Key Takeaways

  • Databricks just commanded an $188 billion valuation — triple its price from December 2024 — without the cash even in hand yet.
  • The company's pivot from big-data plumbing to AI infrastructure turned a mature SaaS business into the hottest proxy for enterprise AI adoption.
  • CEO Ali Ghodsi's own benchmarks show Chinese open-weight model GLM 5.2 matching proprietary giants on hard coding tasks at lower cost, validating the open-model thesis Databricks has championed.
  • The funding frenzy reflects investor desperation for AI exposure more than Databricks' fundamentals; the company is raising because it can, not because it needs to.

Databricks didn't wait for the wire to clear. It announced a $188 billion valuation on Thursday — led by Coatue, roughly $3 billion fresh — while admitting the money won't land until summer. That's the move of a company that knows its stock is currency. Five months ago the price was $134 billion. Five months before that, $100 billion. Nine months before that, $62 billion. The trajectory is vertical. The alphabet is exhausted; memes now joke about Series AA.

This is not a startup story. Databricks was founded in 2013. It won the big-data era, building the plumbing that let enterprises store massive datasets in the cloud and query them fast. That business was real, profitable, and thoroughly BC — Before ChatGPT. Most companies from that vintage are now legacy line items. Databricks refused the fate.

The pivot worked because the asset was already there. Enterprises don't just want AI; they want AI that respects the same security, governance, and compliance frameworks they built around their data. Databricks sat on that data. It didn't need to acquire a position. It just needed to build upward.

And build it did. Lakebase, a database engineered for AI agents. Unity, an AI gateway. Omnigent, a meta-harness orchestrating multiple agents. The product cadence is relentless. Each release reinforces the narrative: Databricks is where enterprise AI runs.

The narrative has a specific flavor. Databricks became the standard-bearer for Chinese open-weight models — code published, free to modify, dramatically cheaper than proprietary alternatives. It championed Z.ai's GLM 5.2 for coding workloads. Last week Ghodsi published internal benchmarks run on his own 3,000 engineers. The result: open models, GLM 5.2 included, now handle the hardest coding tasks at lower total cost than Anthropic or OpenAI. That claim matters. It comes from a buyer, not a vendor.

The surprise buried in the same post: the harness matters more than the model. The agentic coding tool wrapping the model — Codex, Claude Code, or Databricks' own — swung outcomes more than the underlying weights. That insight reframes the competitive landscape. Model access is commoditizing. The differentiation lives in the orchestration layer.

Investors are not funding roadmap. They are buying exposure. The round was oversubscribed; the company announced early because it had no reason to hide the number. Every fund needs an AI winner in the portfolio. Databricks is liquid, legitimate, and already inside the enterprise perimeter. It is the safest proxy on the market.

That safety has a price. $188 billion prices in perfection. The revenue multiple is stratospheric. Growth must stay exponential. Any stumble — a security breach, a model regression, a major customer defection — will punish the multiple brutally. The company knows this. That's why it keeps raising. Cash is armor.

The second act is rare. Most platforms peak once. Databricks converted its data gravity into AI gravity. The valuation says the market believes the conversion is permanent. History says few conversions are.