AI, a powerful growth driver for publishers
Impact on EU GDP
An additional annual growth of +0.7%
Philippe Aghion, Nobel Prize-winning economist, estimates that AI could “increase the EU’s GDP by 0.7% per year for ten years.” This impact, while significant, remains manageable and would be comparable to that of ICT in the 1990s-2000s.
The GDP of the European Union is approximately 18,000 billion euros (2024), the total value of all goods and services produced across the 27 member states. It is worth noting that services account for 74% of EU GDP (and even more in France).
+0.7% per year represents an additional 126 billion euros per year. This is considerable (on the order of the ICT impact in the 1990s/2000s), but does not represent an unsustainable shock.
Two key takeaways:
- Risk of concentration: The lack of competition due to the dominance of a few players (Nvidia, Amazon, Microsoft, Google…) could limit the equitable distribution of created value, especially in Europe.
- Positive impact on employment: The automation of administrative tasks would be offset by an increase in overall productivity, creating new jobs in other positions.
Obviously, like all predictions, these are destined to be contradicted by facts. But the experience and data acquired during the transformation of the economy by ICT seem solid enough to support these reasonable indications.
The key role of applications
Applications are part of every professional activity. It is natural that they progressively integrate AI-related functions, whether to increase productivity, simplify usage, or innovate by introducing previously unimagined features.
Growing daily usage
Business applications are omnipresent in the professional world. On average:
- An office worker uses 11 to 15 applications per day (Gartner, Stratapp).
- A sales professional uses 8 to 12, while a developer can use up to 15.
To elaborate further:
An administrative employee uses 5 to 8 applications (email, ERP, Excel, video conferencing tools, etc.).
A sales professional uses 8 to 12 applications (CRM, PowerPoint, Excel, LinkedIn, prospecting tools, etc.).
A developer uses 10 to 15 applications (IDE, Slack, Jira, GitHub, monitoring tools, etc.).
Industry estimates suggest application usage durations ranging from 1.5 to 3 hours/day on average, depending on role and sector. This time can rise to 4-5 hours/day for highly digitalized professions (tech, data, marketing).
Analysis by company size
Furthermore, research on application portfolios (SaaS + internal apps) shows that:
- Very large organizations (1,000+ employees) often have 200 to 500 applications in total.
- SMEs (10-250 employees) typically have between 20 and 100 applications.
- Micro-businesses (fewer than 10 employees) generally use fewer than 20 applications.
Even if an employee does not use all of them, a larger application portfolio will mechanically increase the number of tools to which an employee is exposed.
Publishers, key players in economic impact
Why do software publishers concentrate the benefits of AI?
Business application publishers have a specific role: pooling expertise and R&D.
Their objective is clear: accelerate deployments, simplify updates, and reduce costs for their clients.
Through this specialization, every company can access the best technologies without having to develop its own skills or bear the associated costs.
This model creates a powerful multiplier effect. A single publisher can reach hundreds of mid-size companies, thousands of SMEs, and tens of thousands of micro-businesses. As a result, thousands of organizations benefit directly.
Quickly reaching millions of users
In France, SMEs and micro-businesses represent more than 50% of salaried employment. These companies have neither the time nor the resources to develop their own applications. They therefore prefer to rely on SaaS solutions offered by publishers. This also allows them to avoid technical constraints.
Even large companies, although they have more resources, often use these solutions. They then adapt them to their specific needs.
Economic consequences
Significant expenditure
Within the EU, IT spending represents approximately 1,300 billion dollars, or 1,100 billion euros, which represents 6.1% of the Union’s GDP. Within this spending, applications alone account for 250 billion euros, or 1.4%.
AI impact: a doubling of publisher revenues
One way or another, publishers will capture their share of this newly created value.
If we accept Philippe Aghion’s forecast, and that the +0.7%/year is shared equally between ICT providers and their business customers, and also that current proportions between software and IT services remain identical, then we can estimate that:
- Applications will contribute to an annual GDP increase of 290/(290+490) = 37% * 0.7% = +0.26%/year, or (for the EU) +47 billion euros/year;
- The revenue of application publishers will increase by 47/2 = 23.5 billion euros/year, or +8.1%/year. That is 2.2x over 10 years, solely from the added value carried by AI services integrated into applications.
Obviously, there will be losers and winners. The former will fall behind, the latter will significantly outperform.
- Losers: those who do not seize the opportunity in time, or who limit themselves to decorative features;
- Winners: those who will exploit their monopolies (notably the GAFAMs), and also publishers who start early and build over time.
An opportunity to seize
The big players will try to build and maintain their monopolies and find revenue sources to absorb their (unreasonably?) massive investments in data centers.
In truth, we can already see this in Microsoft’s price increases. It will not be long with OpenAI and others. And if not directly (token prices), it will be indirectly (inflation of the number of tokens needed to execute tasks that are ultimately not that complex).
Fortunately, there are other solutions, equally effective, more economical, less energy-intensive, sovereign, and performant.
In conclusion: there are always good reasons to watch the trains go by — or to resolutely board them. This was the case for the Internet and the cloud, and it will be the same for AI. To each their own path.