The Round Databricks Turned Down
Most funding stories lead with the valuation because the valuation is the easy number to print. The more interesting number in this round is the gap between what Databricks wanted and what it took. TechCrunch reported that the company set out to raise roughly $1 billion, that investor interest in the syndicate reportedly reached as high as $15 billion, and that the final round closed at $5 billion, with Coatue leading and Blackstone, MGX, T. Rowe Price and new investor Sixth Street Growth joining alongside a long list of existing backers including Andreessen Horowitz, Fidelity, GIC, Insight Partners, NEA, Temasek and Thrive Capital.
Turning away three dollars for every one taken is a specific kind of signal. It says the constraint was never appetite, it was allocation, and Databricks chose to keep the cap table smaller rather than take the money simply because it was offered. That is a different posture than a company racing to lock in every dollar it can before a market window closes, and it is worth noticing precisely because most AI-era mega-rounds read the opposite way.
A Valuation That Barely Moved
A term sheet reported in July had already put Databricks near $188 billion. The round that closed on August 13 priced the company at $190 billion, a rise of roughly 1 percent in the space of a month, even as reported investor demand for the round itself reportedly tripled from what the company initially sought. If Databricks had wanted to maximize price, oversubscription of that scale is exactly the leverage a company uses to push valuation higher. It did not use it that way.
Read against the product numbers, the restraint makes more sense. Revenue run-rate above $7 billion, growth above 80 percent, a Lakehouse product past $1.5 billion in run-rate on its own and a newer Lakebase database product already above $100 million: Databricks priced this round on operating metrics that were already public knowledge in its own growth disclosures, not on a valuation war it could have started with a bidding syndicate willing to pay far more.
The Register Every DORA Officer Now Has to Fill In
The EU's Digital Operational Resilience Act requires financial entities to maintain a Register of Information on every critical ICT third-party provider, covering concentration risk, exit strategies and, in practice, an assessment of the provider's own financial soundness. Supervisors moved from dialogue to active review in 2026, and firms can no longer argue the requirements are unclear. That assessment is straightforward for a listed vendor: pull the 10-K or annual report, check the debt schedule, check the audit opinion.
Databricks offers none of that. It discloses growth metrics through its own press releases, on its own timetable, with no independent audit attached that a bank's third-party risk team can cite. A $190 billion private valuation is not proof of financial soundness in the sense DORA means it; it is proof that investors expect future cash flows to justify the price, which is a different claim entirely. As Databricks becomes the layer running agents that, in CEO Ali Ghodsi's own words, are meant to 'work across their business' and 'execute work without blowing through budgets,' the gap between how deeply embedded the vendor is and how little a regulated customer can independently verify about it gets wider, not narrower.
What a DORA Officer Should Actually Do With This
The practical response is not to distrust Databricks, whose growth numbers are real and whose product build-out is substantive. It is to stop treating 'the vendor raised another round at a higher valuation' as reassurance and start treating it as a prompt: does your Register of Information entry for Databricks still reflect current concentration risk now that Genie, Unity Catalog and agentic workloads sit deeper in your data estate than they did a year ago? Contractual covenants requiring periodic financial disclosure, SOC 2 and ISO 27001 attestations, and a documented exit plan are the substitute for public filings that will not arrive while the company stays private.
None of this requires assuming Databricks is fragile. Coatue's own co-founder called it 'the infrastructure the industry builds and scales AI on,' and the revenue numbers back that framing. The point is narrower and more durable: scale and disclosure are not the same thing, and a vendor can become simultaneously more important to your operations and no more transparent about its own finances, which is precisely the condition DORA was written to make institutions stop overlooking.
Read next: Bosch Sold Its Robot to the Startup It Funded | OVHcloud Warns of Price Hikes Up to 87 Percent



