Loading
Loading
An ambitious but disciplined path from a Romanian software company to a regional AI-infrastructure and platform business serving the European sovereign-AI market.
NOVACORE AI develops secure software, artificial intelligence, and high-performance computing infrastructure that enables European enterprises, governments, and research institutions to innovate with AI while retaining full sovereignty over their data, models, and operational environments.
The convergence of three structural tailwinds creates a generational opportunity for an infrastructure-first AI company anchored in Southeast Europe.
| Market segment | TAM estimate | Serviceable opportunity | Maturity |
|---|---|---|---|
| Enterprise AI services (consulting, integration) | €700M (RO) | €140M | Late growth |
| Private managed AI platform (SaaS) | €2.5B (CEE) | €250M | Early growth |
| GPU rental / bare-metal capacity | €3.8B (SE Europe) | €190M | Nascent |
| Multi-tenant GPU cloud | €8B (EU) | €400M | Nascent |
| Specialised Romanian NLP models | €300M (RO) | €60M | Emerging |
| AI-enabled data-centre colocation | €12B (EU edge) | €240M | Nascent |
| Sovereign AI infrastructure (govt. & defence) | €5B (EU) | €150M | Early growth |
NOVACORE's structural advantages in the sovereign-AI market stem from geography, regulatory alignment, and a capital-disciplined build-up strategy.
| Capital category | Risk profile | Typical financing instruments | Estimated range |
|---|---|---|---|
| Software working capital | Low | Revenue, equity, bank credit lines | €200K – €500K |
| AI research & model development | Medium | Equity, EU research grants, academic partnerships | €500K – €2M |
| GPU equipment procurement | Medium | Leasing, vendor financing, pre-commitment contracts | €3M – €15M |
| Multi-tenant GPU cloud launch | Medium-High | Equity, strategic investors, infrastructure funds | €10M – €40M |
| Data-centre site & shell development | High | Project finance, real estate, strategic partners | €20M – €80M |
| Data-centre fit-out & commissioning | High | Project debt, equipment asset-backed lending | €15M – €50M |
| Operating reserves & contingency | Low | Retained earnings, standby equity facilities | €1M – €3M |
Capital is raised in phases matched to de-risked milestones. Software and services generate positive cash flow before material infrastructure capex is committed. GPU capacity is expanded only against contracted or highly probable demand — not speculatively. This sequencing contains downside while preserving upside exposure to infrastructure-scale returns.
Revenue generation and customer relationships. Custom software development, cloud consulting, and AI readiness assessments for corporate and public-sector clients. Positive unit economics and cash-flow generative from inception.
Selection of 3–5 enterprise or government clients for private AI deployments. Validated demand signals, technical reference cases, and operational playbook development. De-risks the managed-platform investment decision.
Multi-tenant managed platform for fine-tuning, inference, and RAG workflows. Recurring SaaS revenue with usage-based pricing. Target: 20+ enterprise tenants with average contract values above €50K ARR.
Bare-metal GPU capacity leased to enterprises with committed take-or-pay contracts. High capital efficiency: customer pre-commitments offset procurement cost. Entry-size cluster of 32–64 H100-equivalent GPUs.
Scaled, metered GPU infrastructure with orchestration layer. Self-serve provisioning for AI training and inference workloads. Target: 500+ GPUs under management with utilisation above 70%.
Proprietary NLP, speech, and computer-vision models optimised for Romanian language and public-sector use cases. Differentiated IP with licensing revenue and platform integration.
Purpose-built or colocated data-centre capacity optimised for AI workloads (high-density power, liquid cooling). Foundation for scaled GPU cloud and sovereign-AI colocation services.
Large-scale model training only after infrastructure, revenue, and differentiated data assets are established. This is an option, not a necessity — the platform business can thrive as an infrastructure and fine-tuning layer.
Investors should evaluate each risk in the context of the capital discipline and phased execution strategy described above.
NOVACORE's operating model is structured to retain viability across the full range of plausible outcomes. Software services produce cash from day one. GPU and data-centre investments are modular — each tranche requires demand validation before the next is committed. If sovereign-AI adoption underperforms expectations, the company scales back infrastructure capex and operates profitably as an enterprise AI services and software firm. If adoption meets or exceeds projections, the company captures infrastructure-scale economics with a lower entry cost base than Western European competitors. Large-scale AI-cloud businesses illustrate both the transformative upside and the existential risks of overcapitalised, demand-agnostic expansion — NOVACORE plans accordingly.
Secure AI and high-performance computing for enterprises, governments and research.