The current discourse around artificial intelligence often oscillates between breathless predictions of transformative impact and cautious warnings of an overheated "bubble." Yet, as Databricks CEO Ali Ghodsi articulated in a recent CNBC interview, the real story lies in the tangible, albeit often overlooked, foundational shifts enabling AI's practical integration into enterprise operations.
Ghodsi spoke with CNBC’s Deirdre Bosa and Jon Fortt on ‘Closing Bell Overtime,’ discussing Databricks’ robust Q2 performance, the nuanced reality of AI adoption rates among large enterprises, and the critical role of secure data management in unlocking AI’s full potential.
Databricks, a leading data and AI company, reported impressive Q2 milestones, including a $4 billion revenue run-rate—a substantial 50% year-over-year increase—and AI products alone crossing a $1 billion revenue run-rate. This financial strength is underpinned by a fresh $1 billion funding round, pushing its valuation past $100 billion. Ghodsi directly addressed concerns about an AI "bubble" or over-investment, acknowledging past over-promises. However, he emphasized that "just now the actual use cases are starting to work and they're actually starting to get deployed," signaling a maturation of AI applications from theoretical potential to concrete business value.
The company's growth is largely fueled by two key product areas. Agent Bricks, for instance, focuses on developing AI agents that automate "everyday tasks" in the workplace, diverging from the industry’s initial fixation on "super intelligence" challenges like Math Olympiads. This strategic pivot towards practical, immediate utility is resonating with customers. The second product, Lakehouse, is a database specifically designed to support these AI agents, recognizing that AI-driven processes utilize data differently from human users.
This perspective stands in contrast to recent research, such as that by Torsten Sløk of Apollo, which suggests a decline in AI adoption among large firms with over 250 employees. While such data might paint a picture of investor skepticism or a plateau in initial enthusiasm, Ghodsi’s insights underscore that the nature of AI adoption is evolving, moving beyond experimental phases to more deeply embedded, revenue-generating applications within the enterprise. Databricks’ own positive free cash flow over the last twelve months further supports the notion that practical, value-driven AI solutions are gaining traction, even as the broader market adjusts its expectations.
Delving deeper into the challenges, Jon Fortt raised the issue of data silos and unstructured data as potential bottlenecks for AI adoption in large, established enterprises. Ghodsi, however, pinpointed a more fundamental impediment: "It's actually the governance of the data that's the problem. It's not like we can't access that data, that data is there... It's because of privacy." Corporations are acutely concerned about protecting their most sensitive data from breaches or leaks, a significant hurdle that must be overcome before AI systems can fully leverage this valuable information. The fear of "getting hacked" or experiencing a "leak" is actively "hampering access" to data that could otherwise power transformative AI.
This critical focus on data privacy and security directly informs Databricks’ product strategy. Their Unity Catalog solution is explicitly designed to govern data in a secure, safe, and privacy-preserving manner, acting as a foundational prerequisite for successful AI implementation. Without robust governance, the potential ROI from AI remains elusive due to inherent risks and regulatory compliance concerns. This approach enables enterprises to manage who accesses what data and how, fostering trust and enabling broader AI deployment.
Ghodsi further elaborated on a profound shift in the database market. Traditionally, humans (database administrators) created and managed databases. However, he revealed that "for the first time, over 80% of the databases are being created by AI agents." This signifies a revolutionary change, where enterprises are increasingly using AI for "vibe-coding" – building their own software in-house, thanks to AI dramatically reducing development costs. These AI-generated applications, in turn, demand purpose-built databases like Databricks' Lakehouse to support their unique operational needs, fundamentally disrupting the established database landscape. This trend suggests a future where software development becomes significantly more automated and tailored, driven by AI's capabilities.
Despite its rapid growth and substantial valuation, Databricks is not rushing towards an initial public offering. Ghodsi stated that the company operates "like a public company" with rigorous financial audits but is content to remain private for now, driven by the immense demand for its software in the burgeoning generative AI space. This strategic patience allows Databricks to continue focusing on product innovation and addressing the core challenges of enterprise AI without the immediate pressures of public market scrutiny, ensuring a deliberate and sustainable path forward in a dynamic industry.
