Data centralization
"Centralizing all your data is the future." This claim is now the biggest strategic risk for your industry.
The future is not centralization. It is controlled decentralization.
On September 12, 2025, the European Data Act came into force. Many read it as just another regulatory burden. It is exactly the opposite: for the first time, Europe officially states that your industrial data belongs to you, that you can retrieve it, share it, circulate it, and that no one can decide for you.
A few figures to measure the stakes:
€829 billion: the economic value of data produced in the EU by 2025;
+530%: growth in the global volume of data between 2018 and 2025;
95%: AI projects in companies that never reach production;
58% of SME and mid-cap company executives believe AI is a medium-term survival issue;
26% of SME and mid-cap companies already use at least one AI solution on a daily basis as of March 2026, double the figure from two years earlier.
These data do not come out of nowhere. They come from your machines, your sensors, your controllers, your production lines. They come from you.
The real question is not: do these data have value?
The real question is: who benefits from them?
€829 billion. Your resource. Their infrastructure.
Today, your data go to AWS, Azure or Google Cloud. They are processed under extraterritorial legislation, with the Cloud Act allowing American authorities to access them without even informing you.
They then feed artificial intelligence models that you will buy as SaaS. And when you ask the contractual question "who can use my usage data to train their models?", most contracts are either silent or drafted against your interests.
Each AI tool activated in your company consumes your operational data, sends them to third-party servers for inference, and stores them in logs that you do not control.
The MIT study is unequivocal: if 95% of AI projects fail, it is almost never for technological reasons. It is architectural: poorly controlled data, dispersed across third-party infrastructures, with no traceability, no governance, no defined lifecycle.
AI only makes worse what you have not fixed upstream.
Three dimensions handled separately, even though they are linked.
The AI Act, which came into partial force in February 2025, now requires companies to ensure their teams understand the AI systems they use.
This is not trivial: it is legal recognition that deploying AI without mastering the data it consumes is a compliance risk, not just a technical risk.
The three-dimensional problem, almost always handled in silos:
Accessibility: who can access which data, from which system, under what conditions?
Usage: for what purpose are these data processed, by which algorithm, with what traceability?
Lifecycle: how long do they exist, where, and who decides their deletion?
No American cloud provider guarantees control over these three dimensions. It is not a value judgment. It is a structural constraint linked to their business model.
What Europe has started to build: decentralization.
Leading industrial players EDF, Airbus, Renault, Engie, BMW have joined projects built on the GAIA-X standards: Catena-X for automotive, Nuclear-X for nuclear, Data4Industry-X for manufacturing. Donald Trump's return has accelerated things. "The new geopolitical landscape reminds us every day of the importance of protecting our digital sovereignty," notes the deputy director for space and software at the Directorate General for Enterprises.
The France GAIA-X hub held its plenary session in March 2025 at Bercy with an explicit mandate from Institut Mines Télécom: move from think tank to do tank. In December 2025, GAIA-X presented its first catalog of operational services, with 600 certified services from 15 providers.
The principle of a dataspace is simple: share data between actors in a sector without losing control of it, in a trusted framework, without centralized infrastructure subject to external laws. A data space relies on technical building blocks, servers, processing tools, connectors. However, if these building blocks are supplied by actors subject to the Cloud Act, the sovereignty of the data space is illusory.
These dataspaces cover areas ranging from automotive to aerospace or manufacturing, creating data exchange platforms based on transparency, trust and openness. These are no longer experiments. They are operational infrastructures in which sectors are now positioning themselves.
SMEs, mid-sized companies: you are not off topic. You are out of time.
The temptation is to think that these issues concern only Airbus or EDF, not an industrial mid-sized company of 300 people or a subcontracting SME. That's a strategic misreading.
GAIA-X explicitly aims to democratize access to advanced tools for European SMEs: AI as a service, secure storage, collaborative computing, while keeping control of the data.
But this integration does not happen automatically. It requires that you come to the table with a decentralized architecture that lets you share what you want to share, keep what you want to keep, and track every access.
According to Bpifrance Le Lab, 43% of SME and mid-sized company executives have already implemented an AI strategy. This means that 57% are still navigating by sight, testing tools without an overall vision. In most cases, these tools consume data whose destination no one knows exactly, nor how long they remain accessible to third parties.
But let's be clear: entering a dataspace is, today, a technical wall.
EDC connectors (Eclipse Dataspace Connector), verifiable identities (SSI / W3C VC), GAIA-X compliance, fine-grained access governance, end-to-end traceability, OIDC4VC (OpenID Connect for Verifiable Credentials) integration, EBSI notarization...
For a large group with 50 architects, it's a structuring project. For an SME or subcontracting mid-sized company, it's an impassable wall on its own.
That's precisely the problem we decided to tackle with KaptnGO.
KaptnGO today: the sovereign foundation already operational.
KaptnGO is a sovereign M2M communication platform, designed from the outset for demanding industrial environments: aerospace, defense, manufacturing, logistics, sovereign public sector.
What the platform delivers right now, in production at our first customers:
Sovereign M2M communication between machines, PLCs, sensors and industrial systems, without dependence on a hyperscaler;
Execution close to the data: processing happens where the data is produced, not in a remote cloud;
Zero infrastructure modification: KaptnGO is deployed without redesigning your SI or your existing OT networks;
Dynamic automation of machine-to-machine flows, without an intrusive agent on the equipment side;
Integrated LicenGuard module, for governance and usage control of industrial software licenses (FlexLM, DSLS, RLM, Sentinel RMS, LM-X);
Native security posture: where your ZTNA (Zero Trust Network Access) stops, KaptnGO begins.
Concretely, you can already regain control of your M2M flows, your software licenses, and your execution perimeter without throwing away your existing infrastructure.
KaptnGO in 2027: the roadmap "Dataspace Compliant"
The next step, already underway in R&D, aims for a clear goal: allow an SME or mid-sized company to join a sector dataspace without bearing the technical complexity alone.
What we will add by 2027:
Native dataspace connector, compliant with IDSA specifications and compatible with Eclipse Dataspace Connector (EDC)
Management of verifiable identities (SSI) based on W3C Verifiable Credentials standards, to authenticate each participant in the exchange
OIDC4VC / OIDC4VP integration for issuing and presenting credentials in dataspace flows
GAIA-X compliance with listing in the catalog of certified services
Fine-grained governance of usage contracts (Usage Policies): who can do what with the data, for how long, under what conditions
End-to-end traceability of accesses and processing, aligned with the requirements of the Data Act and the AI Act
Optional EBSI notarization for exchanges requiring evidence with European legal value
Simplified onboarding into sector dataspaces, with preconfigured templates for aerospace, automotive and manufacturing
The guiding idea remains the same: the customer keeps their data with them. Execution stays close to the source. Sharing becomes a sovereign choice, no longer a default leak.
The difference with a "classic integrator" approach: you do not have to assemble the EDC, SSI, OIDC4VC, Usage Control and GAIA-X building blocks yourself. KaptnGO includes them, updates them, and guarantees their consistency over time.
Your industrial data is worth 829 billion at the European scale. The question is not whether you will join the dataspace economy. It's whether you will do so with your own architecture, or with someone else's.
Your digital transformation has everything except results.
Your software still doesn't talk to your machines…
53% of MCP servers are hacker nests
Have you given the keys to your company to the whole world?
Heathrow, Brussels, and Berlin paralyzed.
The cause: two unchanged identifiers…
DeepSeek: the app that devours your company...
You hire quickly, your teams adopt AI even faster…
Don't your machines communicate with your information system?
The real cost of industrial "silence"…
Tech in Industry
Facing the challenges of disconnected environments…
