AI - OUTSCALE blog

Sovereign AI and sovereign cloud: how can public-sector projects be accelerated?

OUTSCALE — Tue, 29 Sep 2026 12:56:55 +0000

At OUTSCALE EXPERIENCES 2026, Mireille Mege, Public Sector Customer Success Manager at OUTSCALE (Dassault Systèmes), and Clara Pascolo, Chief Operating Officer at SMLB Next, explained how Nuage Public simplifies public-sector access to sovereign cloud, drawing on its partnership with OUTSCALE and the generative AI bundles from Mistral available on the OUTSCALE Marketplace.

For public administrations and parapublic organizations, public cloud provides a practical entry point to secure, compliant and scalable environments capable of adapting to business needs.

The success of these projects relies on a structured approach: identifying a relevant initial use case, ensuring the protection of sensitive data, selecting the appropriate level of support and anticipating industrialization from the earliest stages.

Nuage Public as a starting point

The UGAP Nuage Public marketplace provides access to cloud services for more than 26,000 public- and parapublic-sector beneficiaries. It enables organizations to define their needs, specify their technical and regulatory requirements, and then be directed to cloud providers capable of meeting them.

This approach facilitates the launch of artificial-intelligence projects, often in the form of a proof of concept. It responds to two main trends. Some organizations want to implement the Cloud au Centre doctrine promoted by DINUM. Others want to use AI in a more secure and controlled environment.

The Nuage Public marketplace combines several factors that are important for public procurement: negotiated prices, a simplified process, guidance toward relevant offerings and shorter deployment times. In some cases, a project can start in less than a month, while the purchase order can be processed very quickly when the requirements are clearly defined.

Public-sector priorities

Organizations launching a sovereign-AI project generally have three main expectations.

Controlling data location

The first question concerns where data is located and how it is processed. Users want to prevent sensitive data from being transferred to environments they do not control. This concern applies in particular to administrative data, health information, identity documents, financial data and defense-related information.

Sovereignty therefore becomes a central decision criterion. Data must be processed and stored in an environment that meets the organization’s requirements, with particular attention paid to access, permissions and the platform’s operating conditions.

Reducing shadow-AI risks

Employees may be tempted to use AI services directly available online, particularly to query language models with professional information. This practice creates a risk of losing control over the data entered and where it is sent.

A sovereign-AI environment provides users with a controlled framework. The organization can define authorized services, manage access and encourage the use of models hosted in an environment that meets its security requirements.

Preparing for industrial deployment

A project often begins with a limited use case. Once the initial results are positive, other needs emerge. The organization may then want to analyze more documents, automate additional steps or use several models.

The selected architecture must therefore be able to evolve. A scalable offering facilitates the transition from proof of concept to production and then to multiple use cases. This industrialization capability must be considered from the outset to avoid having to redesign the entire technical environment later.

Three AI project offerings

Public-sector organizations can choose between several approaches. The right option depends on the desired level of control, the number of models required, the available resources and the expected level of support.

Consuming compute resources

The first use case consists of providing the resources required to run language models. The organization consumes compute including CPU, RAM and GPU and then uses the models at its disposal.

This approach offers significant technical freedom. It is suitable for teams capable of deploying and administering their own environment. However, they must manage several components themselves, including resource configuration, Nvidia drivers and the components required to run the models.

Deploying a Mistral bundle

The second use case is based on Mistral bundles. This offering provides a ready-to-use solution, with a Mistral model deployed in a tenant dedicated to the customer. Deployment takes place through the Marketplace and can be completed quickly after the order has been placed.

The customer receives a directly accessible endpoint to begin its work. Mistral models are supplied with their licenses, and the environment is dedicated to the organization. Data therefore remains within the designated environment in France, in accordance with the characteristics of the offering.

The bundles offer unlimited token consumption. However, capacity remains determined by the available hardware resources, particularly the number of GPUs associated with the model and the use case. Depending on requirements, one or two GPUs may be offered.

This option is especially suitable for organizations that want to use a language model without managing the entire technical stack themselves. It reduces initial complexity and accelerates access to an operational solution.

Building a Mistral AI Studio platform

The Mistral AI Studio platform is designed for projects requiring multiple models and several usage scenarios. It may be appropriate when an administration wants to analyze documents, perform optical character recognition, verify information or cross-check a form against an identity document.

In this configuration, the offering is built together with the customer. Use cases are analyzed in advance to select the appropriate models. It is generally advisable to begin with two or three clearly defined use cases and then design a coherent solution around those needs.

This approach creates a customized and scalable environment. Mistral also supports the customer in tuning the solution and adapting it to the identified use cases. Additional support may be provided for Kubernetes-related operations, including model updates and platform upgrades.

Toward hybrid model consumption

Organizations may need several model families depending on their use cases. A Mistral model may be suitable for one scenario, while a Llama model or another open-source model may be more effective for another task.

This is leading toward hybrid model consumption. The stated objective for the first half of 2027 is to offer an AI Factory capable of adapting GPU consumption to the intensity of the use cases and the models being used.

Such an approach should make it possible to allocate resources according to actual needs. It opens the way to more flexible use of AI infrastructure, with greater adaptability across models, workloads and expected performance levels.

Moving from proof of concept to production

Choose a manageable first use case

The choice of the first use case influences the rest of the project. An organization benefits from starting with a clearly defined need that is useful enough to produce measurable results and simple enough to be mastered quickly.

This initial stage helps identify technical constraints, security requirements, support needs and monitoring arrangements. It also provides valuable feedback before AI is extended to more complex processes.

The approach can follow three priorities:

select a concrete and relatively simple use case;
measure results and resource consumption;
prepare for expansion to other use cases.

Organize security workshops

Security workshops play an important role in projects involving several models or sensitive data. They help define authorized users, roles, access rights and the conditions governing use of the platform.

This preparation should take place before operational deployment. It helps establish a trusted framework and align the technical environment with the organization’s internal security requirements.

Security also concerns project governance. Teams must know what data may be processed, by which users and under what conditions. These rules facilitate adoption and limit uncontrolled use.

Plan for operational support

The required level of support depends on the selected offering and the skills available internally. An organization consuming compute resources directly must have the capabilities needed to administer its environment. A turnkey solution reduces this burden. A multi-model platform requires closer collaboration around use cases, models and their evolution.

Support may also cover Kubernetes operations, model updates, version upgrades and platform optimization. This becomes particularly important when the solution moves from experimentation to production.

Managing costs and environmental impact

Controlling consumption is a major issue in public procurement. The available budget must be considered alongside the resources used, including GPUs, tokens and other cloud components.

Monthly monitoring makes it possible to identify resources that have been ordered but are seldom used, and then adjust the environment. This consumption review also helps prepare for project expansion using actual data rather than initial estimates.

FinOps provides a framework for financial management. It makes it possible to compare deployed resources with observed usage and arbitrate between performance, sovereignty, cost, support and scalability.

Green IT complements this analysis. A carbon calculator can be used to track changes in the footprint associated with resource consumption. It may reveal overconsumption caused by virtual machines left running during inactive periods or by resources being deployed without a corresponding operational need.

Financial and environmental consumption monitoring should begin in the early stages of the project. It promotes more precise resource usage and supports decisions related to scaling.

Balancing sovereignty, performance and scalability

The choice of an AI environment cannot be based on a single criterion. Organizations must consider data location, security, expected compliance, performance, cost, support quality and scalability at the same time.

Sovereignty is particularly important when data is sensitive. In this context, an environment located in France and designed to meet security requirements provides an appropriate framework for public- and parapublic-sector projects.

Performance must then be assessed according to the use case. A model intended for document analysis does not necessarily require the same resources as an environment running multiple models and several simultaneous processes.

Finally, scalability must be incorporated into the decision. A solution suitable for a proof of concept may become insufficient as the number of users, documents or use cases increases. The initial choice should therefore preserve the possibility of gradual industrialization.

Conclusion

Sovereign AI on public cloud provides public-sector organizations with a practical framework for experimenting with, securing and industrializing their use of artificial intelligence. The Nuage Public marketplace simplifies access to cloud providers, cloud offerings and conditions suited to public procurement.

The most effective approach is to start with a manageable use case, secure access and data, monitor consumption and then prepare for expansion to multiple models and scenarios. Compute-resource offerings, Mistral bundles and the Mistral AI Studio platform make it possible to adapt the level of freedom and support to the project’s objectives.

Discover the solutions available on the OUTSCALE Marketplace

L’article Sovereign AI and sovereign cloud: how can public-sector projects be accelerated? est apparu en premier sur OUTSCALE blog.

OUTSCALE accelerates sovereign AI with NVIDIA HGX H200 GPUs in its SecNumCloud infrastructure

OUTSCALE — Mon, 14 Sep 2026 08:30:53 +0000

OUTSCALE announces the availability of NVIDIA HGX H200 GPUs within its SecNumCloud-qualified cloud infrastructure, operated in France, to accelerate sovereign AI and high-performance computing projects in a trusted environment.

Discover NVIDIA HGX H200 GPUs available at OUTSCALE.

Designed for the most demanding uses in generative AI and high-performance computing (HPC), the Nvidia HGX H200 architecture is built on multi-GPU configurations (8x H200) optimized for large language models and memory-intensive workloads.

The HGX H200 platform delivers up to 1.1 TB of aggregate HBM3e memory across its 8 GPUs, with each GPU offering up to 4.8 TB/s of memory bandwidth delivering higher performance for both training and extended-context inference.

With these new capabilities, OUTSCALE enables public and private organizations to accelerate the deployment of their AI projects, streamline the transition from experimentation to production, and maintain control over their strategic data.

Highly regulated sectors such as the public sector, healthcare, finance, and defense can therefore deploy their AI projects with a high level of security, compliance, and sovereignty.

To discover how these new capabilities can accelerate your artificial intelligence projects within a sovereign and secure environment, discover NVIDIA HGX H200.

L’article OUTSCALE accelerates sovereign AI with NVIDIA HGX H200 GPUs in its SecNumCloud infrastructure est apparu en premier sur OUTSCALE blog.

Running an organization in real time: What it demands from Cloud infrastructure

OUTSCALE — Fri, 28 Aug 2026 15:15:59 +0000

A customs tariff decision made thousands of miles away can, within a few weeks, upend an entire organization’s financial balance. Faced with this kind of shock, an organization’s ability to recalculate in real time the impact of a scenario on its value chain (suppliers, costs, lead times, margins) becomes a decisive competitive advantage. This is precisely what Virtual Twins of organizations are addressing today, an approach presented by Morgan Zimmermann, CEO of 3DEXPERIENCE, Dassault Systèmes, at OUTSCALE EXPERIENCES 2026: simulate before deciding, rather than endure. But this agility cannot be decreed by software alone, it depends on a Cloud infrastructure capable of absorbing it.

Simulating before deciding

The principle behind a Virtual Twin of an organization is to continuously represent how a company actually operates: its flows, dependencies, and constraints, so that scenarios can be tested before they are executed. Faced with a sudden rise in customs tariffs representing several billion euros in potential impact, an organization equipped with such a system can query its Virtual Twin, obtain a quantified assessment of several supply chain reorganization scenarios within moments, and make an informed decision. This kind of steering relies on specialized agents, capable of representing a situation, understanding it, engaging with teams, reasoning through alternatives, and triggering concrete actions.

Agility that comes with a technical price

This level of responsiveness places strong demands on the infrastructure that supports it. First, it requires sufficient compute elasticity to absorb occasional, unpredictable simulation spikes without excessive latency: a Virtual Twin that takes several hours to respond loses some of its value. Second, it requires strict protection of data and intellectual property: an organization’s Virtual Twin aggregates strategic information on suppliers, costs, and internal processes, often shared with an ecosystem of partners. Rigorous isolation therefore becomes a condition of trust, not an option. Morgan Zimmermann himself raises this requirement for pooled yet controlled compute power at the close of his demonstration, pointing to OUTSCALE’s ability to pool costly compute resources to make them economically efficient at the point of use: a principle also found in the dynamic GPU resource allocation implemented within OUTSCALE’s AI Factories.

OUTSCALE, the foundation that makes this possible

It is precisely on these requirements that OUTSCALE builds its response. Philippe Miltin, CEO of OUTSCALE, sums up the criteria for this sovereignty: recognized certifications and qualifications, including SecNumCloud qualification, reversibility and interoperability of environments without dependency on a single vendor, and long-term control over the technology investments that underpin it. Arnaud Bertrand, CTO of OUTSCALE and Senior Vice President of OUTSCALE, Dassault Systèmes, details the technical translation of this within the AI Factories: better, pooled use of GPU resources, and physical and logical segregation guaranteeing each organization the isolation of its workloads, even when several partners are working on the same scenario.

Running an organization in real time is therefore not only a software challenge: it is an infrastructure challenge. By providing the compute power, sovereignty, and isolation needed for this kind of steering, OUTSCALE positions itself as the Cloud and AI foundation that enables organizations, including in the most critical sectors, to turn uncertainty into decision.

To watch the replays of Morgan Zimmermann’s and Philippe Miltin’s talks, find the links below.

Organizations steerable through their Virtual Twins – Morgan ZIMMERMANN at OUTSCALE EXPERIENCES 2026:

OUTSCALE: the sovereign Cloud and AI foundation for critical organizations – Keynote by Philippe Miltin at OUTSCALE EXPERIENCES 2026:

Technology vision and new services – Keynote by Arnaud Bertrand at OUTSCALE EXPERIENCES 2026:

L’article Running an organization in real time: What it demands from Cloud infrastructure est apparu en premier sur OUTSCALE blog.

Why Industrial AI Needs Dedicated Infrastructure

OUTSCALE — Thu, 27 Aug 2026 13:00:40 +0000

Generative AI has long been envisioned as a service simply plugged into standard Cloud infrastructure. That assumption no longer holds once AI is expected to deliver reliable results on industrial topics: sizing a mechanical part, anticipating a supply chain disruption, validating a regulatory configuration. A general-purpose language model, however powerful, lacks an understanding of physical reality and can therefore produce results that are not physically correct. This observation, made by Florence Hu-Aubigny, Executive Vice-President Research & Development at Dassault Systèmes and Chief Architect of the 3DEXPERIENCE platform, during OUTSCALE EXPERIENCES 2026, reflects the same rationale behind the design of AI Factories by OUTSCALE: an infrastructure built from the ground up for training and running industrial AI models, within a sovereign and secure framework.

An architecture designed to retain nothing

The first guarantee offered by AI Factories rests on a principle that is simple to state but demanding to implement: no data is stored there on a lasting basis. The AI Factory operates as a compute space, not a storage space: data passes through, computation runs, and everything is then returned to OUTSCALE’s sovereign IaaS infrastructure, where certifications reside and where data persistence is fully guaranteed. Arnaud Bertrand, CTO of OUTSCALE and Senior Vice President of OUTSCALE, Dassault Systèmes, sums up this architectural choice plainly: the goal is to ensure that, at no point, any trace of customer data remains in the compute space.

This non-persistence is paired with physical and logical segregation by customer: each organization has its own dedicated AI endpoint, guaranteeing the isolation of its workloads and its intellectual property. This architecture was built in partnership with Nvidia (based on a generation of GB200 and GB300 components, paired with H200 GPUs) and with Bull for hardware integration, providing state-of-the-art compute capacity while retaining full control over its implementation.

Elastic access to multiple model families

Through a dedicated control plane, AI Factories provide access to a pool of models that includes Nvidia’s NIM models, such as Nemotron, as well as the models OUTSCALE already works with, such as Mistral. OUTSCALE will also add Aleph Alpha and LightOn to this pool to round out its sovereignty offering, and will provide access to LLaMA and OpenAI models as well. This pool powers a multi-model agentic platform that can orchestrate, depending on the use case, whichever model is best suited, whether open, sovereign, or international.

This elasticity comes with a shift in how GPU resources are allocated. Rather than static allocation (where a customer pays for reserved capacity that isn’t always fully used), OUTSCALE is building dynamic allocation, adjusted to actual usage.

One standard, two deployment scales

AI Factories are deployed according to two complementary approaches. On one side, international zones in Europe, the United States, and Asia offer maximum elasticity and allow workloads to be distributed across several continents. On the other hand, a strictly SecNumCloud zone, isolated and self-contained, is reserved for the most sensitive workloads, with no possibility of spillover to international zones.

It is this combination: non-persistence, per-customer segregation, model diversity, dynamic allocation, and dual-speed deployment – that allows OUTSCALE to offer a credible response to organizations operating in critical, regulated sectors: healthcare, energy, aerospace, defense, and the public sector. These AI Factories, initially designed to meet Dassault Systèmes’ internal needs, are now being gradually opened up to all companies and institutions that share the same requirement: industrializing AI without ever compromising on the sovereignty of their data.

To find out more, watch the replays of Arnaud Bertrand’s and Florence Hu-Aubigny’s talks.

Technology vision and new services – Arnaud Bertrand’s keynote at OUTSCALE EXPERIENCES 2026:

Dassault Systèmes: Industrial AI platform architecture – Florence Hu-Aubigny at OUTSCALE EXPERIENCES 2026:

L’article Why Industrial AI Needs Dedicated Infrastructure est apparu en premier sur OUTSCALE blog.

ÆTHER Consortium: Building Europe’s Industrial Capability for AI at Scale

OUTSCALE — Wed, 29 Jul 2026 08:22:06 +0000

Every year, European companies pay tens of billions of euros to American players for their Cloud software and services: a dependency now documented by several parliamentary reports, which call for a reconquest of European digital sovereignty.

But a dynamic of sovereignty cannot be decreed by a report: it is built through example, and through concrete industrial capabilities. This is precisely the ambition carried by the European ÆTHER consortium, in which OUTSCALE, the Cloud and AI brand of Dassault Systèmes, takes part.

What Is the ÆTHER consortium?

ÆTHER brings together complementary European players in energy, construction, high-performance computing, semiconductors, Cloud, and AI: 2CRSi, SiPearl, Axelera AI, Viridien, ÉS, Socomec, Haffner Energy, Equans, Demathieu Bard, Nhood, and Projex Ingénierie, alongside OUTSCALE.

The consortium pursues a clear ambition: to demonstrate that Europe possesses all the expertise required to design, build, power, operate, and scale AI infrastructure.

The consortium has announced its application for the European Commission’s “AI Gigafactory” call for proposals, launched as part of the EU’s digital and industrial sovereignty strategy. The project targets two sites in the Strasbourg region, FR-SXB1 and FR-SXB2, currently in advanced negotiations with the long-term ambition of reaching a total electrical capacity of more than 400 MW.

“Industrial AI will only deliver on its promise if Europe builds infrastructure equal to its ambitions”, Florence Verzelen, Executive Vice President EMEA, Dassault Systèmes

OUTSCALE's role: turning compute into sovereign, secure, and industrialized Cloud and AI services

Concretely, OUTSCALE brings to ÆTHER its expertise as a sovereign Cloud and AI operator: the ability to turn large-scale compute power into secure, governed, and industrialized services for businesses, institutions, and research organizations.

“As the Cloud and AI foundation of Dassault Systèmes, OUTSCALE brings to ÆTHER unique expertise at the intersection of sovereign cloud, high-performance computing, and mission-critical industrial applications. Our ambition is to enable businesses, public institutions, and research organizations to develop and operate their AI models in high-performance, secure, and sovereign environments that protect data, know-how, and intellectual property. With ÆTHER, Europe is equipping itself with a strategic industrial capability to design, host, and operate AI at scale”, Arnaud Bertrand, Chief Technology Officer, OUTSCALE, and Senior Vice President, Cloud R&D, Dassault Systèmes.

A capability that reinforces an already established strategy

Joining ÆTHER doesn’t start from scratch: it reinforces the AI Factories strategy that Dassault Systèmes has entrusted to OUTSCALE, which already aims to provide the Cloud, orchestration, and execution capacities needed to train, infer, and operate AI models at scale in critical environments. ÆTHER brings an additional European capability to this trajectory, aligned with the needs for sovereignty, performance, and scale in industrial AI.

All the pieces already exist: now they need to be connected

Europe today has every piece in the AI value chain: energy, Cloud, high-performance computing, semiconductors, industrial software, and research. The value of the ÆTHER consortium, like that of any AI Gigafactory project, will be measured by its ability to connect these links to serve concrete use cases: training models, adapting them, operating them, securing them, governing them, and integrating them into organizations’ business processes.

This is the ambition that OUTSCALE, Dassault Systèmes, and all the partners of the ÆTHER consortium carry forward together.

View the press release.

L’article ÆTHER Consortium: Building Europe’s Industrial Capability for AI at Scale est apparu en premier sur OUTSCALE blog.

Overcoming the challenges of implementing agentic AI

OUTSCALE — Tue, 28 Jul 2026 13:05:33 +0000

Intelligent automation is entering a new phase with agentic AI. By combining the execution capabilities of automation with the reasoning of generative AI, it enables autonomous agents to act directly at the core of business processes. Organizations expect three major benefits: cost reduction, faster turnaround times, and improved service quality.

Led by Simon LUSINCHI (Partner at Artefact) and Marie-Astrid de Foucauld (AI Ecosystem Manager at OUTSCALE) during the OUTSCALE EXPERIENCES event, this session analyzes the major obstacles related to deploying agentic AI in companies. It highlights the importance of going beyond a purely technological vision to jointly orchestrate human, business, and technical levers, which are key to sustainable transformation.

Understanding the stakes of agentic AI

The emergence of agentic AI is part of the ongoing evolution of technological uses, following RPA automation, traditional analytical AI, and then generative AI. Characterized by its ability to reason and chain actions, this new form of AI marks a turning point: it no longer limits itself to content generation, but decides, acts, and adapts based on its context.

This technological shift fundamentally transforms how projects are approached. The objective is no longer simply to optimize the time allocated to an isolated task, but to rethink and transform the overall functioning of an entire process.

Three methods to get started

To initiate an agentic transformation, organizations can choose between three methods:

Global strategy: map all processes, quantify potential value, and prioritize use cases to build an enterprise-wide agentic roadmap.
Function-based approach: first target a cross-functional area (HR, legal, marketing, administrative) and identify processes to optimize with agents.
Targeted pilots (PoCs): launch pilots on process fragments to build a methodology and quickly demonstrate value (e.g., commitment management).

A progression of benefits

The deployment of agentic AI can be understood across three levels of maturity: individual use, mutualization, and end-to-end autonomy. At each level, ROI becomes clearer, more collective, and more strategic.

Individual level : An assistant automates specific tasks for a person or function. The gain is real but remains diffuse, as it mainly improves local productivity.

Mutualization level : An agent orchestrates multiple actions around a common objective, serving a team or shared workflow. The value created becomes collective and easier to measure in operations.

Autonomous level : The agent takes charge of an entire business process end-to-end. This is where the impact becomes systemic, with performance indicators focused on execution speed, quality, cost, and scalability.

A systemic approach to transformation

To avoid the common mistake of trying to deploy artificial intelligence systematically, Artefact structures its approach around a global vision. Very often, the ideal solution does not lie in integrating an autonomous agent, but rather in simplifying automation or restructuring an existing process.
This multidimensional approach is structured around three strategic axes:

The human dimension: defining a global vision, building the internal narrative, setting appropriate safeguards, and continuously developing skills.
The business dimension: relying on precise process mapping to determine whether the added value will truly come from AI, an RPA solution, or simple organizational optimization.
The technological dimension: focusing on rigorous data management, robust architecture, and the establishment of clear governance.

Data governance and security requirements

The effectiveness of an AI agent is closely tied to the quality of its data architecture. To be fully usable, information must follow a rigorous process of collection, cleaning, structuring, and classification. However, it is crucial to avoid paralyzing the system through excessive classification into “confidential” or “restricted” categories, which could block access to useful data and significantly degrade overall performance.

At the same time, security is a fundamental pillar of the approach. Before considering the deployment of an agentic solution, any organization must anticipate and formalize its requirements in terms of network constraints, hosting conditions, integration protocols, and access rules. Beyond purely technical aspects, this reflection represents a true challenge of trust and overall governance.

Selecting the right solution

To address different levels of maturity and organizational constraints, OUTSCALE offers three complementary approaches based on the desired level of autonomy.

GPU as a Service is aimed at highly technical profiles seeking full control over infrastructure and models.
LLM as a Service enables rapid deployment of a private API within a sovereign and secure framework.
OUTSCALE offers a turnkey software solution designed to be accessible to non-technical users and to accelerate large-scale adoption.

OUTSCALE also relies on an ecosystem of cloud and consulting partners, including Artefact, to support clients at every stage of the project. This support covers initial assessment, use case scoping, deployment, and ongoing optimization of usage and performance.

Key Takeaways

1. Agentic AI fundamentally changes how processes are executed. It no longer simply assists users: it acts, orchestrates, and can make certain decisions within a defined framework. This requires organizations to rethink business processes as a whole, rather than automating only isolated tasks.

2. Success relies on a truly cross-functional approach. Three complementary dimensions must be aligned: the human dimension, with adoption and upskilling; the business dimension, with the selection of the most relevant use cases; and the technological dimension, with data quality, architecture, and integration. The challenge is therefore not only technical, but also organizational and operational.

3. The main obstacles relate to data and governance. Data quality, accessibility, and security directly determine agent performance. Choosing solutions adapted to the organization’s level of maturity is also critical to delivering reliable, measurable, and sustainable results.

L’article Overcoming the challenges of implementing agentic AI est apparu en premier sur OUTSCALE blog.

AI Factories: how OUTSCALE is industrializing AI for the most demanding organizations

OUTSCALE — Tue, 28 Jul 2026 09:12:16 +0000

Artificial intelligence is entering a new phase, that of industrialization, and calls for an infrastructure tailored to its own requirements: intensive computing, training, and running models at scale. For organizations in sensitive sectors (healthcare, finance, the public sector, defense), this equation comes with an added requirement: accessing this computing power while retaining full control over their data.

This is precisely the equation that AI Factories by OUTSCALE, Dassault Systèmes, are designed to solve.

A strategy born for the most demanding use cases

AI Factories were not designed in the abstract. They were first developed to meet Dassault Systèmes’ internal needs, combining the group’s virtual twin technologies with Nvidia’s AI infrastructure, open models, and accelerated computing libraries, to build science-validated industrial models.

Dassault Systèmes entrusted OUTSCALE with carrying this strategy forward: providing the infrastructure, orchestration, and security capabilities needed to train, infer, and operate AI models in critical environments. An expertise built to serve the world’s most demanding industrial players, now extended to all businesses and institutions.

“With Dassault Systèmes, OUTSCALE serves the most demanding industrial players on the planet […]. That’s our uniqueness: what is designed for the most demanding becomes an advantage for regulated markets,” Philippe Miltin, CEO, OUTSCALE.

Three structuring guarantees

Announced at OUTSCALE EXPERIENCES 2026, AI Factories combine three guarantees that matter to sensitive organizations:

Sovereignty, ensured notably through data non-persistence: transient training data is not stored persistently within the AI Factory, which functions as a computing space rather than a storage space. Data persistence remains guaranteed within OUTSCALE’s sovereign, SecNumCloud-qualified Cloud infrastructure.

Optimized resources and usage: dynamic management of GPU and CPU resources, made possible by the strategic partnership with Nvidia, which limits chronic overcapacity and pools scarce resources.

The ability to scale on global resources, as close as possible to each organization’s needs for both training and inference at scale.

One building block within a broader continuum

This strategy fits within a continuum specific to Dassault Systèmes’ industrial AI: one that connects virtual twins, science, simulation, real-world data, AI models, and IP protection. OUTSCALE provides the execution layer, enabling the training, running, and orchestration of the models and agents behind new product categories: virtual companions, generative experiences, virtual twins, and Industry World Models.

Unlike a simple GPU resource provider, OUTSCALE brings a complete layer of service, orchestration, security, and governance: the goal isn’t just to supply computing power, but to make it usable within critical environments, with the guarantees expected by regulated organizations.

Built for regulated and sensitive sectors

AI Factories are designed for businesses, institutions, researchers, and organizations in regulated sectors that want to develop, adapt, and operate their own artificial intelligence models without giving up control over their data, know-how, and intellectual property.

This approach builds on a compliance foundation already proven at OUTSCALE: SecNumCloud qualification, ISO 27001-27017-27018, HDS, and TISAX certifications, CISPE membership, and SOC 2 Type 2. Guarantees that make AI Factories a credible answer for Europe’s most demanding sectors.

Learn more about OUTSCALE’s sovereign Cloud infrastructure.

L’article AI Factories: how OUTSCALE is industrializing AI for the most demanding organizations est apparu en premier sur OUTSCALE blog.

How can Europe ensure the deployment of sovereign and trustworthy AI?

OUTSCALE — Wed, 22 Jul 2026 13:20:58 +0000

Artificial intelligence is now entering the stage of industrialization. Beyond technical challenges, this scaling phase raises a strategic question for the European continent: the conditions required for the emergence of high-performing, autonomous, and trustworthy AI.

As part of OUTSCALE EXPERIENCES 2026, Igor Caron (Co-founder, Lighton), Emmanuel Leroux (CEO, Bull), and Mathieu Feuillet (CISO, Dassault Systèmes) shared their perspectives on the pillars of this European success.

Infrastructure at the heart of technological sovereignty

Digital independence cannot be real without full control of the value chain. Emmanuel Leroux emphasizes that sovereignty necessarily requires control over “compute,” encompassing physical components, firmware, and software platforms alike.

Today, Europe faces strong dependence on foreign GPUs and the uncertainties of global supply chains. To address this, it is crucial to strengthen industrial capabilities and consolidate infrastructure within the continent.

Ultimately, without solid hardware and software autonomy, Europe’s capacity for innovation will remain fragile and dependent on external decisions.

Ethics as a lever for competitiveness

Far from being a mere constraint, trust is becoming the key driver of technology adoption. It cannot be imposed; it must be built patiently through tangible proof.

At Dassault Systèmes, this approach takes the form of a rigorous framework built on fundamental pillars: the primacy of human oversight, system security, strict respect for intellectual property, as well as fairness, transparency, and environmental responsibility.

For businesses, the question has shifted from the technical feasibility of AI to its reliability. Addressing this challenge requires providing concrete assurances regarding data governance, effective control of decision-making processes, and full traceability of outcomes.

Leveraging domain expertise through AI

The effectiveness of an AI solution is not measured solely by the sophistication of its algorithms, but above all by its ability to mobilize a company’s informational assets.

As Igor Caron points out, needs originate in the field: business units want AI tools integrated into the enterprise, capable of securely leveraging internal knowledge and sensitive data.

To address this challenge concretely, LightOn focuses on a technical architecture designed for real-world business use. The idea is not to systematically rely on heavy and costly infrastructure, but to favor more efficient processing, particularly using CPU-based approaches when they meet business needs.

This approach also relies on the RAG methodology, which consists of connecting the AI model to the organization’s internal documents and knowledge. Instead of depending solely on the raw power of a large model, the AI retrieves relevant information from company sources, then reformulates it in a usable, traceable, and contextualized way.

The benefits are twofold: on one hand, infrastructure and computation costs are significantly reduced; on the other, the relevance of responses improves because the system relies on up-to-date internal data. This makes it possible to build AI that is more affordable, easier to deploy at scale, and more profitable for organizations.

Beyond sovereignty: the imperative of excellence

While sovereignty is an essential foundation, it cannot be an end in itself.

To compete with American and Chinese powers, Europe must develop technologies whose competitiveness extends internationally. This ambition requires moving beyond a purely defensive posture to embrace a true culture of overall performance.

The fundamental objective is no longer just to protect digital borders, but to bring superior solutions to market.

Prioritizing use cases to catalyze deployment

The lack of concrete use cases is a major, though often overlooked, barrier to AI development in Europe.

Beyond infrastructure gaps, the continent suffers from limited real-world adoption of technological tools. Since usage drives demand, it is the only true guarantee of return on investment.

Rather than focusing exclusively on increasing computing power, efforts should concentrate on designing specialized business applications rooted in the needs of research, healthcare, and industry.

Architecture serving strategic autonomy

Achieving independence does not rely on using a monolithic tool, but on deploying intelligent structures.

The example of Dassault Systèmes highlights this approach: by orchestrating a multitude of specialized models, each fulfilling a defined role, it becomes possible to guarantee high-level results. This approach drastically reduces dependence on a single provider.

This organizational model thus makes it possible to reconcile technical performance requirements with sovereignty imperatives.

Trust as the cornerstone of European AI

The emergence of trustworthy European AI does not depend on a single factor, but on a synergy of requirements: controlled infrastructure, a robust ethical framework, the use of company-specific data, strong technological capabilities, and the development of concrete use cases.

The continent already has the expertise and partners needed to win this race. The main challenge now is to unite these strengths behind a common goal: to build AI that protects sovereignty, performs on a global scale, and is widely integrated into our economies.

Key takeaways

1. Sovereignty starts with control of infrastructure

Europe cannot claim autonomous AI without controlling the entire technological stack, from hardware (especially GPUs) to software. Today, dependence on foreign players weakens its capacity for innovation. The priority is therefore to strengthen industrial capabilities and build sovereign infrastructure, an essential condition for independent decision-making and sustainable innovation.

2. Trust and ethics become a competitive advantage

European AI must differentiate itself through reliability. This requires strong guarantees: human oversight, security, transparency, respect for data and intellectual property. The challenge is no longer just to make AI work, but to prove that it is trustworthy. This requirement can become a lever for competitiveness against other global powers.

3. Impact depends on real-world use cases and performance

Beyond sovereignty, Europe must create real value. This involves business use cases (industry, healthcare, etc.), leveraging company-specific data, and efficient architectures (such as specialized models rather than a single general model). The dual objective is to accelerate adoption and deliver globally competitive solutions.

These three pillars infrastructure, trust, and use cases together form the foundation of a European AI that is sovereign, credible, and high-performing.

L’article How can Europe ensure the deployment of sovereign and trustworthy AI? est apparu en premier sur OUTSCALE blog.

How French Customs Deployed a Sovereign AI for 17,800 Agents

OUTSCALE — Mon, 20 Jul 2026 14:25:22 +0000

As part of OUTSCALE EXPERIENCES, this plenary session highlighted a major transformation led by the French Directorate General of Customs and Indirect Taxes (DGDDI). Undertaking a complete overhaul of its customs code from 500 to 1,200 articles, the institution had to support a rapid upskilling of its agents, who were faced with significant regulatory changes.

To address this operational challenge, the DGDDI deployed a natural language artificial intelligence assistant, accessible on mobile devices, to facilitate the application of the new rules in the field.

On this occasion, Jérôme Sandrini, CRO at OUTSCALE, Éric Laurent, Head of the “IT Support” Office at DGDDI, and Quentin Lebourgeois, Public Sector AI Services Lead at Sopra Steria, shared their feedback and perspectives on this project.

Challenges and Issues

The legislative restructuring caused a loss of reference points for agents. The expansion to 1,200 articles made the application of rules less predictable, raising numerous operational questions, particularly during field missions.

After several pilot experiments starting in 2024, the DGDDI launched an industrialization phase in 2025 with the goal of a full-scale rollout by May 1, 2026.

Strategic Requirements and Priorities

To guide its technological and contractual decisions, the DGDDI established a set of core requirements. The protection and confidentiality of sensitive information were top priorities, alongside a strong sovereignty requirement involving hosting in France and guaranteed reversibility of systems.

From an operational standpoint, the tool needed sufficient performance and robustness to support approximately 17,800 users, while remaining flexible enough to adapt to evolving usage and regulatory changes. Given these security and compliance constraints, the administration opted for a sovereign architecture based on open source, using the UGAP procurement channel.

Systems and Infrastructure Implemented

The solution relied on close collaboration between three key entities: the DGDDI, which defined business needs; Sopra Steria, responsible for technical integration and software design; and OUTSCALE, which provided hosting on a sovereign SecNumCloud-certified Cloud infrastructure, ensuring computing power (GPU), flexibility, and rapid deployment.

To enable fast rollout and efficient scaling without heavy hardware investment, OUTSCALE provided an industrialized Cloud infrastructure optimized for GPU resources.

Jérôme Sandrini, Chief Revenue Officer at OUTSCALE, emphasized that for any large-scale AI project, Cloud usage becomes essential. OUTSCALE delivers “this elasticity combined with security and the well-known SecNumCloud certification.” He also highlighted the critical role of a partner ecosystem in accelerating time-to-market and enabling large-scale integration of new services.

Acting as the bridge between technical infrastructure and operational needs, Sopra Steria industrialized all components: conversational interfaces, ingestion pipelines, retrieval-augmented generation (RAG), and vectorization. To ensure security and operability while preventing “shadow AI,” the company integrated the solution into the information system, including directory services, authentication, and the SOC.

According to Quentin Lebourgeois, Head of AI for Public Services at Sopra Steria, “the use of managed sovereign solutions made it possible to build a response tailored to business challenges while avoiding the complexities of infrastructure management.”

According to the DGDDI, the project’s success was based on several strategic pillars: strong executive support, clear prioritization of needs with security and sovereignty at the forefront, and a pragmatic approach informed by past experience. To facilitate industrialization, deployment began with a simplified use case. As Éric Laurent, Head of the “IT Support” Office at the DGDDI, noted, “the operational goal was to provide agents with a natural language assistant directly accessible on their mobile devices during their daily operations.”

Key Takeaways

The value of generative AI lies in its ability to address immediate business needs. By focusing on regulatory support for field agents, the DGDDI removed a major operational bottleneck, enabling agile integration and organization-wide adoption.
The project’s success relied on a demanding technical foundation: the use of a SecNumCloud-certified Cloud, open-source technologies, and strict adherence to sovereignty criteria such as local hosting, reversibility, and interoperability with the information system. This strategy ensured a balance between technological progress and regulatory compliance.
The success of this large-scale deployment was driven by the complementarity of a strategic trio: deep understanding of business challenges, tailored technical integration, and scalable infrastructure. This close collaboration between the DGDDI, Sopra Steria, and OUTSCALE enabled rapid service expansion while maintaining a high level of security.

L’article How French Customs Deployed a Sovereign AI for 17,800 Agents est apparu en premier sur OUTSCALE blog.

Securing Banking: Compliance, SecNumCloud, and DORA Resilience

OUTSCALE — Fri, 15 May 2026 12:40:51 +0000

For a bank, any migration of critical applications to the cloud starts with the same question: will we be at least as well protected as we are on-premises?

Compliance as a starting point

Before discussing cost or performance, risk, compliance, security, and IT teams assess whether a cloud solution can meet regulatory requirements and supervisory expectations. These requirements: operational resilience, control over outsourcing, incident management, and data protection determine which applications can be migrated and how deep the transformation can go. Compliance thus becomes the foundation of the cloud strategy.

OUTSCALE: compliance “by design”

At OUTSCALE, we chose to embed compliance “by design” from the company’s inception over 15 years ago. Our sovereign infrastructure is SecNumCloud-qualified by ANSSI, a certification that attests to a high level of technical, operational, and legal rigor, particularly regarding sovereignty, sensitive data protection, and resilience against extraterritorial laws. This qualification is based on a highly demanding framework that complements ISO/IEC 27001.

Why SecNumCloud matters for banks

For a financial institution, SecNumCloud is a key trust marker: it ensures that the evaluated cloud offering meets a high level of security, that data is hosted within the European Union, and that the operator is not subject to extraterritorial laws that could compromise confidentiality or service availability. At OUTSCALE, this foundation is reinforced by certifications such as ISO 27001, 27017, 27018, HDS, and SOC 2 Type 2, further strengthening trust in the infrastructure for critical use cases.

DORA: a new framework for digital operational resilience

The European DORA regulation (Digital Operational Resilience Act) harmonizes ICT risk management requirements across the EU for financial entities. It introduces stronger ICT risk governance, regular resilience testing, stricter oversight of critical service providers, and detailed reporting obligations in the event of major incidents. Leveraging a qualified sovereign cloud offering transparency, audit rights, traceability, and environment isolation becomes a key enabler for demonstrating DORA compliance and securing relationships with supervisory authorities.

Sovereign SaaS solutions beyond infrastructure

At OUTSCALE, we offer SaaS solutions deployed on our sovereign cloud to manage fund compliance and structure end-to-end data value chains—from ingestion to delivery in a controlled environment. Risk, compliance, and business teams thus benefit from ready-to-use application building blocks designed from the outset to meet regulatory expectations.

Once this foundation of security, compliance, and sovereignty is in place, banks can focus on transforming their use cases and leveraging their data to make better decisions. The third article in this series will address this dimension: how to “transform and decide” using a sovereign, hybrid infrastructure and well-governed AI.

L’article Securing Banking: Compliance, SecNumCloud, and DORA Resilience est apparu en premier sur OUTSCALE blog.