Frontier Radar #5: If AGI is so close, why is everyone leaving? The neolabs and their bets
What happened
OpenAI and Nvidia executives are signaling that artificial general intelligence (AGI) is imminent or already here. Yet many of the engineers and researchers behind today’s leading AI models are leaving OpenAI, Google DeepMind, and Meta to launch independent labs. These “neolabs” have collectively raised billions of dollars despite showing little in the way of finished products. They are betting the current widely used training methods won’t be enough to reach the next AI breakthrough.
Why it matters
The exodus exposes growing skepticism about the current AI development path. While today’s models excel at specific tasks, major limitations remain. Neolabs aim to tackle fundamental technical barriers, such as scaling costs, model robustness, and reasoning abilities. Their divergence signals that investors and top talent expect the next advances in AI will require new architectures, training methods, or hardware designs beyond the industry incumbents’ roadmaps.
This trend pressures the established AI giants by threatening to fracture talent and funding pools. It also raises costs and risks for startups chasing the current recipe with no guarantee it leads to AGI. For builders and investors, it spells a more fragmented AI ecosystem where no single approach dominates yet. The market becomes less predictable as multiple experimental bets compete, making strategic planning tougher for companies banking on one technology path.
What to watch next
Track the neolabs’ research outputs and funding rounds closely. Their early work may signal which technical approaches will gain traction or fade away. Watch Nvidia’s moves since their hardware leadership strongly influences the cost and feasibility of training larger AI models. If neolabs start shipping meaningful improvements or new models, it could reshape where AI innovation happens and who captures the economic value.
Meanwhile, keep an eye on talent flow between established AI companies and neolabs. That movement reflects where confidence lies in solving today’s AI bottlenecks. The next few years will reveal if these labs finally break free from simply replicating current models and unlock the true promise of AGI or if the industry consolidates around the current tech roadmap.
AI Quick Briefs Editorial Desk