Databricks co-founder and CEO Ali Ghodsi said on June 16, 2026 [1], that AI lacks the necessary context to reach artificial general intelligence.

This perspective shifts the debate over AGI from a question of raw computational power or algorithmic "intelligence" to one of data accessibility. If the primary barrier to general intelligence is context rather than capability, the focus of the industry may move toward how companies organize their proprietary data.

During a Bloomberg Tech interview, Ghodsi discussed the gap between current AI capabilities and the goal of AGI [1]. He said, "AI is already smart enough, but it lacks the right context to reach the level of AGI" [1]. According to Ghodsi, the path to a general-purpose system relies on providing AI with richer contextual data, and more effectively structured databases [1].

There is a slight contradiction in how Ghodsi characterized the current state of the technology across different platforms. While the Bloomberg interview suggested that AI still needs context to reach AGI, he said to MSN that "we have already reached artificial general intelligence, but the AI is just lacking context to be more productive" [2].

Regardless of whether the threshold has been crossed, Ghodsi said that the utility of these systems depends on their ability to understand the specific environment in which they operate. He said that the intelligence is present, but the application is limited by the information available to the model at the time of the request [2].

This approach highlights a growing trend in the tech sector toward retrieval-augmented generation and other methods that ground AI in real-time, specific data. By focusing on the infrastructure of data, companies like Databricks aim to bridge the gap between a model that can reason and a model that can execute complex, real-world tasks accurately [1].

“AI is already smart enough, but it lacks the right context to reach the level of AGI.”

Ghodsi's comments suggest that the 'intelligence' component of AI has plateaued or reached a sufficient level, moving the bottleneck to the data layer. This implies that the race for AGI may no longer be about building larger models, but about building more sophisticated data pipelines that allow models to access and utilize specific organizational knowledge.