AI & Defense: Have we entered a new strategic era?

In late February 2026, several international media outlets highlighted a major shift: the explicit alignment between leading American AI companies and the U.S. Department of Defense.

This is not merely a technology headline.

It is a strategic turning point.

Since the Cold War, technology has been a lever of power.

But with generative AI, we are entering a new dimension.

We are no longer speaking solely about infrastructure (networks, satellites, semiconductors).

We are now discussing systems capable of analyzing, recommending, classifying, predicting, and potentially influencing critical decisions.

In a context of intensified Sino‑American rivalry, Washington now regards AI as a national strategic asset, an approach that aligns with a historical logic.

The rise of AI is embedded within a structural technological rivalry between the United States and China. Since 2023, Washington has tightened restrictions on advanced semiconductor exports to China, explicitly targeting the computing capabilities required for AI and military applications (CNBC, October 2023). In 2024, these controls were expanded to other sensitive technologies, within an openly acknowledged containment strategy (CNBC, September 2024). On the Chinese side, the trajectory has been clear since the 2017 national plan: to become the global leader in artificial intelligence by 2030, backed by massive state support (CNBC).

American think tanks such as CSIS now analyze these measures as structuring geopolitical instruments reshaping global technological alliances.

As TIME summarizes, the “tech war” no longer concerns the economy alone, it conditions the global strategic balance:

AI is no longer simply a market. It is a lever of power at the core of national strategies.

The Pentagon: A discreet engine of american innovation

Silicon Valley is often imagined as independent from military power.

Historically, this is inaccurate.

ARPANET (1960s)

Funded by DARPA, the research agency of the Department of Defense, ARPANET became the precursor to the Internet.

GPS & Microelectronics

GPS was developed by the DoD before being opened to civilian use.
Military investments also fueled the rise of the semiconductor industry.

Cybersecurity & advanced research

Even after the Cold War, DARPA has remained a central actor in financing critical technologies.

The model has been consistent:

• High‑risk military research
• Technological maturation
• Large‑scale civilian diffusion

This framework has structured the modern digital economy.

Timeline: From the Cold War to strategic AI (1960 → 2026)

The Anthropic episode: A structural tension revealed

Before understanding the divergence with Washington, it is essential to understand who Anthropic is and why its position matters.

Who is Anthropic?

Anthropic is an American company founded in 2021 by former OpenAI executives and researchers, including Dario Amodei.

It develops the Claude model, one of GPT’s primary competitors.

In recent international media analyses (notably in France Culture’s international press review), Anthropic appears as the actor that has chosen to maintain a more restrictive stance on certain sensitive uses of AI.

The company advocates:

• The progressive reliability of models
• The necessity of human supervision in critical uses
• The refusal to broadly open its systems to certain military applications

This position is not marginal. Anthropic is backed by major technology players and is among the central companies in the global AI race.

However, regarding military applications, it has maintained clear limits.

The disagreement

According to economic and international press reports (L’Echo, France 24), Anthropic refused to relax restrictions concerning:

• Mass domestic surveillance
• Integration into fully autonomous weapons systems

This refusal contributed to its exclusion from certain federal markets, while OpenAI concluded an agreement with the Department of Defense.

What stands out in analyses by France Culture is not commercial rivalry, but what this episode reveals:

“We are entering a phase where AI is no longer merely an optimization tool, but a strategic element within a state’s architecture of power.”

Anthropic did not adopt a dramatic activist stance. It simply considered that the current maturity of systems does not justify certain critical delegations.

And this is where the debate becomes structural.

What changes in 2026

The link between technology and defense is not new. What differentiates the current moment lies in three factors:

1. The nature of the systems

We are moving from technical infrastructure to architectures of algorithmic authority.

2. Geopolitical intensity

Technological rivalry with China accelerates strategic decision‑making.

3. Speed of deployment

Generative AI evolves in cycles of months rather than decades.

Beyond the military question: governance

The real debate is not:
“Should AI collaborate with Defense?”

The real question is:
At what level of maturity do we entrust critical responsibilities to probabilistic systems?

This is a matter of:

• Robustness
• Auditability
• Supervision
• Legal accountability

And this is precisely where design intersects with technology.

What this means for companies

We are entering a phase where:

• AI becomes strategic infrastructure
• Governance becomes a competitive advantage
• The ability to render complexity understandable becomes essential

This is no longer merely a question of innovation.

It is a question of decision architecture.