960 billion-yuan AI unicorn is set to secure a new financing round

Latest news: AI data unicorn Databricks is seeking funding again.

This time, its maximum possible valuation could reach $175 billion (equivalent to approximately 1.26 trillion RMB).

With SpaceX going public and OpenAI and Anthropic secretly submitting their prospectuses, Databricks may be the last giant AI unicorn still active in the primary market.

Databricks is an AI company that helps enterprises manage data. It raised $5 billion in February and currently has a valuation of $134 billion (equivalent to approximately 964.8 billion RMB).

01 Making Corporate Private Data "Valuable"

The story of Databricks began with a piece of code in the Berkeley Lab.

In 2013, several researchers from the AMPLab at the University of California, Berkeley founded Databricks. Their core technological asset is Apache Spark, a software engine that enables hundreds or thousands of servers to process massive amounts of data simultaneously.

The relationship between Apache Spark and Databricks. Source: Public information

If data is compared to ore, Spark is the mining machine, and Databricks has built an entire modern mine.

Internet companies, banks, retailers, and automotive companies generate massive amounts of data every day. User clicks, transaction records, inventory changes, sensor signals, log files, customer profiles, and advertising campaign results are all stored in the system. The problem is that the more data there is, the more difficult it is to process. Databricks helps enterprises extract value from their data.

After the AI boom, this role suddenly became extremely important.

Because large models themselves do not understand the specific business of a company. They don't know which areas of a retailer have insufficient inventory today, which transactions of a bank are abnormal, or which batch of battery test data of an automotive company has problems. For a model to truly work for an enterprise, it must access the enterprise's internal data.

However, enterprise internal data is often in a mess.

Some data is stored in the cloud, while some is on local servers. Some data is in data warehouses, and some is in business systems. Some data is in structured tables, while some is in the form of customer service recordings, contract texts, pictures, and logs. What's more troublesome is that not all data can be freely used by AI. The finance, healthcare, manufacturing, and retail industries all have strict requirements for permissions, security, and compliance.

This is exactly where Databricks sees an opportunity.

It can tell enterprises: You don't need to move all your data again, nor do you need to build AI infrastructure from scratch. You can manage data, train models, deploy AI applications, and establish governance rules on a unified platform, enabling AI to truly utilize the company's own data.

In the AI era, the most expensive thing may not be the model itself, but the connection layer between the model and real business. This is exactly what Databricks is doing.

02 Annual Revenue of $5.4 Billion

Databricks' way of making money is different from that of traditional software companies.

Traditional software is more like selling licenses. Enterprises buy a system and pay an annual fee, and their employees can use it. Databricks is more like a cloud computing company. Customers don't simply buy a software account; instead, they process data, train models, run AI applications, and call computing resources on its platform. The more they use, the higher the bill.

Databricks Data Intelligence Platform. Source: Databricks official website

This is also what makes Databricks most attractive.

An enterprise may initially use it for data analysis. For example, it can upload sales, inventory, order, and user behavior data to the platform to generate reports, check trends, and predict demand. Later, the enterprise starts training machine learning models. Then, with the arrival of the AI era, the enterprise wants to develop AI assistants, AI agents, intelligent customer service, and risk control systems based on its own internal data. Each additional scenario leads to an increase in the usage of Databricks.

Therefore, what Databricks sells is not a one - time software, but a set of "enterprise data and AI infrastructure".

Its revenue growth comes from two sources.

First, there is an increase in new customers. More and more large enterprises need to organize their data and build AI capabilities, so they will purchase Databricks.

Second, existing customers use more. This is even more crucial. Databricks' disclosed net revenue retention rate exceeds 140%, which means that if a group of existing customers spent $100 last year, they may spend more than $140 this year. For investors, this is a very impressive indicator. It shows that customers don't just try it out and then stop; instead, they use it more deeply and at a higher cost.

There is a strong business logic behind this.

Once an enterprise's data is connected to Databricks, it's not just about uploading a few tables. Instead, data pipelines, permission management, analysis models, and AI application development processes are all built on it. The sales department, finance department, customer service department, and R & D department are all using it. As the amount of data and AI applications increases, the migration cost also becomes higher.

This creates strong customer stickiness.

One important reason why investors continue to give Databricks a high valuation is that it has proven that it can not only tell AI stories but also actually make money.

The company has disclosed that its current annual revenue has exceeded $5.4 billion. More importantly, many customers spend more and more money after their first purchase. Because once an enterprise's data, AI models, and business systems are connected to Databricks, new usage scenarios will continue to emerge.

For example, a retail enterprise may initially use it to analyze sales data. Later, it starts training AI models, deploying intelligent customer service, and developing AI assistants. Each new function incurs more costs.

This means that Databricks doesn't rely on constantly finding new customers to make money; instead, existing customers keep increasing their spending on their own.

Currently, Databricks has more than 800 customers with an annual consumption of over $1 million and more than 70 customers with an annual consumption of over $10 million. For an enterprise software company, this shows that it has entered the core systems of many large companies, rather than being just an optional small tool.

This is also the business model that investors like the most: customers can't do without it, revenue continues to grow, and there is even more room for growth as AI becomes more widespread.

03 Becoming the AI General Manager for Enterprises

In the past, enterprises bought Databricks mainly for data processing.

For example, a retail company wants to know which stores are selling well, which products have overstocked inventory, and which customers are likely to churn. It can put its sales, inventory, membership, and logistics data into Databricks and then let the data team conduct analysis.

This is still the business of a traditional data platform.

However, after the emergence of AI, Databricks' goal has changed. It not only wants to help enterprises "understand data" but also wants to help them "use AI to mobilize data".

"Understanding data" is mainly used by data analysts, engineers, and business leaders. It solves problems related to reporting, prediction, and analysis.

"Using AI to mobilize data" means that every ordinary employee can directly interact with the company's data. Salespeople can ask: What has this customer bought in the past? Customer service representatives can ask: Why is this user's order delayed? Financial staff can ask: Which expenses are abnormal? Supply chain personnel can ask: Which warehouse may be out of stock?

Databricks official website: Your data. Your AI. Your future. Source: Databricks official website

In the past, these questions required the data team to write SQL, generate reports, and create dashboards. In the future, Databricks hopes that AI agents can directly handle them.

This is why it has launched products such as Genie One and Agent Bricks. Databricks doesn't want to be an ordinary chatbot; instead, it wants to be an AI assistant that can access an enterprise's real - world data, understand the business context, and help employees make decisions.

In other words, OpenAI and Anthropic are building general - purpose large models. Databricks wants to build "business - savvy AI" within enterprises.

No matter how powerful a large model is, if it can't access an enterprise's internal data, it's just an external tool. No matter how advanced an AI agent is, without permission management, data governance, cost control, and a security system, it's difficult to enter the core business. Databricks wants to cover all these aspects and become the unified operation layer for enterprise AI.

It can develop AI assistants at the upper level, allowing employees to directly interact with the company's data.

It can build databases at the lower level, connecting business systems and AI systems.

It can horizontally enter scenarios such as marketing, security, customer service, and developer tools.

It can also manage AI costs. As enterprises use a large number of AI agents, the bills become increasingly difficult to predict. An employee, an agent, or an automated process may continuously call models in the background, resulting in huge expenses. Databricks' launch of AI expenditure control tools is essentially an attempt to become the "main gate" for enterprise AI budgets.

This is similar to the early days of cloud computing. At first, enterprises simply moved their servers to the cloud. Later, cloud providers not only sold servers but also databases, data warehouses, AI services, security services, development tools, and cost - management tools. The more customers use, the harder it is for them to leave.

Databricks also wants to follow this path.

This article does not constitute any investment advice.

This article is from the WeChat official account “Pencil News” (ID: pencilnews). The author is Pencil News, and it is published by 36Kr with authorization.