Nvidia Partners With AI Startup Ineffable Intelligence on Reinforcement Learning Systems

Nvidia has announced a partnership with Ineffable Intelligence, an AI startup founded in late 2025 by a former lead of Google DeepMind's reinforcement learning team. The London-based company is pursuing superintelligence through systems that learn by trial and error.

The two organizations will enter an engineering-level collaboration to build AI systems that learn continuously from experience, the company said on Wednesday.

Ineffable Intelligence announced a record $1.1 billion seed round in April, co-led by the U.S. Sovereign AI Fund. The collaboration will focus on reinforcement learning, in which AI models learn from experience rather than being trained primarily on human data.

The system will train on rich forms of experience that are distinct from human language and other human data, and may require novel model architectures and training algorithms, the company said. The work will center on building a pipeline that can feed reinforcement learning systems at scale. Engineers from both companies will team up on the project.

It will use Nvidia's Grace Blackwell chips alongside its Vera Rubin platform. "The next frontier of AI is superlearners — systems that learn continuously from experience," said Nvidia CEO Jensen Huang. " Researchers have largely solved the easier problem of building systems that know what humans already know, according to the company's statements.

The harder challenge is developing systems that discover new knowledge independently. Ineffable Intelligence is among several AI labs launched in recent months by former researchers from major technology companies, which have attracted billions in investment. On Wednesday, another months-old startup announced it had raised $650 million.

A separate lab announced a $1 billion raise in March. In the past year, former staff from several leading AI organizations have raised hundreds of millions of dollars for new ventures.