AlphaTON Capital Secures $82.5 Million Strategic GPU Infrastructure to Power Privacy-First Decentralized Artificial Intelligence Cluster after Successful Pilot Launch of Telegram’s Cocoon AI

Rhea-AI Summary

AlphaTON Capital (Nasdaq: ATON) announced agreements to acquire and deploy a 1,000+ Nvidia B200 GPU compute cluster as a revenue-generating asset, funded by approximately $30M equity and $52.5M debt, for total project capitalization of about $82.5M. The company says $70M of Nvidia hardware will be added to the balance sheet and the debt facility is amortized over 36 months.

AlphaTON reported a successful pilot of Telegram’s Cocoon AI on Nov 25, 2025 and plans dual revenue from priority Cocoon workloads and external GPU rentals via CUDO Compute. Management published model metrics including a projected IRR 59.7%, NPV $59.6M, ROI 615%, 90% target utilization, and EBITDA margins of 64–73% over five years.

Positive

• $70M Nvidia hardware added to balance sheet
• Total project capitalization of $82.5M
• Successful Cocoon AI pilot on Nov 25, 2025
• Projected IRR 59.7% and NPV $59.6M

Negative

• $52.5M debt facility amortized over 36 months
• Revenue model assumes 90% utilization target
• Financials assume 20% first-year price compression

Insights

Infrastructure Strategist

AlphaTON secures large Nvidia GPU cluster and partners to host Telegram's Cocoon AI, creating dedicated sovereign compute capacity.

The deal adds a 1,000+ Nvidia B200 GPU cluster and names CUDO Compute, SNET Energy UK Ltd, and Vertical Data as deployment and operations partners, positioning AlphaTON as a direct compute provider for Cocoon AI while reserving priority capacity for Telegram workloads. The transaction structure places the hardware as a balance-sheet asset and targets $70M in Nvidia hardware on the company books.

Primary dependencies include timely hardware delivery and operational integration under CUDO's orchestration, plus maintaining the stated 90% target utilization. Key operational risks are single-cluster concentration until planned expansions and execution of multi-tenant governance. Watch the Q1 2026 hardware delivery and validation milestones and the transition to full Cocoon workloads in Q2 2026 as near-term proof points for uptime and multi-tenant controls.

Financial Modeler

Projected returns are aggressive but anchored to explicit model inputs and a mixed capital structure with leverage.

The project is financed with $30M equity and $52.5M debt, creating total capitalization of $82.5M; the debt is amortized over 36 months with covenanted debt service coverage ratios above 1.0x. The release lists explicit outputs: projected IRR 59.7%, NPV $59.6M, equity multiple 5x, and a five-year total cash return of $150M.

Key sensitivities are utilization, price compression assumptions (stated as 20% in year one then 10% thereafter), and operating cost inflation embedded in the model. Monitor realized utilization, actual pricing trajectory against the assumed compression, and covenant compliance during the 36-month amortization; early confirmations in Q1 2026 and operational EBITDA margins in year one will materially affect covenant headroom and equity returns.

This Deal adds $70M of Nvidia GPUs to AlphaTON’s Balance Sheet, combining with the $27M Digital Asset Treasury to total $97M in Total Assets.

Dover, DE, Nov. 26, 2025 (GLOBE NEWSWIRE) -- AlphaTON Capital Corp. (Nasdaq: ATON) ("AlphaTON" or the "Company"), a digital asset technology company focused on the Telegram and TON ecosystem, today announced it has entered into agreements to acquire and deploy a 1,000+ Nvidia B200 GPU high-performance compute cluster valued at approximately $68 million. This strategic infrastructure investment positions AlphaTON as a critical compute provider for Telegram’s Cocoon AI - the world’s largest decentralized, privacy-preserving artificial intelligence network, while creating a diversified revenue stream through high-performance compute and GPU rental services.

TRANSACTION STRUCTURE AND FINANCING

The transaction will be financed through a strategic capital structure consisting of:

• $30 million in equity capital from AlphaTON Capital Corp.
• $52.5 million in debt financing to be facilitated by Vertical Data’s GPU Financing division
• Total project capitalization of approximately $82.5 million (includes additional working capital and deployment costs)
• Total of $70M of high-value Nvidia hardware assets added to the balance sheet

The debt facility facilitated by Vertical Data to be fully amortized over 36 months, with the project maintaining debt service coverage ratios above 1.0x throughout the amortization schedule. The compute cluster will be held as a revenue-generating hard asset on AlphaTON's balance sheet.

STRATEGIC INFRASTRUCTURE PARTNERSHIP

AlphaTON has assembled a consortium of best-in-class infrastructure partners to deploy and operate the GPU cluster:
CUDO Compute – A leading European NVIDIA Cloud Partner (NCP), will serve as the primary deployment and operations partner for the GPU infrastructure. CUDO will lead cluster installation, orchestration, workload management and lifecycle operations, while also providing the commercialization engine through its GPU marketplace and AI services platform. CUDO’s operational stack ensures production-grade uptime, enterprise-grade telemetry and scalable multi-tenant governance across the full cluster.

SNET Energy UK Ltd – Providing expertise in sustainable-power data center development and AI infrastructure optimization, SNET Energy's involvement ensures technical excellence and alignment with decentralized AI principles.

Vertical Data – Beyond debt financing, Vertical Data will provide the GPU cluster and deployment services, and hardware sourcing.

HARDWARE SPECIFICATIONS AND CAPABILITIES

The compute cluster will feature 1,000+ Nvidia B200-class GPUs, representing cutting-edge AI inference and training capabilities. The infrastructure is designed with a modular, upgradable architecture to accommodate rapid evolution in AI hardware and model architectures. The cluster will operate at 90% target utilization, serving dual revenue streams:

1. Dedicated Workloads for Cocoon AI – Priority allocation for Telegram/Cocoon AI's proprietary, privacy-preserving AI network
2. External GPU Rentals – Monetization through CUDO Compute's marketplace, initially for bare metal compute and progressively for AI services

COMPELLING PROJECTED PROJECT ECONOMICS

The investment presents institutional-grade returns with strong downside protection:

• Projected IRR: 59.7%
• Projected NPV: $59.6 million
• Projected Equity Multiply: 5x
• Projected ROI: 615%
• Projected Total Cash returned (5-year projection): $150 million

Revenue projections incorporate conservative assumptions, including 20% annual price compression in the first year and 10% in the following years, 90% utilisation rates, and comprehensive operating cost inflation. The financial Model demonstrates consistent EBITDA margin ranging from 64-73% over five year forecast horizon, with Year 1 gross profit margins of 32%

SUCCESSFUL PILOT LAUNCHED ON NOVEMBER 25, 2025

AlphaTON Capital announced the successful completion of a landmark pilot program utilizing Telegram’s innovative Cocoon AI infrastructure, a milestone made possible through a strategic collaboration that granted the firm early access. "Our strategic partnership with Telegram and Cocoon provided early access to the technology, and we successfully launched a Cocoon pilot on November 25, 2025," said  Logan Golema, CTO of AlphaTON Capital. "Following this validation, we are now going to deploy the technology into our secure and decentralized AI hosting infrastructure to further enhance our network's scalability and performance."

STRATEGIC RATIONALE

Democratizing AI with “AI For Good”

As Artificial Intelligence becomes increasingly centralized among a few global technology giants, AlphaTON is committed to an “AI For Good” mandate. This initiative prioritizes ethical access, user data sovereignty, and environmental responsibility. By supporting Cocoon, AlphaTON is helping to build a sovereign compute layer where user data remains encrypted and private, and where content rights are respected—a direct alternative to the surveillance-capitalism models prevalent in the industry today.

“We believe the future of AI must be decentralized, private, and accessible to all, not hoarded within the walled gardens of Big Tech,” said Brittany Kaiser, Chief Executive Officer of AlphaTON Capital. “This investment is not just about hardware; it is a commitment to ethical infrastructure. By powering Project Cocoon, we are giving Telegram’s nearly one billion users access to AI tools that respect their privacy and freedom, while ensuring our investors have direct exposure to the most valuable asset class of the next decade: sovereign compute.”

ALIGNMENT WITH COCOON AI'S DECENTRALIZED VISION

Cocoon AI, leveraging Telegram's 1 billion+ monthly active user base, is building a privacy-first alternative to centralized AI services dominated by Meta, Google, Amazon, and Microsoft/OpenAI. Cocoon's approach—running AI models on decentralized infrastructure with user data sovereignty—requires compute resources that are not owned, monitored, or controlled by Big Tech competitors.

AlphaTON's GPU cluster provides Cocoon AI with:

• Guaranteed Performance and Uptime – Dedicated capacity free from cloud provider throttling or resource contention
• Data Sovereignty – Complete control over data boundaries, ensuring user privacy is architecturally protected
• Infrastructure Independence – Freedom from platform risk and competitive intelligence exposure inherent in Big Tech cloud services
• Economic Value Capture – Direct ownership eliminates cloud intermediary margins, improving unit economics

DEPLOYMENT TIMELINE AND COMMERCIAL MILESTONES

1. Q4 2025: Transaction close, establishment of asset holding structure, Cocoon AI testing and launch on first fleet of machines
2. Q1 2026: Hardware delivery, validation of workloads, performance benchmarking, security audits, and soft launch for Cocoon AI inference
3. Q2 2026: Implementation of full Cocoon AI workloads; expansion of cluster orchestration capabilities; Planned expansion with 2-4 additional clusters, subject to market conditions and initial performance validation

MARKET CONTEXT AND COMPETITIVE POSITIONING

The investment comes at an inflection point in AI infrastructure markets. According to McKinsey analysis, data center AI workload capacity will grow 3.5x between 2025 and 2030, adding 124 gigawatts of incremental capacity. However, value capture is heavily skewed toward technology and chipmakers (60% of spending) versus real estate (15%) and power (25%). AlphaTON's model captures hardware asset appreciation while generating recurring rental revenue, positioning across multiple value pools.

The U.S. government's unprecedented 10% equity stake in Intel and the characterization of AI infrastructure as "buyer of last resort" by federal policymakers signal systematic support for the sector. As Google CEO Sundar Pichai stated to the BBC, an AI infrastructure bubble burst would represent "a total market contagion event," effectively designating the sector as too integral to fail.

About AlphaTON Capital Corp. (Nasdaq: ATON)

AlphaTON Capital is a specialized digital asset company focused on developing the Telegram ecosystem and managing a strategic reserve of TON tokens. The Company implements a comprehensive M&A and treasury strategy that combines direct token acquisition, validator operations, and strategic ecosystem investments to generate sustainable returns for shareholders. Through its operations, AlphaTON Capital provides public market investors with institutional-grade exposure to the TON ecosystem and Telegram's billion-user platform while maintaining the governance standards and reporting transparency of a Nasdaq-listed company. Led by Chief Executive Officer Brittany Kaiser, Executive Chairman and Chief Investment Officer Enzo Villani, and Chief Business Development Officer Yury Mitin, the Company's activities span network validation and staking operations, development of Telegram-based applications, and strategic investments in TON-based decentralized finance protocols, gaming platforms, and business applications.

AlphaTON Capital Corp is incorporated in the British Virgin Islands and trades on Nasdaq under the ticker symbol "ATON". AlphaTON Capital, through its legacy business, is also advancing potentially first-in-class therapies that target known checkpoint resistance pathways to potentially achieve durable treatment response and improve quality of life for patients. AlphaTON Capital actively engages in the drug development process and provides strategic counsel to guide development of novel immunotherapy assets and asset combinations. To learn more, please visit https://alphatoncapital.com/.

Forward-Looking Statements

This press release contains forward-looking statements within the meaning of the Private Securities Litigation Reform Act of 1995. These statements relate to future events or AlphaTON's future financial performance and involve known and unknown risks, uncertainties and other factors that may cause actual results to differ materially from those expressed or implied by these forward-looking statements. Factors that could cause or contribute to such differences include, but are not limited to, the development and adoption of AI technologies, cryptocurrency market volatility, regulatory developments, technical challenges in infrastructure deployment, and general economic conditions. AlphaTON undertakes no obligation to update any forward-looking statements, except as required by law.

Investor Relations:
AlphaTON Capital Corp
AlphaTON@icrinc.com
(203) 682-8200

Media Inquiries:
Richard Laermer
RLM PR
AlphaTON@rlmpr.com
(212) 741-5106 X 216

Richard Laermer
AlphaTON (at) rlmpr.com

FAQ

What did AlphaTON announce on November 26, 2025 regarding GPU infrastructure for ATON?

AlphaTON announced agreements to acquire and deploy a 1,000+ Nvidia B200 GPU cluster with ~$82.5M total capitalization funded by $30M equity and $52.5M debt.

How much Nvidia hardware will AlphaTON add to its balance sheet (ATON)?

AlphaTON said it will add approximately $70M of Nvidia hardware assets to the balance sheet.

What financing terms did AlphaTON disclose for the GPU project (ATON)?

The project is financed with $30M equity and a $52.5M debt facility that is fully amortized over 36 months.

What commercial use cases did AlphaTON list for the new GPU cluster (ATON)?

AlphaTON plans priority capacity for Telegram’s Cocoon AI and external GPU rentals via CUDO Compute’s marketplace.

What financial projections did AlphaTON publish for the GPU investment (ATON)?

Reported model metrics include a projected IRR 59.7%, NPV $59.6M, ROI 615%, and EBITDA margins of 64–73% over five years.

When did AlphaTON complete the Cocoon AI pilot and what is the next milestone (ATON)?

AlphaTON completed a Cocoon AI pilot on Nov 25, 2025; Q1 2026 milestones include hardware delivery, validation, and security audits.