Databricks has closed a financing round of more than 7 billion dollars (5,000 million in capital and 2,000 million in debt) that values the company at 134 billion dollars. It’s a dizzying figure for a company that the vast majority of people have never heard of.
The San Francisco firm is not, technically, an AI company either. Its business is enterprise-scale data management and analysis. What Databricks does is provide the invisible infrastructure that allows other companies to store, process and extract value from enormous amounts of information.
Without that, training AI models would be impossible.
Why is it important. Databricks is the cover of the boom of AI. OpenAI, NVIDIA or Google grab the headlines, but it’s companies like this that build the plumbing that makes everything else possible.
Its valuation is 134,000 million. Without ever having gone public. That places it even above established technology giants. It is at the level of Qualcomm or Sony. Beats Xiaomi or Adobe. And it does so with a less business model sexy but more profitable: B2B infrastructure than it leaves gross margins greater than 80%.
In figures. The Databricks numbers They explain a growth that justifies the enthusiasm of its investors.
- Annualized revenues exceeding $5.4 billion in the fourth quarter, with 65% year-over-year growth.
- More than 800 clients that generate more than a million dollars annually.
- Positive free cash flow over the last year.
- Its AI product line has surpassed $1.4 billion in revenue with a net retention rate of over 140%.
Between the lines. The participation of JPMorgan Chase, Goldman Sachs, Morgan Stanley, Microsoft and sovereign funds like Qatar’s in the latest round says a lot: these large investors are betting on the infrastructure, not the final application.
The implicit message is something we’ve been hearing since the first few months after the ChatGPT moment: in the AI race, those who sell picks and shovels can earn more than those who pan for gold. Databricks provides the platform where companies store their proprietary data and train their custom modelssomething that the public APIs of OpenAI or Anthropic cannot offer.
Yes, but. Its CEO, Ali Ghodsi, has said that “now is not a good time to go public,” even though his company meets all the financial requirements to do so. The strategy is to accumulate enough cash enough to withstand any market correction like the one in 2022.
And seen the vertigo it produces any headlines on capex figuresit makes sense to make a cushion for what may happen.
The context. Databricks represents an important change in how the technology sector is structured.
- For years, traditional SaaS companies dominated the B2B landscape.
- Now, AI infrastructure and data platforms are achieving similar or higher valuations.
The company is also expanding beyond its traditional business with products such as Lakebase, a database designed specifically for AI agents. Or with Geniea conversational assistant that allows employees to query business data using natural language.
- If Databricks achieves a strong IPO in an environment where technology valuations are more closely monitored than ever, it would demonstrate that markets are willing to pay very large premiums for AI infrastructure, not just flashy models.
- And that would change the rules of the game for dozens of similar companies operating in the shadows.
In Xataka | Spain, on the verge of adding another AI unicorn: Multiverse negotiates a round to exceed 1.5 billion euros
Featured image | Databricks and Xataka with Mockuuups Studio
GIPHY App Key not set. Please check settings