Databricks Hits $188B Valuation — What Developers Need to Know

Databricks raised $3 billion at a $188 billion valuation, a 40% increase from five months ago, driven by its pivot from data analytics to an enterprise AI control plane. CEO Ali Ghodsi revealed that Databricks' 3,000 engineers now default to Chinese open-source model GLM 5.2 over Anthropic's flagship, signaling a shift in enterprise AI preferences. The company surpassed Snowflake in annualized recurring revenue, reaching $5.4 billion with 65% growth.

Databricks just raised $3 billion at a $188 billion valuation — a 40% jump from its $134 billion mark just five months ago. That number will generate plenty of coverage about AI’s private market frenzy. However, the more interesting detail in this announcement is buried a few paragraphs down: Databricks CEO Ali Ghodsi quietly revealed that his company’s 3,000 engineers are now defaulting to a Chinese open-source model instead of Anthropic’s flagship. That is the real story inside the funding announcement https://techcrunch.com/2026/07/17/databricks-hits-188b-valuation-extending-its-run-as-ais-favorite-second-act/ . From $62B to $188B in 18 Months To understand what is happening here, look at the trajectory. Databricks was valued at $62 billion in December 2024. It raised at $100 billion in September 2025, then $134 billion in February 2026 with a $5 billion round. Now it has closed a $3 billion round led by Coatue at $188 billion. Three-x in eighteen months, with each round arriving faster than the last. The justification is real. Databricks crossed $5.4 billion in annualized recurring revenue as of February, growing at 65% year over year. Snowflake, its closest rival, sits at $4.68 billion in annual revenue growing at 29%. Databricks has now surpassed Snowflake in ARR — a reversal that would have seemed unlikely two years ago. Moreover, the 35x ARR multiple the market is assigning reflects a clear bet: Databricks has successfully pivoted from data analytics platform to enterprise AI control plane. Ghodsi put it plainly in the official announcement https://www.databricks.com/company/newsroom/press-releases/databricks-raising-strategic-round-funding-188-billion-valuation : “Enterprises are moving from tokenmaxxing to valuemaxxing — they want the best outcome per dollar.” That phrase is doing a lot of work. It is both a product thesis and a pricing argument, and it points directly to the three products Databricks is now building its future around. Three Products Developers Should Know The funding is earmarked for three specific bets. Each one targets a gap that enterprise development teams actually face right now. Lakebase is Databricks’ answer to the question every AI agent team faces: where does the agent store its working memory? Lakebase is a serverless Postgres database integrated directly with the lakehouse. It scales to zero when idle, supports instant branching and zero-copy clones for safe testing, and ships with pgvector built in for vector search. It is now generally available on Azure. If your team runs a separate Postgres instance alongside Databricks, Lakebase eliminates that dependency. See the full feature list on the Lakebase product page https://www.databricks.com/product/lakebase . Unity AI Gateway is the governance layer for teams managing multiple AI models and agents simultaneously. It handles spend caps, smart routing, policy enforcement, and unified telemetry across all LLM calls and MCP tool invocations. Think of it as the control plane for a team now running five different AI vendors with no visibility into what any of them cost. That describes most enterprise data teams right now. Omnigent is the wildcard. It is an open-source meta-harness that sits above existing agents — Claude Code, Codex, custom agents — and orchestrates them as a unified system. The managed version on Databricks entered beta this month. The pitch is that teams do not need to rewrite their agent infrastructure to get cross-agent coordination. Instead, they layer Omnigent on top and get stateful governance and shared memory without dismantling what they already built. The Databricks $188B Valuation’s Buried Bombshell: Open Models Beat Anthropic Here is where it gets genuinely interesting. Databricks published internal benchmarks from its own multi-million-line codebase — deliberately avoiding public datasets where models can game results from training data — and found that GLM 5.2, a Chinese open-source model from Zhipu AI, matched Claude Opus 4.8 in coding quality at $1.28 per task versus $1.94. That is a 34% cost reduction with no measurable quality penalty. Databricks has now switched GLM 5.2 to its default coding model https://www.databricks.com/blog/benchmarking-coding-agents-databricks-multi-million-line-codebase . The implications extend beyond Databricks. Anthropic and OpenAI have been the default assumption for serious coding workloads. Furthermore, Databricks ran a rigorous benchmark on its own production code — not a synthetic leaderboard — and found that assumption wrong for its use case. If a company with 3,000 engineers can drop to an open model and cut costs by 34%, the case for paying frontier model prices becomes much harder to make across the board. The Competitive Stakes: Databricks vs Snowflake, Redefined For years, Snowflake and Databricks competed in adjacent markets. That is no longer true. Both now pursue the same enterprise data and AI budget. Snowflake holds approximately 18% of the cloud data warehouse market versus Databricks at roughly 9%, but the growth rates tell a different story. Databricks is growing at twice Snowflake’s pace and has already crossed it in total ARR. The valuation premium reflects that divergence. Databricks trades at roughly 35x ARR while Snowflake trades at around 12x. The market is pricing in a permanent premium for AI-native data platforms, and Databricks is betting its entire product roadmap on that thesis being correct. What This Means for Your Team If you are already on Databricks, this round accelerates your near-term options. Lakebase moving to general availability removes a dependency for teams waiting on it. Unity AI Gateway becoming a first-class product means centralized AI spend visibility is available without third-party tooling. And the GLM 5.2 default signals that Databricks is willing to challenge its own vendor relationships to win on cost — which should prompt every team to revisit its own AI model assumptions. The $188 billion price tag does raise a question Databricks has not yet answered: when does it go public? At this valuation, a Databricks IPO would rank among the largest tech listings in years. The company keeps raising private rounds and running the valuation higher, but ultimately the returns have to close somewhere. Watch that tension play out over the next eighteen months. In the meantime, the products are real, the growth is real, and the open-model benchmark is worth taking seriously regardless of where the stock eventually prices.