AKAWAN NEWSLETTER
Before leaving, let's stay in touch.
Once a month, receive practical insights from our AI and digital experts - covering use cases, real-world experiences, and trends - with no jargon, just useful ideas.
Your address remains confidential. You can unsubscribe at any time with a single click.
Data centralization
"Centralizing all your data is the future." This statement is currently the greatest strategic risk for your industry.
The future does not lie in centralization. It lies in controlled decentralization.
On September 12, 2025, the European Data Act came into force. Many read it as just another regulatory constraint. It is exactly the opposite: this is the first time that Europe officially asserts that your industrial data belongs to you, that you can retrieve it, share it, circulate it, and that no one can make that decision for you.
A few figures to measure what is at stake:
€829 billion: economic value of data produced in the EU by 2025;
+530%: growth in global data volume between 2018 and 2025;
95%: enterprise AI projects that never reach production;
58% of SME-ETI executives believe that AI is a medium-term survival issue;
26% of SME-ETIs are already using at least one AI solution daily as of March 2026, a doubling in two years.
This data does not come from nowhere. It comes from your machines, your sensors, your PLCs, your production lines. It comes from you.
The real question is not: is this data valuable?
The real question is: who benefits from it?
829 billion euros. Your resource. Their infrastructure.
Today, your data goes to AWS, Azure, or Google Cloud. It is processed under extraterritorial laws, with the US Cloud Act allowing authorities to access it without even informing you.
It then feeds artificial intelligence models that you will purchase 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 to your disadvantage.
Every AI tool activated in your company consumes your operational data, sends it to third-party servers for inference, and stores it in logs that you do not control.
The MIT study is clear: if 95% of AI projects fail, it is almost never for technological reasons. It is architectural: poorly managed data, scattered across third-party infrastructures, without traceability, without governance, without a defined lifecycle.
AI only worsens what you have not resolved beforehand.
Three dimensions treated separately, even though they are linked.
The AI Act, which partially entered into force in February 2025, now requires companies to ensure that their teams understand the AI systems they use.
This is not anecdotal: it is the legal recognition that deploying AI without mastering the data it consumes constitutes a compliance risk, not just a technical risk.
The problem with three dimensions, almost always treated in silos:
Accessibility: who can access what data, from which system, under what conditions?
Usage: for what purpose is this data processed, by which algorithm, with what traceability?
Lifecycle: how long does it exist, where, and who decides on its deletion?
No American cloud provider guarantees you control over these three dimensions. This 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 such as EDF, Airbus, Renault, Engie, and BMW have partnered on projects built on GAIA-X standards: Catena-X for automotive, Nuclear-X for nuclear, and Data4Industry-X for manufacturing. The return of Donald Trump has accelerated things. "The new geopolitical reality reminds us every day of the importance of protecting our digital sovereignty," points out the sub-director of 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 the Institut Mines Télécom: to move from a think tank to a do tank. In December 2025, GAIA-X presented its first catalog of operational services, featuring 600 certified services from 15 providers.
The principle of a dataspace is simple: to share data between stakeholders in an industry without losing control of it, within a trusted framework, and without a centralized infrastructure subject to external legislation. A data space relies on technical building blocks, servers, processing tools, and connectors. However, if these blocks are provided by actors subject to the Cloud Act, the sovereignty of the data space is an illusion.
These dataspaces cover fields ranging from the automotive industry 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 industries are now positioning themselves.
SMEs, Mid-caps: you are not off-topic. You are out of time.
The temptation is to think that these issues only concern Airbus or EDF, and not an industrial mid-cap with 300 employees or a subcontracting SME. This is a strategic misreading.
GAIA-X explicitly aims to democratize access to advanced tools for European SMEs: AI as a service, secure storage, and collaborative computing, while maintaining control over data.
But this integration does not happen automatically. It assumes that you come to the table with a decentralized architecture that allows you to share what you want to share, keep what you want to keep, and track every single access.
According to Bpifrance Le Lab, 43% of SME and mid-cap leaders have already implemented an AI strategy. This means that 57% are still navigating blindly, testing tools without an overall vision. In most cases, these tools consume data and no one knows exactly where it goes, or how long it remains accessible to third parties.
But let's be clear: entering a dataspace is currently a technical wall.
EDC (Eclipse Dataspace Connector) connectors, verifiable credentials (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, this is a structuring project. For a subcontracting SME or mid-cap, it is an insurmountable wall to face alone.
This is 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 ground up for demanding industrial environments: aerospace, defense, manufacturing, logistics, and the sovereign public sector.
What the platform delivers right now, in production with our first clients:
Sovereign M2M communication between machines, PLCs, sensors, and industrial systems, without dependence on an hyperscaler;
Execution close to the data: processing is done where the data is produced, not in a remote cloud;
Zero infrastructure modification: KaptnGO is deployed without redesigning your existing IT or OT networks;
Dynamic automation of machine-to-machine flows, without an intrusive agent on the equipment side;
Integrated LicenGuard module, for the 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, software licenses, and execution perimeter without discarding your existing infrastructure.
KaptnGO in 2027: the "Dataspace Compliant" roadmap
The next step, already underway in R&D, has a clear objective: to enable an SME or mid-cap to join an industry dataspace without bearing the technical complexity alone.
What we will add by 2027:
Native dataspace connector, compliant with IDSA specifications and compatible with the Eclipse Dataspace Connector (EDC)
Verifiable credentials management (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 registration in the certified services catalog
Fine-grained governance of usage contracts (Usage Policies): who can do what with the data, for how long, and under what conditions
End-to-end traceability of access and processing, aligned with Data Act and AI Act requirements
Optional EBSI notarization for exchanges requiring proof of European legal value
Simplified onboarding in industry dataspaces, with pre-configured templates for aerospace, automotive, and manufacturing
The guiding idea remains the same: the client keeps their data locally. Execution remains close to the source. Sharing becomes a sovereign choice, no longer a leak by default.
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 embeds them, updates them, and guarantees their coherence 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 is whether you will get there with your own architecture, or with someone else's.
The demonstrator who has never seen a cadence
If your AI pilots are not making it to production, the problem is not the AI
How many open houses in your workshop?
No one in the company knows the exact number…
The digital health record you don't have
The digital health record that your industry has never opened.
Multi-sites, zero visibility...
The nightmare of blind industrial management.
Industrial M&A: The Billion-Dollar Bug
IT architecture secretly dictates the success of your mergers
Successful merger, ERP in ruins
The true price of your redemptions…
Why 80% of factory mergers verge on a digital crash
and how to escape it…
The deadly trap of industrial hyper-growth
+28% to +66% in cadence…
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?
