★★★ Investor pitch · Series A open
Invest upstream of Europe's move to sovereign AI.
We own the whole chain. We source and train the talent, build AI companies with them, and run the strongest open-source AI on our own servers in France. State-funded training puts us in front of every enterprise before it chooses where its AI lives, then converts into recurring revenue from the data centers we own.
★★★ The shift
Regulation is pushing European enterprises off the US cloud, and AI compute is a high-margin business.
The Cloud Act and Patriot Act expose any data held by American providers to US authorities, wherever it sits. The EU AI Act adds obligations with penalties up to 7% of global turnover. European companies now need trusted, self-hosted AI, and dedicated AI infrastructure has proven to be one of the most profitable layers of the value chain. The demand exists. Reaching the client first, and owning the servers, is the hard part.
★★★ The model
One company owns the whole chain: the talent, the software and the servers.
We internalized the entire value chain rather than renting any part of it. Three assets we operate directly, each feeding the next, with the workforce paid for by public funding instead of our cash.
The School
Our own accredited training organization sources and trains AI talent, financed by France Travail and OPCO. Headcount becomes a profit center, not a payroll cost, and every trained enterprise becomes a hosting prospect.
The Studio
A start-up factory where that funded talent builds proprietary AI companies. We hold equity in each one and turn training into an asset that compounds.
The Data Centers
Sovereign AI hosting on French soil. The tenants are the enterprises we trained and the start-ups we built. Revenue is contractual, recurring, and at infrastructure margins.
Every part we own sends demand to the next. The servers are filled by clients we trained and companies we built, which is why our commercialization risk is resolved upstream, not downstream.
★★★ The enterprise funnel
Three steps to the client. The first two cost them nothing, and they hand us the account.
Every enterprise we audit and train becomes a qualified candidate for sovereign hosting before any competing infrastructure provider has met them.
Sovereignty audit
A dedicated session maps a company's exposure under the Cloud Act, Patriot Act and AI Act, with a written legal opinion from our French-licensed jurists. It costs the client nothing and opens the relationship.
Funded AI training
Teams are trained to deploy and govern a trusted AI system through our Qualiopi-certified courses, financed by public funding. Zero cost to the company. Months spent inside our ecosystem.
Sovereign hosting
The trained client deploys its own open-source AI in our French data centers. Data stays under French jurisdiction. Revenue becomes contractual and recurring.
★★★ The start-up studio
We don't just train students. We build companies with them, and keep a stake.
Our best talent builds AI start-ups inside the studio, in the model of Y Combinator, Hexa and eFounders. Because the workforce is funded by public schemes, we create equity without burning cash, and each company we build becomes a tenant of our data centers.
Build
Talent sourced through a France Travail work-placement period, then hired on work-study, develops proprietary AI products under our roof. State and OPCO funding cover the cost, so this R&D force does not weigh on the burn rate.
Keep equity
At the first outside round, the founding team takes the majority and we retain a strategic minority stake. Each cohort turns a training margin into a portfolio of equity positions.
Feed the data centers
Every studio company hosts its AI with us and grows into a recurring tenant. The demand we create for our own infrastructure, we also capture.
Recruiter margin, training margin, equity upside and hosting revenue, from the same person we were paid to source.
★★★ The moat
Assets a competitor cannot rent: funded people, owned distribution, owned curriculum.
Three proprietary layers keep our cost of growth structurally low and hard to copy.
A funded workforce
France Travail placement pay, OPCO funding and work-study state aid cover our people. Under-26 talent is cash-positive to onboard. Our sales and R&D capacity scales without the payroll burn that sinks classic start-ups.
Owned distribution
A network of 120+ proprietary websites ranks organically and brings enterprise and candidate leads every day. No dependence on paid ads, and a digital estate that does not depreciate.
Accreditation & curriculum
State-recognized (RNCP) programs and fully owned pedagogical IP give us direct access to public funding without third-party certifiers, and a balance-sheet asset in their own right.
★★★ The infrastructure
Our own sovereign AI data centers, built with a proven real-estate partner.
We develop capacity with Compagnie de Phalsbourg to own, rather than rent, the infrastructure our trained clients and studio companies will use.
InnovVal Data Hills — Aulnay-sous-Bois
Our first AI data center, developed with Compagnie de Phalsbourg on a 100+ hectare brownfield converted from the former PSA Citroën industrial site. Located inside the InnovVal economic hub in the Paris region, with a secured RTE 225 kV grid connection. Designed for AI workloads, GDPR and NIS2 compliant, on French soil.
View the InnovVal project →Ecotone — Southern France
A campus vision around the Ecotone project near Sophia-Antipolis, designed by Jean Nouvel. Our planned "Station F for AI" in the south, combining the business school, the start-up studio and on-site AI data-center capacity under one roof.
View the Ecotone project →Compagnie de Phalsbourg, founded by Philippe JOURNO, is our real-estate and infrastructure partner, bringing access to a portfolio approaching one million square meters to convert into sovereign AI data-center capacity. See compagniedephalsbourgia.com.
★★★ The stack
The strongest open AI model, running on silicon we own.
We don't rent a foreign API. We run GLM 5.2, today's leading open-weights model, on our own accelerators, so client data never leaves French jurisdiction and the margin on every token stays with us.
The model
GLM 5.2 is an MIT-licensed open model with 744 billion parameters and a one-million-token context, competitive with the leading proprietary models on public benchmarks. We hold the weights, not a license, and it runs with the same tooling enterprises already use.
The silicon
At production quality the model runs on a multi-accelerator node. The Series A funds that hardware. Through our Lenovo partnership we procure accelerators at competitive rates, and stay multi-vendor across Nvidia and alternatives such as Huawei Ascend to control both cost and supply.
The margin
Owning the model and the chips turns AI serving into an infrastructure business: fixed hardware, recurring tenants, and a cost per token we set ourselves instead of paying a US provider.
★★★ The plan
Build the talent-and-client engine first. Own the data centers second.
Accreditation is secured. The Series A funds the engine and the first compute, run in two execution phases; the Series B then scales the data-center capacity across France and Europe.
Scale the school, the studio and the pipeline
With Qualiopi accreditation in hand, we grow the funded cohorts, the sales engine and the first studio companies, and formalize the France business school. Each trained enterprise is a pre-sold hosting client. This phase de-risks everything that follows.
Acquire the hosting capacity
With demand demonstrated and a client base already inside the ecosystem, we buy the AI accelerators and finance dedicated data-center capacity at InnovVal with Compagnie de Phalsbourg, then convert trained enterprises and studio companies into recurring sovereign-hosting contracts.
★★★ The rounds
Enter at the Series A. Scale at the Series B.
Structured as a club deal within the AMF framework. The Series A is open now; the Series B follows once recurring hosting revenue is proven. Terms are indicative and open to discussion under NDA.
Series A
Build the whole chain: the funded engine and the first compute.
- Raising€15M+
- Pre-money valuation≈ €30M
- Equity offered≈ 33% (indicative)
Use of funds. Scale the Qualiopi funded-training pipeline and the first studio cohorts; hire across sales, delivery and legal; buy the AI accelerators to run GLM 5.2 in-house through our Lenovo partnership; open dedicated data-center capacity at InnovVal with Compagnie de Phalsbourg.
Series B
Scale the compute and the campus once the model is proven.
- Raising€50M+
- Pre-money valuation≈ €150M (indicative)
- Triggerproven recurring revenue
Use of funds. Scale the accelerator fleet and data-center capacity to the full InnovVal Data Hills build-out and the Ecotone campus; expand across Europe and through the Madagascar hub; grow the studio portfolio into a compounding book of equity.
Illustration only, and dependent on hitting the milestones that trigger the Series B. Valuations, round sizes and equity are indicative, pre-negotiation and subject to due diligence and definitive documentation; the Series A stake is diluted at the Series B as shown. Figures are not a forecast or a guarantee of returns. This page is a summary for discussion and does not constitute an offer to sell, or a solicitation to buy, securities, nor investment advice.
★★★ Business plan
Three revenue streams, one demand engine.
Our commercial model rests on three streams, all fed by the same proprietary demand engine.
Corporate AI training
Delivered to French enterprises via OPCO funding channels (AFDAS, Atlas, OPCO EP, OPCO 2i, OPCOMMERCE) and the EU FSE+ programme. Zero out-of-pocket cost for the client. High margin, no receivable risk since OPCOs settle direct with us.
AI data-center colocation
Recurring hosting revenue from the tranches co-owned at InnovVal Data Hills and Ecotone. Tenants are pre-sourced from our training portfolio, who migrate their AI workloads to sovereign French hosting once trained.
Start-up studio equity
Minority equity in AI ventures incubated at our facilities, harvested at exit. Portfolio companies also become anchor tenants of our own data centers, closing the loop.
The demand engine: a proprietary 120+ domain network optimised for OPCO-related keywords, paid search coverage at scale, and IA Afterwork Paris, the monthly gathering of 120+ AI decision-makers we co-organise.
★★★ Financial projections
Three-year path from ramp-up to €22M revenue and 45% EBITDA margin.
Indicative projections tied to the Series A milestones. Data-center revenue starts partial in 2027 and reaches full occupancy in 2028 with the tranche 1 activation.
€3.0M revenue
First tranche partially operational. Training scaled through OPCO pipeline.
- Training€1.2M
- Data-center colocation€1.8M
- EBITDA€0.9M · 30%
€11.2M revenue
Tranche 1 at full occupancy. First studio equity gains.
- Training€3.5M
- Data-center colocation€7.2M
- Studio equity€0.5M
- EBITDA€4.5M · 40%
€22.5M revenue
Tranche 2 partial. Africa deployment via Madagascar hub live.
- Training€7.0M
- Data-center colocation€14.0M
- Studio equity€1.5M
- EBITDA€10.1M · 45%
Underlying assumptions: data-center yield 15 to 20% operational, project IRR 18 to 22% over 10 years, training gross margin 65%, OPCO conversion rate 12% on qualified leads. Figures are indicative and subject to due diligence.
★★★ Use of proceeds
€15M mostly deployed into tangible AI infrastructure assets.
Two thirds of the Series A goes into real-estate and data-center equity tranches, not into intangible spend. The Series A is asset-backed by design.
★★★ The structure
Founder-owned today, restructured at closing into a French SAS that receives the round.
The group is currently organised around two entities, both fully controlled by the founder. At Series A closing, a new French operating company is incorporated to receive the funds and consolidate all commercial activities.
Founder-owned group
Franck PARIENTI holds 100% of the group. Zero external capital raised to date. Zero debt.
- BusinessDigital Holding (Madagascar)100% Franck PARIENTI
- Croissance Personnelle (French association)Qualiopi RNQ-25/10/2579
- Angels · VCs · convertiblesNone
- ESOPNot yet allocated
BusinessDigital AI SAS (France)
Newly incorporated French operating company receives the Series A investment and consolidates commercial activities.
- Founder67%
- Series A investors≈ 33%
- ESOP pool at closing10% (from founder shares)
- Qualiopi operationsLong-term services agreement with Croissance Personnelle
Restructuring plan financed by tranche 1 of the Series A: incorporation of BusinessDigital AI SAS, transfer of operational assets from BusinessDigital Holding Madagascar via contribution, execution of a services agreement with Croissance Personnelle for continued Qualiopi operations under French regulatory framework.
★★★ Series A terms
€15M in three tranches, tied to structuring and operational milestones.
Standard early-stage governance and preferred rights. Founder equity fully vested against Series A closing to align long-term interests.
€5M at signing
Released upon incorporation of BusinessDigital AI SAS, transfer of operational assets from BusinessDigital Holding Madagascar, and execution of the services agreement with Croissance Personnelle for Qualiopi operations.
€5M at milestone 1
Released upon executed co-investment agreement with Compagnie de Phalsbourg on Data Hills tranche 1 and confirmed OPCO training pipeline of €800K qualified opportunities.
€5M at milestone 2
Released upon operational launch of AI data-center tranche 1 at InnovVal and first paying colocation tenant migrated from the training portfolio.
- Board1 seat + 1 observer for the lead Series A investor
- Information rightsMonthly management reporting, quarterly financials, annual audit
- Liquidation preference1x non-participating
- Anti-dilutionWeighted-average, broad-based
- OtherROFR on future issuances, tag-along and drag-along at standard early-stage terms
- Founder vesting100% subject to 4-year vesting with 12-month cliff from closing
Terms are indicative, pre-negotiation and subject to due diligence and definitive documentation. Final terms will be captured in a term sheet and long-form investment agreement to be executed under NDA.
★★★ Regional expansion
Madagascar hub for Africa. Optional MENA structure for Gulf capital.
Two regional layers extend the platform beyond France while keeping the French operating entity as the group anchor.
Madagascar training centre
Locally recruited AI specialists trained on our proprietary curriculum deliver services at approximately 40% of French operational cost. The hub serves as the launchpad for francophone Africa (Côte d'Ivoire, Sénégal, Cameroun, RDC) and anglophone Africa via partner network. African corporate clients trained by the hub migrate their AI workloads to our sovereign French data centres, creating a defensible South-to-North data flow with EU-grade compliance.
Optional MENA holding (QFC)
BusinessDigital AI is open to structuring a Middle East and North Africa holding entity under Qatar Financial Centre jurisdiction, or equivalent Gulf regulatory framework, to co-develop the platform with regional strategic investors. QFC provides tax exemptions on cross-border activities, a dedicated regional deployment vehicle for GCC enterprise clients seeking sovereign AI hosting outside US Cloud Act exposure, and alignment with regional AI sovereignty visions.
The French operating entity BusinessDigital AI SAS remains the group anchor, licensed to any regional holding via a brand and methodology royalty agreement, ensuring French regulatory continuity while unlocking regional capital deployment.
★★★ The team
AI expertise, legal authority and certified training in one team.
A rare combination: the people who train the client, secure their compliance, and can host their AI are the same people.
Franck PARIENTI
Jurist by training, he leads the group's legal-sovereignty positioning and built the proprietary pedagogical engine behind the AI Business School, which trains through work-study. Founder of BusinessDigital.fr and co-founder of IA Afterwork, the monthly Paris gathering of 120+ AI decision-makers. He drives the training machine and the start-up studio.
Charlie EISSEN
Twenty years turning legacy information systems into real-time, AI-ready architectures, the foundation of a solid AI strategy. He works at both C-level, advising executive committees, and hands-on in delivery, with deep expertise in enterprise and cloud architecture, data streaming and agentic AI. Trained at Université Paris Descartes.
Board and advisors include a director at Google (name disclosed under NDA) and our infrastructure partner Compagnie de Phalsbourg, founded by Philippe JOURNO. Additional senior hires across sales, delivery and legal are planned immediately after the round.
★★★ Investor relations
Get in before the client pipeline is priced in.
We share the full financial model and business plan under NDA. The Series A is open now. Start with a conversation.