Starcloud Raises $250M to Scale AI Beyond Earth

Starcloud has raised $250 million in a Series A extension, boosting its valuation to $2.3 billion as it accelerates plans to build orbital data centers for AI workloads. The funding underscores growing investor confidence in space-based computing, with Nvidia and Cisco joining as strategic backers.

Starcloud aims to build orbital data centers leveraging continuous solar power and radiative cooling for large-scale AI compute.

Founded in 2024 2024 (originally as Lumen Orbit, later rebranded to Starcloud), Starcloud is a U.S.-based startup pioneering orbital data centers. Its founders are Philip Johnston (CEO, ex-McKinsey), Ezra Feilden (CTO, ex-Airbus Defence and Space), and Adi Oltean (Chief Engineer, ex-SpaceX and Microsoft Azure). The company has raised $450M to date, with investors including Benchmark, EQT, Manhattan West, NVIDIA, Cisco Investments, and others. 

Funding and Valuation

• Amount Raised: $250 million (Series A extension)
• Valuation: $2.3 billion post-money
• Total Capital Raised: $450 million since founding in 2024
• Lead Investor: Manhattan West
• Other Backers: Benchmark, EQT, Soma, NFX, 776, Nvidia, Cisco Investments, Cedar Capital, Goanna Capital, Standard Capital

Orbital Data Center Vision

• Core Idea: Move AI processing into orbit, reducing latency by analyzing data closer to where it is collected
• Advantages Over Earth-Based Centers:
  • No need for complex cooling systems
  • Continuous solar power without weather disruptions
  • Lower latency for space applications
• Target Scale: Constellation of 88,000 satellites delivering 20 gigawatts of compute capacity

Technology and Partnerships

• Nvidia Collaboration:
  • First NVIDIA H100 GPU flown to orbit in 2025
  • Development of Space-1 Vera Rubin Module, designed for radiation-heavy orbital environments
  • Future satellites expected to deliver 25x more compute than current H100 GPUs
• Cisco Role: Providing networking and AI infrastructure expertise for orbital systems
• Upcoming Hardware:
  • Starcloud-2 (2027): AI chips, storage, and backup modules
  • Starcloud-3: 200 kW satellites with advanced cooling and heat dissipation
  • Starcloud-4 (future): Cylindrical orbital data center with a 2.5-square-mile solar array

Manufacturing Expansion

• Facility: 100,000-square-foot plant in Woodinville, Washington
• Purpose: Mass production of Starcloud-3 spacecraft

Risks and Challenges

• High Capital Needs: Orbital infrastructure requires billions in long-term investment
• Launch Constraints: Reliance on SpaceX’s Starship as Falcon 9 phases out by 2028
• Regulatory Hurdles: FCC approval sought for 88,000 satellites
• Competition: SpaceX’s “Starmind” project envisions up to a million orbital data center satellites

Market Context

• SpaceX IPO Impact: Renewed investor enthusiasm for space-tech startups
• Global Trend: Orbital data centers seen as a solution to terrestrial bottlenecks in land, power, and water usage. 

Intel Launches Starfire: AI‑Powered Chip to Transform Space Computing

Intel has unveiled Starfire, its first space‑grade chip built on the advanced Intel 18A process, designed to power AI workloads directly aboard satellites and spacecraft. The processor is engineered to withstand radiation, extreme temperatures, and long missions, marking Intel’s bold entry into the aerospace computing market.

Currently, most satellites and spacecraft use  radiation‑hardened processors like BAE Systems’ RAD750 and RAD5545, but newer missions are increasingly adopting commercial off‑the‑shelf (COTS) chips such as NVIDIA Jetson Orin and ARM‑based SoCs for AI workloads. NASA and Microchip are also developing next‑generation high‑performance spaceflight computing systems.

The space computing market has long relied on BAE Systems’ RAD750 and RAD5545, with Microchip developing NASA’s next‑gen processor. Starfire is Intel’s bid to disrupt this dominance.

Starfire chip was developed under Intel Government Technologies with strong alignment to U.S. defense and aerospace programs, but it is not restricted to government use alone. While U.S. manufacturing and security programs are central, Intel has signaled broader availability for international partners once qualification is complete.

Starfire is part of Pentagon‑linked initiatives like RAMP‑C and SHIP, ensuring trusted supply chains and radiation‑hardening standards. Intel has positioned Starfire for private operators who want onboard AI inference for navigation, imaging, and scientific data processing.

Starfire is designed first for U.S. government and defense applications, but Intel intends it to be a dual‑use technology — serving both national security and commercial satellite markets worldwide. Its U.S. manufacturing under the Trusted Foundry program ensures compliance with defense standards, while its AI capabilities make it attractive for global space operators.

Key Highlights of Intel’s Starfire Chip

• Space‑grade design: Built to survive radiation, thermal cycling, and 10+ year missions in orbit.
• AI acceleration: Up to 75 TOPS of performance via a three‑tile neural processing unit.
• CPU architecture: Eight cores (4 performance + 4 efficiency) on Intel’s 18A node.
• GPU integration: Four‑core Xe GPU with 64 execution units, built on Intel 3.
• Packaging: Uses Intel’s Foveros 3D stacking for compact, resilient design.
• Variants: Low‑power (10 W, up to 45 TOPS) and Performance (35 W, up to 75 TOPS).

Key Specs from Intel Starfire Datasheet

Feature	Details
CPU	8 cores (4 performance + 4 efficiency) on Intel 18A
GPU	4 Xe cores, 64 execution units (Intel 3)
NPU	3‑tile design, up to 75 TOPS
Memory	LPDDR5 / DDR5 support
Connectivity	12 PCIe Gen4 lanes
Thermal Range	−55°C to 125°C
Mission Life	10+ years

Source - Intel 

Comparison: Starfire vs Legacy Space Chips

Chip	Performance	Process Node	AI Capability	Mission Lifespan
Intel Starfire	Up to 75 TOPS	Intel 18A (CPU/NPU), Intel 3 (GPU)	Dedicated NPU for AI inference	10+ years
BAE RAD750	110–200 MHz	150–250 nm	None	Proven on Mars rovers
BAE RAD5545	Multi‑core, higher throughput	Mature node	Limited	Long‑duration missions
NASA/Microchip Next‑Gen	100× current throughput (in development)	TBD	AI‑ready	Future missions

Strategic Impact

• National Security: Aligns with U.S. government’s emphasis on space as a defense domain.
• Commercial Satellites: Enables autonomous navigation, onboard image processing, and scientific data analysis without ground reliance.
• Market Disruption: Competes with aerospace chips like BAE’s RAD750, offering modern AI capabilities.

Challenges & Risks

• Radiation qualification pending: Validation against total ionizing dose and single‑event effects still in progress.
• Yield concerns: Intel’s 18A node may face production yield issues until 2027.
• Adoption timeline: Engineering samples ship in Q3 2026, broader deployment post‑qualification.

Bottom Line

Intel’s Starfire chip represents a major leap in space computing, bringing modern AI and high‑performance processing into orbit. If radiation testing succeeds, it could redefine how satellites and spacecraft handle data, shifting from ground‑based reliance to autonomous, onboard intelligence.