Core Insights - AlphaTON Capital Corp has signed a strategic $46 million deal to expand its AI compute infrastructure, adding 576 NVIDIA B300 chips, which is expected to achieve a projected equity multiple of 3.82x [1][7] - The deployment is part of AlphaTON's initiative to provide privacy-preserving, decentralized AI infrastructure integrated with Telegram's ecosystem, addressing a critical market gap for AI workloads that cannot be run on Big Tech infrastructure [2][5] Financial Structure - The project is financed through a combination of $4 million in cash, $32.7 million in closed non-recourse debt financing, and $9.3 million in equity, with a two-month grace period for installation [3] - The projected economics include a 27% internal rate of return (IRR), a total return on investment (ROI) of 282%, and a net present value (NPV) of $11 million [7] Market Opportunity - The AI infrastructure market is projected to require up to $7 trillion in investment by 2030, with 70% of new demand driven by AI workloads, indicating a significant growth opportunity for AlphaTON [11] - Major tech companies are expected to spend over $600 billion on AI infrastructure in 2026, highlighting the critical nature of AI infrastructure as a national priority [10] Strategic Positioning - AlphaTON is positioned at the intersection of three major trends: the $46 billion AI infrastructure opportunity, the global shift toward privacy and data ownership rights, and the rise of decentralized computing networks [2][6] - The company aims to provide a full-stack, vertically integrated approach to data sovereignty and privacy-centric AI technologies, differentiating itself from Big Tech [5][8] Deployment Details - The 576 NVIDIA B300 chips will be hosted at AtNorth's sustainable, hydroelectric-powered data center in Sweden, ensuring low-cost and clean energy usage [4] - Managed services for the deployment will be provided by CUDO Compute and SNET Energy Ltd, with financing arranged by Vertical Data and LEAP [4]

Core Viewpoint - Digi Power X Inc is transitioning from Bitcoin mining to focus on AI data centers, aiming to leverage its existing power infrastructure for higher-value opportunities in the AI sector [1][2]. Company Strategy - The company is rebranding itself as a power infrastructure provider, specifically targeting Tier 3 NeoCloud AI data centers [1][2]. - Digi Power X has four operational sites and approximately 100 megawatts of Bitcoin mining capacity, with plans to develop its first Tier 3 AI data center in Alabama [2][3]. - The economic valuation of AI infrastructure is significantly higher than Bitcoin mining, with AI infrastructure valued at $10 million to $12 million per megawatt compared to $500,000 per megawatt for Bitcoin mining [2]. Infrastructure and Deployment - The Alabama site is expected to be operational by late 2025, initially using 200 Nvidia B200 chips, with plans to upgrade to B300s in early 2026 [3]. - The company can build a Tier 3 data center in six to nine months, significantly faster than the typical three to five years required by new entrants [4]. - Digi Power X's partnership with Super Micro Computer enhances its capabilities, allowing for rapid deployment of AI data centers optimized for high-performance computing [6][7]. Financial Performance - The company's model aims to deliver predictable cash flow and strong margins, with a payback period of less than three years [8]. - Digi Power X's power generation assets have been profitable, supplying electricity to the grid at margins described as "exceptional" during recent heatwaves [11]. - The company is rebating gas at $2.55 per gigajoule, resulting in an effective cost of around $31–32 per megawatt-hour, while the grid pays up to $150 per megawatt-hour [11]. Market Positioning - Digi Power X positions itself as a complementary infrastructure provider to major tech companies, focusing on dedicated AI infrastructure solutions [12]. - The company is refining its execution strategy to convert megawatts into higher-value infrastructure, aiming for a 20 to 25 times multiple [12].