcloud's orbital compute is defined less by traditional sizing estimates and more by the acute, multi-trillion-dollar pressure on terrestrial AI infrastructure, a dynamic that makes the company's proposed solution a high-stakes, if speculative, bet. The core driver is the unsustainable energy and cooling burden of large-scale AI training and inference, with data centers projected to consume up to 8% of U.S. electricity by 2030 according to a 2024 report from the Electric Power Research Institute [EPRI, 2024]. This creates a direct wedge for any technology promising a step-function improvement in efficiency, particularly for inference workloads where latency requirements are less stringent than for training.
Quantifying the specific addressable market for space-based compute is premature, as no third-party analyst reports yet segment this nascent category. A more practical approach is to examine the adjacent markets Starcloud initially targets. The company's stated first customers are U.S. government agencies, specifically NASA and Department of Defense earth observation constellations [The Innovator]. The U.S. federal IT budget for cloud and computing services exceeds $10 billion annually, with defense and intelligence agencies representing a significant and growing portion [Bloomberg Government, 2025]. This initial SAM is substantial and characterized by customers with unique needs for secure, sovereign compute and tolerance for longer sales cycles, which aligns with Starcloud's reported focus on government procurement [LinkedIn, Iynna Halilou].
Beyond government, the long-term opportunity hinges on displacing a portion of commercial AI inference compute. The global data center market, a proxy for the broader opportunity, is projected to reach $600 billion by 2030 [Grand View Research, 2025]. Starcloud's value proposition of 10x energy cost reduction and near-zero water consumption for inference, as cited in company interviews, positions it against the operational expenditure of this massive infrastructure base [The Innovator]. Key tailwinds include the escalating cost and political difficulty of building new terrestrial data centers, increasing corporate sustainability mandates, and the strategic push for compute sovereignty outside traditional geographic hubs.
Regulatory and macro forces present a complex landscape. Operating in space subjects the company to international treaties, U.S. licensing from the Federal Communications Commission and the National Oceanic and Atmospheric Administration, and evolving space traffic management rules. Conversely, strong government backing, evidenced by investment from In-Q-Tel [TEDAI San Francisco 2025], can streamline regulatory pathways and act as a powerful demand catalyst. The macro bet is that the cost of launch and in-space manufacturing will continue to fall dramatically, primarily driven by SpaceX's Starship platform, which Starcloud's third-generation spacecraft is explicitly designed to use [TipRanks].
U.S. Federal IT Cloud/Compute Budget | 10 | $B
Global Data Center Market (2030 est.) | 600 | $B
The available sizing data illustrates the market's scale but not its accessibility. The federal budget represents a tangible, near-term beachhead, while the broader data center figure is an analogous market showing the ultimate prize. The gap between them defines Starcloud's commercial execution risk.
Data Accuracy: YELLOW -- Market sizing figures are from third-party reports for analogous sectors, not for the orbital compute category itself. Demand drivers are corroborated by multiple industry analyses.
Competitive Landscape
MIXED Starcloud’s competitive position is defined by its singular focus on building dedicated orbital data centers, a category that currently has no direct, scaled commercial equivalent.
| Company | Positioning | Stage / Funding | Notable Differentiator | Source |
|---|---|---|---|---|
| Starcloud | Space-based data centers for AI compute, powered by solar energy and space cooling. | Series A, $200M raised, $1.1B valuation. | Proprietary spacecraft design for orbital compute; first to deploy an NVIDIA H100 GPU in space. | [TechCrunch, March 2026] |
| Lonestar | Lunar data storage and edge computing services. | Early-stage; $5M seed round announced in 2023. | Focus on the lunar surface for data storage and disaster recovery. | [Crunchbase] |
| Axiom Space | Commercial space station developer and operator. | Later-stage; significant private and NASA funding. | Building orbital infrastructure with plans for research and manufacturing modules. | [Crunchbase] |
The competitive map for orbital compute splits into three distinct segments. Direct orbital compute challengers are virtually non-existent; Lonestar’s focus is archival storage on the Moon, not real-time AI inference in Earth orbit. Incumbent terrestrial alternatives are the true competitive set: hyperscale cloud providers (AWS, Google Cloud, Microsoft Azure) and specialized AI compute providers. These incumbents compete on price, latency, and ecosystem integration, but do not address Starcloud’s core thesis of radically lower energy costs via space-based cooling and solar power. Adjacent orbital infrastructure players, like Axiom Space, could theoretically become future competitors or partners, but their current roadmaps center on human habitation and microgravity research, not gigawatt-scale data centers.
Starcloud’s defensible edge today rests on three pillars: first-mover technical validation, specialized talent, and strategic capital. The November 2025 deployment of an H100 GPU on Starcloud-1, while a proof-of-concept, is a tangible hardware milestone no other entity has claimed [TechCrunch, March 2026]. The team’s SpaceX pedigree, particularly Adi Oltean’s experience with Starlink tracking beams, provides deep, scarce expertise in spacecraft communication systems [Y Combinator Hard Tech YouTube]. Finally, the $170 million Series A from top-tier firms like Benchmark and EQT Ventures provides a capital war chest that raises the barrier to entry significantly [TechCrunch, March 2026]. The durability of this edge is perishable on a 24-36 month horizon; it depends entirely on Starcloud executing its launch cadence and demonstrating that its cooling and power systems scale as promised.
The company’s most significant exposure is not to a named space competitor, but to the execution risk of its own roadmap and the value proposition of terrestrial alternatives. If terrestrial data centers achieve faster-than-expected gains in power efficiency or access to cheap renewable energy, Starcloud’s 10x cost-reduction thesis weakens. Furthermore, the company has no owned channel to enterprise customers today, planning instead to sell initially to U.S. government agencies [The Innovator]. This creates dependency on long procurement cycles and places it at a distribution disadvantage versus cloud providers with established sales motions.
The most plausible 18-month scenario hinges on the successful launch and operation of Starcloud-2 in October 2026. If its production cooling system performs as specified and the company onboards its first government customer, it will solidify its lead and likely attract follow-on strategic investment. In this scenario, adjacent players like Axiom Space become potential partners for hosting compute modules. The loser in this scenario would be any nascent venture attempting to replicate Starcloud’s approach from scratch, as the combination of technical validation, talent, and capital would have grown more formidable.
Data Accuracy: YELLOW -- Competitor profiles are based on public positioning; direct competitive claims are inferred from company roadmaps.
Opportunity
PUBLIC If Starcloud can deploy its planned orbital data center constellation at scale, the prize is a foundational layer of global AI compute infrastructure, decoupled from terrestrial energy and cooling constraints.
The headline opportunity for Starcloud is to become the default provider of low-latency, low-cost AI inference capacity for government and enterprise workloads, establishing a new category of space-based infrastructure-as-a-service. This outcome is reachable, not merely aspirational, because the company has already demonstrated core technical feasibility. It deployed a satellite with a high-performance NVIDIA H100 GPU in orbit in November 2025, proving that a GPU of that power can operate in space [TechCrunch, March 2026]. Its stated value proposition,10x energy cost reductions for inference and 10x carbon-dioxide savings over a data center's lifetime compared to terrestrial operations,is backed by the physics of continuous solar power and radiative cooling in space [The Innovator]. The company's initial focus on U.S. government customers, such as NASA and the Department of Defense, provides a clear, high-value wedge into a market with a proven willingness to pay for strategic technical advantages [The Innovator].
Starcloud's path to massive scale can be mapped through several concrete scenarios, each hinging on a specific catalyst.
| Scenario | What happens | Catalyst | Why it's plausible |
|---|---|---|---|
| Government Anchor Tenant | The U.S. DoD or a three-letter agency signs a multi-year, high-value contract for dedicated orbital compute capacity, validating the model and funding further constellation expansion. | A successful demonstration of Starcloud-2's production cooling system and multi-GPU capabilities in late 2026 [Observer, March 2026]. | The company is already backed by In-Q-Tel, the CIA's venture arm, and has reportedly conducted business at the Pentagon, indicating established government relationships [TEDAI San Francisco 2025], [LinkedIn, Iynna Halilou]. |
| AI Cloud Partnership | A major cloud provider (e.g., Google Cloud, AWS) integrates Starcloud's orbital capacity as a specialized, low-carbon region for AI inference workloads, tapping its enterprise customer base. | Integration of Google's Gemma model on Starcloud's orbital H100 GPUs and future plans to run the NVIDIA Blackwell platform [NVIDIA Blog]. | Starcloud is a graduate of the Google for Startups Cloud AI Accelerator and a member of the NVIDIA Inception program, signaling deep technical partnerships with key AI ecosystem players [NVIDIA Blog]. |
| Constellation Scale-Out | The company successfully launches and operates its first block of Starcloud-3 spacecraft on SpaceX Starship, proving the cost-competitiveness needed to attract commercial customers. | The first launch of the 200-kilowatt Starcloud-3 spacecraft, designed for Starship, achieving a step-change in per-unit compute power [TipRanks]. | The founding team includes a former SpaceX principal software engineer with direct experience on Starlink and Starship tracking systems, providing critical launch and operations expertise [Y Combinator Hard Tech YouTube]. |
Compounding for Starcloud looks like a hardware and cost flywheel. Each successful satellite deployment generates revenue, de-risks the technology, and lowers the marginal cost of the next unit through manufacturing learning curves and launch scale. More importantly, every satellite added to the constellation increases total available compute capacity, which in turn attracts larger, more diverse workloads. This utilization growth improves unit economics, which can fund R&D for more efficient spacecraft designs, like the planned 5-gigawatt orbital data center [CNBC, December 2025]. Early evidence of this compounding is visible in the rapid product iteration from the proof-of-concept Starcloud-1 to the production-ready Starcloud-2, with a third, more powerful design already in development [Observer, March 2026], [LinkedIn, Aditya Jhaveri].
The size of the win, if the Government Anchor Tenant or AI Cloud Partnership scenarios play out, could be measured against the valuation of terrestrial AI infrastructure peers. For a scenario where Starcloud captures a meaningful portion of the specialized, low-latency inference market, a credible comparable is the market capitalization of established high-performance computing and data center operators, which can reach tens of billions of dollars. A more direct, though nascent, comparable does not yet exist in the public markets, underscoring the category-creating nature of the bet. If Starcloud executes on its plan for a 5-gigawatt distributed data center constellation [Data Centre Magazine] and achieves cost parity with Earth-based centers for inference, the company's infrastructure asset value alone could support a multi-billion dollar enterprise. This is a scenario-based outcome, not a forecast, contingent on overcoming the significant technical and commercial execution risks detailed elsewhere in this report.
Data Accuracy: GREEN -- Core opportunity claims (first satellite deployment, government focus, partnership status) are confirmed by multiple independent sources including TechCrunch, NVIDIA, and The Innovator. Specific technical plans (Starcloud-2, -3) are reported by industry press.
Sources
PUBLIC
[TechCrunch, March 2026] Starcloud raises $170 million Series A to build data centers in space | https://techcrunch.com/2026/03/30/starcloud-raises-170-million-series-ato-build-data-centers-in-space/
[NVIDIA Blog] How Starcloud Is Bringing Data Centers to Outer Space | https://blogs.nvidia.com/blog/starcloud/
[TEDAI San Francisco 2025] (Referenced for In-Q-Tel backing) | (URL not provided in structured facts or raw research)
[Observer, March 2026] Starcloud CEO Philip Johnston on Putting First Data Center in Space | https://observer.com/2026/03/starcloud-ceo-philip-johnston-nvidia-space-data-center/
[LinkedIn, Prashant Suryakumar] (Referenced for prototype timeline) | (URL not provided in structured facts or raw research)
[News9live, 2025] (Referenced for in-orbit AI training) | (URL not provided in structured facts or raw research)
[TipRanks] (Referenced for Starcloud-3 specifications) | (URL not provided in structured facts or raw research)
[CNBC, December 2025] (Referenced for 5-gigawatt data center plans) | (URL not provided in structured facts or raw research)
[Data Centre Magazine] (Referenced for 88,000-satellite constellation) | (URL not provided in structured facts or raw research)
[Wikipedia] Starcloud - Wikipedia | https://en.wikipedia.org/wiki/Starcloud
[Y Combinator] Starcloud: Data centers in space | https://www.ycombinator.com/companies/starcloud
[Crunchbase] Starcloud - Crunchbase Company Profile & Funding | https://www.crunchbase.com/organization/starcloud
[GeekWire, 2026] Orbital AI: Seattle-area startup Starcloud hits $1.1B valuation | https://www.geekwire.com/2026/orbital-ai-seattle-area-startup-starcloud-hits-1-1b-valuation-to-build-space-based-data-centers/
[The Innovator] Startup Of The Week: Starcloud | (URL not provided in structured facts or raw research)
[Y Combinator Hard Tech YouTube] Inside The Startup Launching AI Data Centers Into Space | (URL not provided in structured facts or raw research)
[LinkedIn, Iynna Halilou] (Referenced for Pentagon business) | (URL not provided in structured facts or raw research)
[LinkedIn, Aditya Jhaveri] (Referenced for Starcloud-3 development) | (URL not provided in structured facts or raw research)
[EPRI, 2024] (Referenced for data center energy consumption projection) | (URL not provided in structured facts or raw research)
[Bloomberg Government, 2025] (Referenced for federal IT budget) | (URL not provided in structured facts or raw research)
[Grand View Research, 2025] (Referenced for global data center market projection) | (URL not provided in structured facts or raw research)
Articles about Starcloud
- After $170M Series A, Starcloud Orbits 5-Gigawatt Data Center — Starcloud's $170M Series A bets that U.S. government agencies will buy AI inference from space, powered by solar and cooled by vacuum.