Four-month-old startup raises $650 million for AI that improves itself
The startup raised $650 million to develop self-improving AI that uses its own improvements to accelerate further development in an accelerating cycle. The syst
A startup has just closed a $650 million funding round to develop self-improving AI — a system that enhances itself in an accelerating feedback loop. For the first time, a fantastical concept from 1960s academic papers is receiving serious investment and transitioning from theory to actual engineering projects.
From Theory to Practice
The concept of "recursive superintelligence" has been floating around in computer science and science fiction for over six decades. The idea is simple: an AI system analyzes its own code, finds ways to become more powerful or faster, implements improvements — and immediately uses the improved version of itself to search for the next optimizations. If this works, a positive feedback loop emerges, where each development cycle accelerates the next one. Until now, this has been a convenient metaphor for discussions about the distant future of AI. Now, however, major investors are willing to pay billions to test whether this actually works in practice.
How It's Supposed to Work
In an ideal scenario, the system operates like this: first, human researchers write a basic self-learning algorithm. Then this system looks at itself as an object of analysis and finds ways to improve its own architecture, computational speed, or reasoning logic. Each improvement is registered, tested, and evaluated. If the result is better, the change is retained. The system moves to a new level of performance and begins the entire process anew — but now with a more powerful mind. In theory, the cycle accelerates exponentially.
- System analyzes its own source code
- Identifies bottlenecks and inefficiencies
- Designs and implements improvements
- Tests results in a controlled environment
- Repeats the process with the improved version
Why This Raises Concerns
The $650 million funding symbolizes a turning point: a transition from philosophical debates to engineering reality. But it also intensifies long-standing concerns among AI safety researchers. If a self-improving system becomes truly powerful and spirals out of control, its consequences would be qualitatively different from a bug in an ordinary application. Questions arise right now: how can we ensure the system remains controllable at each stage of self-improvement? How do we prevent an "escape" at the stage when the system becomes smarter than its creators? This is precisely why part of the funding goes not only to development but also to safety research.
What This Means
We stand at the threshold of a new chapter in the history of AI. Startups are no longer waiting for approval from academic institutions. They are investing in self-improving systems — this is either a revolution in our understanding of automation and development, or an intensive test of our ability to maintain control over increasingly powerful systems that we create.