Meta, Microsoft, and the Illusion of “Superintelligence for All”

Starting Point: Quarterly Figures and Big Promises

At the end of July 2025, Meta released its latest quarterly figures – alongside strong revenues (22% growth to $47.52 billion, profit increase of 36% to $18.34 billion), Mark Zuckerberg primarily delivered one message: “Superintelligence” is within reach. Meta aims to “create a personal superintelligence for all people in the world.”

In parallel, Microsoft also presented impressive figures: a revenue jump of 18% to $76.4 billion, a profit increase of 22% to $27.2 billion, and cloud growth of 25%. With these figures, Microsoft joins the four-trillion-dollar club, alongside Nvidia, while Meta moves towards a market capitalization of two trillion dollars. The AI boom is driving the stock market – and is being sold as a historical turning point.

But beyond the euphoria, the question arises: What does it mean when Big Tech talks about “superintelligence” – and what is really behind it?

What Superintelligence Means in This Context

Zuckerberg defines “superintelligence” as a form of artificial intelligence that surpasses human intelligence in all areas. It is no longer just about specialization in individual tasks (like language models or image analysis), but an AI capable of solving complex problems independently and improving itself.

The crucial point is recursive self-improvement: systems that optimize their own algorithms, adjust training methods, and enhance their performance – without direct human intervention. This has been considered a critical tipping point in research for decades, as technological advances can accelerate exponentially.

Opportunities and Risks of Recursive Self-Improvement

An AI that can improve itself changes the rules of the game.

• Opportunities: Accelerated research in medicine, energy, or climate models, automation of knowledge work, a productivity boost that could transform entire economies.
• Risks: Loss of control over systems whose internal decision logic detaches from humans. Concentration of power in a few companies with the resources to train such systems. A possible dependency on actors who are not democratically legitimized but primarily follow capital interests. For companies, this means new challenges in technology strategy.

The transition from weak to self-improving systems is not only a technological issue but also a societal and geopolitical risk.

The Economic Dimension: Market Capitalization and Investments

Meta announced additional investments of at least $66 billion in AI data centers. The goal: gigantic infrastructures that enable the training of models with billions of parameters. These sums are comparable to state infrastructure projects – here, however, entirely in private hands. For Kubernetes-based infrastructures, this means new scaling requirements.

Microsoft, for the first time, provided concrete figures for Azure: $75 billion annual revenue – a growth of 34%. This confirms that Microsoft is the number two in the cloud market behind AWS. Together with Nvidia, which is now also worth over four trillion dollars, these companies form an oligopoly dominating global AI development.

What Is Missing: Regulation and Control

While the US government (under President Trump) has just relaxed regulatory requirements for AI to ensure “competitiveness,” China proposes a global framework. Europe discusses ethics and acceptance, while technological development accelerates.

The central question is: Who controls systems that can improve themselves?

So far, there are no internationally coordinated mechanisms, no binding standards for auditability or security protocols. Development is effectively determined by the investment decisions of a few corporations.

A Look into the Future

When Zuckerberg talks about a “personal superintelligence for all,” it is more than marketing. It is a claim to power: Meta wants to provide the operating system of the future, the interface between individual intelligence and machine superiority.

But the danger is that we are moving towards a technology whose dynamics are no longer fully controllable. Recursive self-improvement could bring advances that surpass human development cycles – and drive us into a dependency on systems whose logic we no longer understand.

Conclusion

Meta and Microsoft celebrate quarterly figures, investors celebrate the AI boom, and politics discusses ethical guidelines. But the real question remains: What happens when we develop AIs that improve themselves – and we no longer set the pace?

Superintelligence may shine as a vision. As a reality, it could become the greatest challenge of our time.