Nvidia Is Moving Beyond LLMs to Superlearners, the Possible Precursor to AGI. What That Means for NVDA Stock.

Jabran Kundi

Tue, May 19, 2026 at 1:54 PM EDT 4 min read

Large language models (LLMs) have dominated much of the AI debate over the last few years. Scaling token prediction and LLM training were considered reliable metrics to measure progress, irrespective of the cost. This factor played right into the hands of Jensen Huang, who makes the best GPUs in the world, a fundamental requirement for training these AI models.

As intelligent as these systems were, they were still dumb in the sense that they had to be fed data in order to become better at predicting the next token. These static datasets could only make the AI so intelligent, giving rise to the need for superlearners: AI systems that continuously learn from experience rather than static datasets.

More News from Barchart

• Dear Sandisk Stock Fans, Mark Your Calendars for May 20

• How Beaten-Down Tempus AI Stock Offers a Lottery Ticket for Traders Here

• NVDA Earnings Bull Put Spread has a High Probability of Success

• Our exclusive Barchart Brief newsletter is your FREE midday guide to what's moving stocks, sectors, and investor sentiment - delivered right when you need the info most. Subscribe today!

Nvidia (NVDA) announced a new engineering collaboration with a London-based startup called Ineffable Intelligence. The startup is led by David Silver, the same man behind the success of DeepMind and AlphaGo. Nvidia is now backing the idea of building an AI that discovers knowledge through interaction rather than just pre-training. This is just another step closer to AI, and one can already imagine the kind of progress a system like this would make in fields such as drug discovery, climate control, cybersecurity, and pretty much any field that progresses based on trial and error.

Huang has already dubbed superlearners as the “next frontier of AI.” By backing Ineffable Intelligence, he is trying to secure a foothold in the technology of the future. He already did this with LLMs. Repeating the same with superlearners could provide an even bigger growth story. That’s because the continuous and real-time feedback that such systems require will test memory bandwidth and interconnects far more than the current systems do. For now, Nvidia’s Grace Blackwell will power the research performed by Ineffable Intelligence, later moving to the Vera Rubin Platform. What comes after that is anybody’s guess, but if Jensen Huang is backing it, you can bet it will be powered by his firm.

About NVIDIA Stock

Nvidia is a fabless semiconductor and AI computing company that designs GPUs, AI accelerators, application programming interfaces (APIs), and system-on-a-chip units. The company operates through the Graphics and Compute & Networking segments. Through its CUDA ecosystem, the company enables industries ranging from autonomous vehicles to scientific research by advancing AI, accelerated computing, and data center infrastructure.