NVIDIA has announced a partnership with Ineffable Intelligence, a London‑based AI startup, to build new AI systems that learn through trial and error, not just from human‑written text or data. The collaboration focuses on “reinforcement learning,” where AI agents improve by interacting with environments, getting rewards or penalties, and adjusting their actions over time.
Ineffable Intelligence, founded by David Silver—one of the key scientists behind Google DeepMind’s AlphaGo—specialises in this kind of self‑learning AI. The startup raised about $1.1 billion in seed funding in April 2026, in a round co‑led by Sequoia and Lightspeed, and valued the company at roughly $5.1 billion. NVIDIA and Ineffable will co‑design large‑scale infrastructure so these reinforcement‑learning systems can train faster and more efficiently.
The project will run on NVIDIA’s Grace Blackwell chips and the upcoming Vera Rubin platform, which are built for heavy‑duty AI workloads and memory‑intensive training. CEO Jensen Huang has called this “the next frontier of AI,” where systems become “superlearners” that keep discovering new knowledge instead of only copying what humans already know. If the approach works, it could help create smarter AI agents for robotics, simulations, complex decision‑making, and scientific discovery.
Industry experts say reinforcement‑learning AI is still early stage but could reshape how machines handle real‑world problems that cannot be solved by static data alone. The NVIDIA–Ineffable deal signals a major bet on this direction, adding the UK startup to a growing list of European AI labs that NVIDIA is backing.
FAQs [Frequently Asked Questions]
1. What does “AI that learns by trial and error” mean?
It means AI improves by trying actions, seeing what works, and adjusting over time, instead of only reading human‑written text or data. This is known as reinforcement learning.
2. Who is Ineffable Intelligence and why is it important?
Ineffable Intelligence is a UK AI startup founded by David Silver, a former DeepMind scientist. It focuses on next‑generation reinforcement‑learning systems and raised around $1.1 billion in seed funding.
3. How will NVIDIA benefit from this partnership?
NVIDIA will help design special hardware and software for large‑scale reinforcement learning, so more AI workloads run on its chips and clouds, strengthening its position in the future AI market.