Databricks hits a $188B valuation by selling the AI story above enterprise data

Databricks announced a new funding round led by Coatue at a $188 billion valuation. TechCrunch says the company did not disclose the exact amount and expects the round to close later this summer, while other outlets cited by the article put it at roughly $3 billion.

The valuation jumped from $62B to $188B in eighteen months

The pace is the point. In December 2024, Databricks raised $10 billion at a $62 billion valuation. In September 2025 it raised $1 billion at $100 billion, followed in February 2026 by another $5 billion at $134 billion. Now it is communicating $188 billion.

TechCrunch frames this as the company’s second act: Databricks grew from big data and analytics, then remade its image as an AI company after ChatGPT. That is not only cosmetic. Databricks sits close to enterprise data, where buyers care about security, governance and real deployment.

The AI halo works because Databricks already holds the data layer

For enterprise customers, the biggest question is not which model wins a benchmark. It is where the model gets safe access to data, who controls permissions and how agentic workflows are audited. Databricks is selling that position through Lakebase for AI agents, Unity as an AI gateway and Omnigent for managing multiple agents.

The second layer is cost. TechCrunch says Databricks is championing open-weight models, including Z.ai’s GLM 5.2 for coding, as a way to control spending. Its internal benchmark for 3,000 software engineers argued that model choice is only part of cost and that the harness, such as Codex, Claude Code or open-source Pi, heavily affects context use.

A cheaper model will not save the bill if the harness leaks context

That is a useful correction to the simple open versus proprietary debate. If the agentic harness cannot manage context, plan steps and reduce unnecessary model calls, cheaper tokens disappear quickly in production use.

Caution still matters. A vendor’s internal benchmark is an interesting signal, not neutral market truth. Databricks has a clear reason to argue that AI belongs to platforms that combine data, governance, model routing and cost control.

The market will watch revenue, not the next round letter

The next proof will not be another meme about Series AA. It will be customer revenue from AI products, repeatable savings on real coding workloads and the ability to keep security promises when agents operate on sensitive data.

If Databricks is only carrying an old data platform on an AI valuation, the market will sober up. If it really controls the point where enterprise data meets agents, $188 billion is a bet on a checkpoint rather than on one model.

Lilith's verdict

Databricks is standing at the turnstile between enterprise data and agents, charging tolls. Investors are not only betting on a model. They are betting on the gate every outside model and internal dataset has to pass.