Databricks CEO: We don’t need AI to be smarter, we just need longer context.

```

Ali Ghodsi, Databricks co-founder and CEO, believes that artificial general intelligence (AGI) has already arrived, but the industry's focus has gone astray—the breakthrough for AI models is not to make them "smarter," but to provide sufficient data context.

At the recent Bloomberg Tech event, Ghodsi shared his views on the current state of AI development. He said that whenever he asks the audience "how many people think AGI has already arrived," usually only about 10% raise their hands; but when he rephrases the question—"how many people think the AI models they use every day are smarter than the people interacting with them"—the proportion rises to about 90%. Ghodsi believes this paradox shows that AI already possesses tremendous capabilities; the issue is not with the models themselves, but how to pair them with the right data.

This perspective directly impacts how enterprises choose their AI investment paths. He pointed out that AI can generate code and automate tasks, but true business value comes from implanting these capabilities into concrete business scenarios and datasets. "We don’t need AI to be smarter, it just needs to be given context," he said. This also means that for enterprise customers, the value of building data infrastructure may surpass ongoing investments in cutting-edge models.

Data Context: The Key to Unlocking AI Potential

Ghodsi’s judgment on AGI is itself challenging. He defines AGI—whether it has arrived—by the dimension of "intellectual capability," i.e. whether AI is smarter than humans, rather than relying on the strict, yet contentious, academic definition. He believes that when 90% of users admit AI has surpassed humans intellectually, continuing to ask "when will AGI arrive" becomes practically meaningless.

His fundamental argument is that AI models are already strong enough; the industry should not overly focus resources on developing more complex models, but instead should provide existing models with relevant data. In his view, AI models are like talented students who need suitable textbooks and curricula to realize their potential, rather than merely enhancing their innate talent.

Based on this logic, Databricks centers its strategy on helping companies inject their proprietary data into AI models. Ghodsi stated the company is collaborating with numerous clients and conducting substantial research, finding that enterprises are increasingly eager to integrate vast datasets with AI to drive business value.

He emphasizes that the real power of AI lies in its combination with specific business contexts and data. The Databricks platform is positioned to play the role of a "comprehensive educational resource"—not only offering data storage and management capabilities but also striving to build an intelligent data system that allows AI models to take root and flourish in enterprise settings.

Collaboration Patterns and Customer Base

Ghodsi also revealed Databricks’ collaboration layout within the AI ecosystem. The company currently partners with Alphabet's Google DeepMind (Gemini), Microsoft-backed OpenAI, and Anthropic, providing a data foundation for these models so they can be trained and fine-tuned based on specific organizational datasets.

On the customer side, Ghodsi presented a roster including major enterprises such as AT&T, Rivian, Adidas, Mercedes-Benz, Unilever, Virgin, and Bayer to demonstrate the wide applicability of its data and AI solutions. He positions Databricks as the data backbone for enterprise AI adoption across industries, not merely as an AI model developer.

On the market side, Ghodsi admitted that the current IPO market for software companies is experiencing a slowdown, but he believes Databricks’ fundamentals are not dependent on this. He stated that the AI era has a fundamental need for data management; as more AI agents are deployed and more software is generated, demand for Databricks’ robust data infrastructure will only continue to grow.

He predicts that in the next nine months to two years, the scale of software output will surpass the sum of all previous human history, viewing this trend as the core driver for long-term expansion of data infrastructure demand. In his view, this structural backdrop is enough to keep the company resilient as the IPO window narrows.

Risk Warning and DisclaimerThe market carries risks, and investments should be made with caution. This article does not constitute personal investment advice and does not take into account any individual user's special investment objectives, financial situation, or needs. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investments based on this are at your own risk. ```