AI and technological sovereignty: understanding the risk
"It's time for founders to realize that when you build a technology company, you need to think about the entire stack."
For many years, technology was treated primarily as a cost center. The priority was to deploy tools that were efficient, reliable, and easy to implement. This approach led many organizations to concentrate their dependencies around a handful of major providers, most of them American.
For David Djaïz, that mindset is now outdated. Business leaders must start viewing their technology architecture as a strategic issue in its own right, not simply as an IT concern. Behind every application, every software solution, and every AI use case lie infrastructure choices with tangible economic and operational consequences.
The objective is not to pursue absolute technological sovereignty at all costs. The first step is simply becoming aware of the dependencies that already exist. David points to a recent comment by Arthur Mensch, founder of Mistral, who noted that token consumption already represents roughly 10% of his company's payroll expenses. This figure illustrates a broader trend: as AI becomes increasingly embedded in business processes, the associated costs will grow and technological dependencies will deepen.
A few ways to reduce risk include:
- Using multiple cloud providers: This restores bargaining power and makes it possible to migrate part of the infrastructure if prices rise sharply or contractual terms change.
- Handling sensitive data differently: Not all data needs to be hosted on American public cloud infrastructure. The most critical information may require more controlled environments.
- Building more resource-efficient AI systems: Leveraging open-source models such as Mistral AI, DeepSeek, or Meta locally can reduce token consumption and limit certain dependencies.
AI and technology: what impact on talent and organizations?
"AI can be an extraordinary enabler, but it must be used in the right way and in the right place."
For David, AI is not merely a technological transformation. By becoming embedded directly into workflows, it is reshaping how companies create value and organize work. What he observes is a reallocation between technical capital and human labor. The challenge is no longer simply how to deploy AI, but how to manage this transformation without weakening collective intelligence or undermining human capital.
He also sees a significant gap between productivity promises and real-world outcomes. AI may streamline one part of a process while creating new requirements elsewhere: oversight, quality control, verification, and coordination. It can "save 50% of the effort here" while generating "40% more work" somewhere else in the process. For now, the overall net impact remains disappointing in most organizations.
The deeper challenge, however, is organizational. Many companies still approach AI through a traditional problem-solving lens, whereas AI requires a broader redesign of workflows, collaboration models, and operating structures. David frames this transformation around three dimensions:
- Technological readiness: Evolving the technology stack without creating new dependencies.
- Organizational readiness: Rethinking workflows and ways of working.
- The social contract: Determining how the value created by AI is shared among shareholders, customers, and, most importantly, employees.
This third dimension is often the most neglected. If AI generates productivity gains, who will ultimately benefit? Companies will need to answer that question not in theory, but through their day-to-day practices.
The impact is already visible in the labor market, particularly for junior talent. Many of the analytical and synthesis tasks traditionally assigned to early-career professionals are among the easiest to automate. Nevertheless, David believes that adopting a "no junior hiring" policy would be a mistake. Those early years of learning are what build the talent pipeline organizations will depend on in the future. Cutting back too aggressively on junior hiring is effectively undermining the very foundation of tomorrow's workforce.
Why recruitment has become his number one priority
After several years in the public sector, David encountered a responsibility he had never truly faced before: building his own team. In public institutions, managers lead teams, but they rarely choose the people they work with. Upon taking over Ascend Partners, he quickly realized how difficult hiring can be, even with highly structured processes in place.
- You have to accept a certain error rate: Human fit and professional fit can only be fully assessed over time. Despite interviews, reference checks, and case studies, it often takes two or three months of working together to know whether a relationship will truly succeed. Even leading consulting firms accept error rates of around one-third despite rigorous hiring processes. The goal is therefore not to eliminate mistakes entirely, but to acknowledge and normalize them.
- You need to understand your strengths and weaknesses: At Ascend Partners, a growing company, uncertainty is part of the journey. While larger consulting firms may offer greater predictability, David highlights different advantages: exposure to strategic topics, greater ownership, significant responsibilities, and a highly entrepreneurial environment.
- Recruitment requires time and energy: If hiring is genuinely a 10/10 priority, it is inconsistent to dedicate only a marginal share of your calendar to it.