-+ 0.00%
-+ 0.00%
-+ 0.00%

Citi: The Databricks conference confirmed the acceleration of corporate AI diffusion and recommended Microsoft (MSFT.US), Snowflake (SNOW.US), Palantir (PLTR.US), and MongoDB (MDB.US)

Zhitongcaijing·11/20/2026 12:09:09
Listen to the news

The Zhitong Finance App learned that after Citibank summed up the key points of participating in the Databricks Data+ AI World Tour event in New York, its core judgment was that the spread of enterprise-grade AI is still in its early stages, but it is accelerating; at the same time, cost optimization has become a dark line throughout.

Citi maintains a “buy” rating for all four companies covered: Microsoft (MSFT.US) target price of $680, MongoDB (MDB.US) target price of $565, Palantir (PLTR.US) target price of $265, and Snowflake (SNOW.US) target price of $490.

Citi pointed out that communication with customers and partners surrounding Databricks shows that data and AI consumption is accelerating, driven by Genie, multi-agent architectures, and continued migration of legacy platforms. Genie is a key growth engine, and feedback from customers and partners is particularly prominent in buyer scenarios such as CFO offices; Unity Gateway has won the favor of companies managing multiple models and high token usage through integrated model access, routing, monitoring, and expenditure control.

A set of data outlines progress on the adoption side: more than 70% of the data pipelines have been written by Genie Code, Genie's accuracy rate on actual workloads has exceeded 80%, and there are hundreds of thousands of active users in some organizations. In terms of performance, the Reyden engine has reduced query delays by up to 50% for a financial services customer during the private preview phase, and is expected to be officially launched in January next year. After a major media company adopted a Lakebase-supported machine learning feature library, the number of models trained by the same team tripled and the number of experiments increased sevenfold.

However, Citi also suggested that the topic of cost optimization is ubiquitous. A FinOps partner that manages approximately $400 million in Snowflake and DataBricks spending revealed that the automated savings that customers can achieve on Databricks are about 20%, higher than Snowflake's 10% to 13%, mainly from shutting down or adjusting inefficient resources; traditional data workloads are still the biggest savings pool. Another migration partner described that the spending structure of a large media customer has evolved from 90 percent Snowflake and 10 percent Databricks a few years ago to today's 55 percent, and Snowflake's spending is relatively flat. Citi believes that cost review and optimization trends are worth wary of, particularly focusing on native AI vendors and large-scale language models.

In the competitive landscape, Citi noticed the competitive relationship between Databricks and Palantir — some customers use Databricks to manage core data and Palantir to carry ontologies or applications. Palantir's explicit modeling body still has a competitive advantage over Databricks's deductive ontology.

Overall, Citi believes that although cost optimization may suppress consumption growth in stages, the accelerated trend of enterprise AI diffusion and the implementation of multi-model and multi-agent architectures still bring positive demand signals to the software sector.