Summary of Starcloud's Conference Call Company Overview - **Company**: Starcloud - **Industry**: Space-based data centers and AI compute infrastructure - **Headquarters**: Redmond, Washington, close to major space and technology hubs like Blue Origin, Amazon, Microsoft, and SpaceX [doc id='16'][doc id='17'] Key Points Industry Context - The construction of large, power-hungry data centers on Earth faces significant challenges, leading to interest in alternatives such as space-based data centers [doc id='1'][doc id='7'] - Space-based data centers are reshaping investment discussions across public and private markets, with notable figures like Sundar Pichai (Google), Jeff Bezos (Amazon), and Elon Musk (SpaceX) advocating for this shift [doc id='3'] Starcloud's Innovations - Starcloud is the first company to train an AI model in orbit, achieving this milestone with its Starcloud-1 demonstrator satellite launched in November 2025 [doc id='1'][doc id='58] - The company plans to launch its first commercial GPU cluster, Starcloud-2, in late 2026, which will feature enhanced capabilities compared to its predecessor [doc id='59'] Market Dynamics - The U.S. power demand from data centers is projected to rise from approximately 9% in 2026 to around 20% by the mid-2030s, indicating a growing need for alternative energy sources [doc id='7'][doc id='10'] - Political pushback against new data center construction is increasing, potentially impacting the future of terrestrial data centers [doc id='7'] Advantages of Space-Based Data Centers - **Power**: Space offers abundant solar power, with a near full solar constant of about 1,361 W/m², which is significantly higher than ground-level solar irradiance [doc id='26'] - **Cooling**: Space's cold environment provides natural cooling advantages, potentially reducing energy costs associated with cooling in terrestrial data centers [doc id='27] - **Scalability**: The rise of reusable rockets and decreasing launch costs will enable larger and more modular deployments of space-based infrastructure [doc id='28'] Challenges and Risks - **Radiation**: Space-based data centers face challenges from radiation, requiring specialized hardware for protection [doc id='33'] - **Security**: The lack of ownership in space leads to potential geopolitical risks, necessitating new defensive strategies for space-based assets [doc id='34'] - **Maintenance**: Hardware degradation due to space conditions complicates maintenance, requiring innovative solutions for remote servicing [doc id='36] Future Plans - Starcloud envisions a modularized 5GW data center design in sun-synchronous orbit, with plans for gradual scaling and deployment of distributed miniaturized data centers [doc id='48'][doc id='61] - The company aims to provide in-space infrastructure rather than being a GPU-as-a-service provider, allowing customers to supply their own GPUs [doc id='50] Partnerships and Competitors - Starcloud has formed strategic partnerships with companies like NVIDIA, Google, and Crusoe to enhance its capabilities and market reach [doc id='68'][doc id='70] - Competitors include Google’s Project Suncatcher, Amazon's Blue Origin, and Axiom Space, all exploring space-based data center solutions [doc id='75'][doc id='78] Funding and Valuation - Starcloud has raised $27.5 million to date, with a recent seed round of $10 million in March 2025, indicating strong investor interest [doc id='88'] Conclusion Starcloud is positioned at the forefront of a transformative shift in data center infrastructure, leveraging the unique advantages of space to address the growing demands of AI compute while navigating significant challenges and competition in the emerging market of space-based data centers.

Core Insights - VEON Ltd. announces the development of Ukraine's national large language model (LLM) in collaboration with Kyivstar and the WINWIN AI Center of Excellence, utilizing Google Gemma as the foundational model [1][2][3] Group 1: Project Overview - The Ukrainian LLM aims to encompass various Ukrainian dialects, terminology, and historical context while ensuring sensitive national data is securely processed within Ukraine [2] - Google was chosen for this initiative to strengthen technological ties between Ukraine and the United States, particularly following Kyivstar's Nasdaq listing in August 2025 [3] - The project is expected to enhance digital services across public and private sectors in Ukraine, with applications in areas such as education, finance, and healthcare [5] Group 2: Technical Development - Kyivstar will lead the operational development of the LLM, focusing on training the model with unique Ukrainian data to minimize linguistic and ethical risks [4] - The initiative will first optimize Google Gemma for the Ukrainian language, refining its tokenizer and training it on curated datasets [6] - Dedicated benchmarks will guide the model's fine-tuning and adaptation for specific applications [6] Group 3: Strategic Importance - This initiative is part of VEON's broader strategy to address the AI language gap in its markets, following similar projects in Kazakhstan and Pakistan [7] - The investment in the Ukrainian LLM is part of a larger commitment by Kyivstar and VEON to invest USD 1 billion in Ukraine from 2023 to 2027, focusing on infrastructure and technological development [8]