Issued by //gPUsh KoM Bond B.V.
Version: April 20, 2026
//gPUsh.ai is deploying 21 NVIDIA RTX PRO 6000 GPU servers (168 GPUs total) via a scalable GPU-as-a-Service model in a Tier-3 Swedish data center powered by sustainable energy. The Amortizing Bond refinances these cash-flow-generating assets, directly targeting surging AI infrastructure demand.
- AI market is booming: projected to reach $1.8 trillion by 2030 (36% annual growth)
- GPU market growing from $65B to $274B (2029)
- Europe investing €200B in AI with 4,000+ AI startups needing computing power
- Businesses shifting from buying hardware to flexible, on-demand access
- High costs: GPUs cost tens of thousands of dollars, creating barriers for startups
- Inefficiency: Companies that buy GPUs often leave them idle and underutilized
- Complexity: Building GPU infrastructure requires expertise, maintenance, and compliance
- Scalability issues: Hard to quickly scale up for short-term projects
- GPU-as-a-Service: Rent GPUs hourly or long-term without massive upfront investment
- Optimized resources: Pool GPUs across clients for maximum efficiency
- Instant scalability: Scale computing power up or down on demand
- Plug-and-play access: No infrastructure complexity—developers focus on innovation
- GPUs are optimized for LLM training, graphic creation and high-performance computing.
- The GPU-servers are operational and running on a GPU-as-a-Service model designed to meet critical market demand for high-performance computing.
- The platform is built on latest‑generation NVIDIA GPUs, providing the best available performance, efficiency, and scalability.
- GPUs are located in a datacenter in Sweden, powered by sustainable energy.
GPUaaS is a cloud-based model where users rent high-performance GPU compute power on-demand. This model generates revenues tied to occupancy and pricing, with downside protection from the physical assets.
- End-user needs computing power
- Accesses GPUs through AI cloud software partner
- AI model runs on GPU cluster
- Utilization is monitored
- Revenue flows to SPV
168x NVIDIA RTX PRO 6000 Blackwell GPUs in 21 servers with 8 GPUs each
Installed end of February 2026, generating revenue from April while ramping-up
Based in a state-of-the-art, Tier-3 (minimal downtime), datacenter in Sweden
5 years straight line depreciation
~6 years expected useful life
3 year warranty from NVIDIA
//gPUsh.ai is a European GPU-as-a-Service provider, specializing in on-demand NVIDIA GPU access for AI workloads, with operational GPU assets in data centers in Norway, Sweden, and Canada.
Offers investment products for investors looking to benefit from AI infrastructure, with asset ownership remaining with //gPUsh.ai.
Provides high-performance NVIDIA GPUs for AI training and deployment, eliminating Capex, delays, and technical complexity for end-users.
Operates proprietary GPU assets in data centers in Norway, Sweden, and Canada for reliable and scalable computing power.
Deploys GPUs through strategic partners, avoiding internal data center construction and intensive sales and service activities.
Collaborates with AI cloud providers and resellers serving over 10,000 customers (enterprises, SMEs, AI developers, research institutions).
$4.2M in operational GPU assets.
Scalable to €100M+ in GPU-server assets within 5 years by attracting investors and leveraging the asset base.
GPUs, especially powerful NVIDIA models, are essential for AI development but come with hefty price tags, often costing tens of thousands of dollars. This creates a significant financial barrier for AI developers and startups.
With limited budgets, organizations often prioritize hiring developers and investing in software over purchasing expensive hardware, leaving a gap in access to critical computing power.
Companies that purchase GPUs frequently face inefficiencies and often don't use their full capacity. Many GPUs sit idle for long periods, leading to wasted resources and inefficiency.
Start-ups may need access to more GPUs for short-term projects (e.g., training Large Language Models) but can't afford to scale up quickly because of long procurement times and high costs.
Building and maintaining GPU-powered infrastructure requires expertise, compliance, security, and ongoing maintenance, which can distract businesses from their core focus: developing AI solutions.
Model
//gPUsh.ai offers GPUs from hourly rental to multi-year contracts, eliminating massive upfront investments. This makes cutting-edge hardware accessible to businesses of all sizes.
Resources
By pooling GPUs and renting to multiple clients, //gPUsh.ai ensures powerful machines are utilized efficiently, reducing waste and maximizing usage.
Scalability
AI companies can instantly scale computing power up or down depending on needs, without worrying about procurement delays or capital expenditure.
Access
//gPUsh.ai provides a plug-and-play solution with no need to build complex infrastructure. Developers access GPU power with minimal setup, focusing entirely on innovation.
enabling them to innovate faster, cheaper, and smarter!
- Sourcing: GPUs are sourced from NVIDIA through a partner, who is an Elite Cloud Partner within the NVIDIA Partner Network. Hewlett Packard delivers rack-mounted servers with 8 GPUs each, while Cisco provides the network equipment that links servers internally and to the internet.
- Operationally Lean: GPUs are deployed through strategic partners, eliminating the need to build in-house data centre capabilities.
- Customer demand: The Company strategically partner with resellers that serve over 10,000 customers. These customers include large enterprises, AI developers at large corporations, SMEs, and research organisations that require compute to train or deploy modern AI models.
- Pricing model: Operate a flexible pricing model; offering end customers pay-as-you-go and contract-based options.
- Pipeline: The active pipeline consists primarily of qualified opportunities sourced through resellers and direct channels and assessed against customer size, expected GPU demand, and deployment timelines. Given that global demand for high-performance GPUs exceeds available supply, commercial demand is expected to increase strongly once additional capacity becomes available.
Rapid adoption of AI, cloud computing, and IoT is creating unprecedented demand for high-performance computing and storage capacity globally. GPUs are the backbone of AI development, with NVIDIA leading. Global GPU market expected to grow from $65 billion in 2025 to $274 billion by 2029.
Global AI market projected to grow at 36% CAGR, reaching over $1.8 trillion by 2030, fuelling massive GPU demand. Global data centre capacity is expected to grow at around an 18% CAGR to 2030, but this lags exponential demand growth, reinforcing a persistent scarcity dynamic that underpins long-term pricing power for GPU capacity.
Businesses increasingly prefer cloud-based and as-a-service models, choosing flexible on-demand GPU access over heavy upfront investments. GPUaaS offers flexibility and scalability for end users and can generate recurring, utility-like revenue streams.
With strict energy regulations and carbon reduction goals, European companies seek efficient, sustainable solutions like //gPUsh's eco-friendly data centers in Norway and Sweden.
European Union committed over €200 billion to AI funding (InvestAI), fostering a growing ecosystem. Europe's push to reduce US dependency creates opportunities for local players. Europe is home to 4,000+ AI start-ups and growing organizations developing AI tools. Many lack capital for high-end GPUs, making GPU-as-a-Service an ideal cost-effective solution.
Demand for GPU capacity and Data center space is outpacing supply, driven by AI adoption across multiple sectors. Constraints include suitable real estate, grid capacity, skilled labour, supply-chain and manufacturing bottlenecks, and regulatory requirements, making scarcity a durable feature of the market.
GPU clouds (“Neoclouds”) sit at the centre of the AI boom as the layer where end users rent GPUs to train models, process data, and run inference.
The market is structured around three groups:
- Hyperscalers (Amazon, Google, Microsoft): broad Infrastructure-as-a-Service/Platform-as-a-Service, strong enterprise tooling, premium pricing, vendor lock-in.
- Specialised GPU clouds: NVIDIA-focused, better price/performance, simpler pricing, developer-centric, but smaller scale.
- Emerging/distributed GPU networks: marketplace models, very low cost but higher variability in latency and support.
- //gPUsh.ai partners with a specialized GPU cloud provider with:
- Self-serve GPU cloud platform and managed orchestration for AI/Machine Learning/High Performance Compute workloads.
- NVIDIA GPU instances and transparent, usage-based pricing as a high-performance, cost-efficient alternative to hyperscalers.
- //gPUsh.ai provides the underlying GPU assets used on this platform, turning AI demand into asset-backed cash flows for investors.
- European AI cloud positioning: Focus on EU workloads, data residency and regulatory alignment for AI and data-intensive use cases.
- Sustainability: Exclusive partnerships with 100% renewable-powered data centers, supporting ESG mandates and net‑zero narratives.
- Democratised, self-service access: Direct-to-compute GPU marketplace with pre-configured "flavours", enabling quick deployment without heavy DevOps.
- Cost-efficiency: Up to 5x lower GPU pricing vs hyperscalers, with no hidden fees and transparent per-usage billing.
The Neocloud market is the transaction layer of the AI boom, where developers and enterprises rent GPUs to train and deploy models. //gPUsh.ai provides the physical NVIDIA GPU assets powering a specialised, EU-centric, 100% renewable GPU cloud partner that offers self‑serve, developer-friendly GPU-as-a-Service with transparent, usage-based pricing up to 5x cheaper than hyperscalers, giving investors asset-backed exposure to a differentiated European AI infrastructure play.
We currently manage 280 high-performance Blackwell GPUs, comprising 56 NVIDIA RTX 5090s and 224 NVIDIA RTX PRO 6000s, representing a combined book value of $4.2 million. A total of $5 million has been committed by shareholders in order to raise $20 million of debt.
Our GPUs are housed in state-of-the-art data centers in Norway, Sweden and Canada. These locations were carefully selected for their advanced infrastructure, low energy costs, and access to sustainable energy sources, critical for optimizing operational efficiency and reducing costs.
Our GPUs are monitored in near real-time through a proprietary, in-house developed dashboard. This ensures efficient resource allocation, proactive maintenance, and proper reporting across all infrastructure.
The projected payback period for our GPU assets is approximately three years. Our occupancy ratio currently exceeds the projection, underscoring strong market demand. Revenues are collected monthly with end-users paying upfront, ensuring predictable cash flow and minimal credit exposure.
With global AI adoption accelerating, demand for GPU-based computing power is skyrocketing. This presents a clear opportunity for //gPUsh.ai to scale beyond €100 million in GPUs under management, positioning as a key enabler of AI innovation in Europe.
//gPUsh KoM Bond B.V. offers a 42‑months amortizing bond to acquire 21 operational NVIDIA RTX PRO 6000 Blackwell GPU servers located in a data center in Sweden. The Amortizing Bond combines linear principal repayments with a Fixed Coupon and Performance-based Coupon, targeting a 14% IRR.
€2,100,000 inaugural bond issue
€100,000
81% bond issue, 19% equity injected by //gPUsh Services B.V.
Linear semi-annual repayments
Guarantee by parent company //gPUsh Services B.V.
42-months investment
14% target IRR, through 7% Fixed Coupon and maximum 7% Performance-based Coupon
Including Performance Coupon, depending on performance
Expected return
Net Revenue: Gross GPU-as-a-Service Revenue minus 20% service fee
Verification: Daily reporting through https://my.gPUsh.ai portal
Payout: Performance Coupon to be paid on a yearly basis, based on the performance of the previous period
Performance Coupon Formula: If Net Revenue falls within the Conservative Case row, the Performance Coupon is paid out on a pro-rata basis calculated as: (Net Revenue – Downside Case) / (Base Case – Downside Case) x 7%
7.0% Fixed + 7.0% Performance Coupon
€100,000
42-months
14% IRR
€128,000
128.0% — MOIC 1.28x
GPU asset depreciation profile and GPU/hr price degradation over time.
This bond investment offers multiple layers of protection and favorable risk characteristics tailored for qualified investors.
Corporate guarantee by //gPUsh Services B.V.
14% IRR based on 7.0% Performance-based Coupon and 7.0% Fixed Coupon (fixed, regardless of performance).
GPU servers are operational and revenue-generating, eliminating supply-chain risks and ramp-up delays.
Semi-annual linear repayments reduce credit risk and structurally decrease principal exposure.
42-months tenor reduces exposure to long-term technology obsolescence and GPU depreciation cycles.
3-year payback period on GPU asset investments with occupancy currently trading above expectations.
19% equity contribution from //gPUsh Services B.V. acting as buffer for bond holders and shows skin-in-the-game.
Steady semi-annual cash flows from interest and principal repayments enhance investor liquidity.
//gPUsh Ventures B.V. is responsible for holding IP and overseeing project management.
//gPUsh Services B.V. manages the assets and finances of the SPVs.
Key responsibilities include:
//gPUsh KoM Bond B.V. is the issuing entity for the Bond Loan, remains the owner of the GPUs and holds contracts with all relevant stakeholders. Activities include accounting and tax filing, interest payments & repayments to bondholders.
GPU
- 6-years GPU-lifetime
- 5-years depreciation schedule
- Base pricelist
- 10% annual GPU/hr. price degradation (from Y2)
- 80% occupancy ratio
Debt
- 42-months tenor
- 81%/19% — debt/equity ratio
- Semi-Annual 7% Fixed Coupon
- Annual 7% Performance-based Coupon
- Linear repayment
Latencies
- 1 month receivables delay
- No utility invoice latency
- No revenue ramp-up
Other
- 0.8682 Euro/Dollar exchange rate
- 10% Management Fee
- Gross Revenue is based on GPU price per hour × 80% occupancy × 720 hours × 168 GPUs = Gross Portfolio Revenue per month.
- Opex includes a 20% Service fee to our GPU Cloud Partner, 10% Management fee for //gPUsh Services B.V., and $126,000 annual Utility fee ($6,000 per server/year).
Being able to invest from €100,000 benefitting from massive market trends. Global AI market growth of 36% CAGR to $1.8T by 2030. GPU demand outpacing supply. EU committed €200B+ to AI funding with 4,000+ AI startups creating unprecedented demand for compute access.
Payback Period: ~3 years on GPU asset investments.
Occupancy Ratio: Currently exceeds projections, indicating strong market demand .
Cash Flow: Monthly upfront payments from end-users = predictable, low creditor risk.
Revenue per GPU: €1.45 GPU/hr × 80% occupancy × 720 hours × 168 GPUs = €140,314 Gross Portfolio Revenue / month.
21 revenue-generating GPU servers (168× NVIDIA RTX PRO 6000 Blackwell GPUs). 81% debt / 19% equity structure demonstrates skin-in-the-game from //gPUsh Services B.V.
Semi-annual linear principal repayments reduce credit risk. Tenor remains within GPU depreciation cycle (3.5-year tenor vs. 5-year depreciation). Debt Service Coverage Ratio (DSCR) averages 1.16x over the first 3 years, well above 1.0 threshold.
EU-regulated operations serving the European AI ecosystem. State-of-the-art datacenter in Sweden (Tier-3) powered by sustainable energy. Reduces EU dependency on US-based GPU providers.
Qualified investors are invited to participate in this unique opportunity to invest in GPU-servers through the Amortizing Bond.
Contact our team to discuss the investment opportunity and conduct due diligence.
Carefully review all bond conditions, subscription forms, and supporting materials.
Complete and submit your subscription form before April 24, 2026.
Transfer funds before April 25, 2026 and start earning returns on your investment.
For questions or to schedule a consultation, please contact any member of the //gPUsh.ai team using the contact information provided.
//gPUsh Ventures B.V.
Fred. Roeskestraat 100
1076 ED Amsterdam
The Netherlands
www.gPUsh.ai
Powering the
AI revolution,
one GPU at a time.
This Memorandum has been prepared by //gPUsh Services B.V. ("gPUsh" or "the Company") based on information furnished by its management and employees. This Memorandum relates to the issue of short term fixed rate bills by the Company ("Bond Issue") and contains confidential information regarding the business of the Company. Every Recipient of this Memorandum is bound by confidentiality in respect of all information contained herein. This Memorandum is only being made available to a selected group of investors who are deemed sufficiently expert and / or sufficiently experienced to understand the aspects and risks involved in investing in general, and specifically a financial instrument like a bond, and who will obtain expert advice where and when needed (“Recipients”). The information contained herein has been prepared to assist prospective investors in making their own evaluation of the Company and the Bond issuance, and does not purport to be all-inclusive or to contain all of the information that a prospective investor may desire. In all cases, prospective investors should conduct their own investigation and analysis of the information and data set forth in this Memorandum and satisfy themselves as to the accuracy, reliability and completeness of such information and data.
The Company makes no representation as to the accuracy or completeness of the information in this Memorandum or any other information made available to Recipients or prospective investors. In particular, no representation or warranty is made as to the achievement or reasonableness of any future projections, management estimates, prospects, returns or market data contained herein. Statements contained in this Memorandum are made in good faith and have been derived from information believed to be reliable as of the date of this Memorandum. Subject to any law to the contrary, and to the maximum extent permitted by law, the Company disclaims and excludes all liability for any loss or damage (whether foreseeable or not) suffered by any person or entity acting on, or refraining from acting because of, anything contained in or omitted from this Memorandum, whether the loss or damage arises in connection with any negligence, default, lack of care or misrepresentation on the part of the Company, its subsidiaries or their Associates or any other cause.
Each Recipient and prospective Investor agrees that it shall not seek to sue or hold the Company, and / or subsidiaries or the Associates so liable in any respect for the provision of this Memorandum and the information contained herein. Nor the Company, nor their Associates accept any responsibility to inform Recipients or prospective investors of any matter arising or coming to any of their notice which may affect any matter referred to in this Memorandum (including but not limited to any error or omission which may become apparent after this Memorandum has been issued). This Memorandum shall not be deemed an indication of the state of affairs of the Company nor shall it constitute an indication that there has been no change in the business or affairs of the Company since the date of this Memorandum or since the date at which any information contained herein is expressed to be stated. If further information in connection with a possible transaction in relation to the Company is provided by the Company, //gPUsh or their Associates or any other person or entity, Recipients and prospective investors acknowledge receipt of such information as though it formed part of this Memorandum.
//gPUsh will arrange for appropriate due diligence by Prospective Investors. In furnishing this Memorandum, //gPUsh undertakes no obligation to provide Recipients or Prospective Investors with access to any additional information. //gPUsh reserves the right, to its sole discretion, to negotiate with one or more Prospective Investors at any time and to enter into a definitive sale and purchase agreement regarding the Company at any time without stating its reasons and without prior notice to any Prospective purchaser. Also, //gPUsh reserves the right to terminate, at any time, further participation in the investigation and proposal process by any party and to modify the procedures without assigning any reason therefore.
The Company intends to operate in the ordinary manner during the evaluation period; however, the Company reserves the right to take any action, whether in or out of the ordinary course of business, which it may deem necessary or prudent in the normal conduct of such business. This Memorandum nor any other information made available to any Recipient, Prospective Investor or its advisors is a binding commitment nor will it form the basis of any contract. A proposal regarding the Company will only give rise to any contractual obligations on the part of the Company upon execution of a definitive agreement. Only those obligations which may be set forth in such definitive agreement will be accepted. This Memorandum does not constitute an offer or invitation for the sale or purchase of securities or any of the businesses or assets described in it in any jurisdiction and no decision has been taken whether and, if so, which of such securities will be offered. This Memorandum does not constitute any form of commitment or recommendation on the part of //gPUsh or any of their respective subsidiaries or associated companies nor does it grant any of the Recipients exclusivity.
This Memorandum shall be governed by the laws of The Netherlands. The distribution of this Memorandum in certain jurisdictions may be restricted by law and, accordingly, Recipients of this Memorandum represent that they are able to receive this Memorandum without contravention of any unfulfilled registration requirements or other legal restrictions in the jurisdiction in which they reside or conduct business. //gPUsh shall not be liable for any violation by any party of such restrictions and limitations.
— By invitation only —