Databricks closed a $4+ billion Series L yesterday, valuing the company at $134 billion—up from $100 billion just three months ago. That's the kind of valuation jump that used to require going public. Except Databricks has no intention of IPOing anytime soon, and they're proving you don't need public markets when private investors will keep writing massive checks.
The Numbers Are Absurd
$4.8 billion revenue run-rate, growing 55% year-over-year. Positive free cash flow over the last 12 months. Over $1 billion in AI product revenue. More than 700 customers spending over $1 million annually. Net retention above 140%, meaning existing customers are expanding their spend dramatically.
These aren't startup metrics. These are mature company numbers that traditionally signal IPO readiness. Yet Databricks just completed its third major funding round in less than a year—$62 billion valuation in early 2025, $100 billion in September, now $134 billion in December.
The funding round was led by Insight Partners, Fidelity, and J.P. Morgan Asset Management, with participation from basically everyone: Andreessen Horowitz, BlackRock, Blackstone, Coatue, GIC, NEA, Temasek, and more. When asset managers like BlackRock and Fidelity are investing at these valuations, it signals that private market valuations have completely decoupled from traditional IPO logic.
Why Stay Private?
CEO Ali Ghodsi told Bloomberg the company wants to keep investing in its core products and AI software without the quarterly earnings pressure that comes with being public. That's the standard line, but there's more to it.
Going public means: scrutiny on margins, questions about growth sustainability, pressure to deliver predictable quarterly results, and restrictions on long-term bets that might hurt near-term profitability. Staying private means: freedom to invest aggressively in R&D, ability to make acquisitions without shareholder approval, and no obligation to disclose competitive information to rivals.
Databricks acquired Neon (a Postgres database startup) for $1 billion earlier this year to build Lakebase, its AI agent database. That kind of deal happens faster in private markets. Public company M&A requires board approval, regulatory disclosure, and shareholder patience. Private companies can move more quickly.
There's also the valuation arbitrage. Private investors are willing to pay $134 billion for a company with $4.8 billion revenue—a 28x multiple. That's high even by SaaS standards. If Databricks went public today, market skepticism might compress that multiple significantly, especially given public market volatility.
What They're Actually Building
The Series L funding is earmarked for three strategic products: Lakebase (a serverless database for AI agents), Agent Bricks (a platform for building and deploying multi-agent systems), and Databricks Apps (application layer for data-intelligent applications).
Lakebase launched six months ago and is already growing revenue at twice the pace of Databricks' data warehousing product. That's remarkable traction for a new product category. The pitch is that AI agents need transactional databases that can handle rapid reads/writes while maintaining data consistency. Traditional data warehouses aren't built for that workload.
Agent Bricks is Databricks' bet that enterprises will deploy hundreds or thousands of specialized AI agents, each handling specific workflows. The platform provides infrastructure to manage, monitor, and orchestrate these agents at scale. It's infrastructure for an agentic future that hasn't quite arrived yet, but Databricks is building for it now.
The broader strategy is becoming the foundational data layer for enterprise AI. Every company needs: data infrastructure (Databricks' core business), data warehousing (over $1 billion revenue run-rate), AI-ready databases (Lakebase), and agent platforms (Agent Bricks). Databricks wants to own the entire stack.
The Competitive Landscape
Snowflake is the obvious comparison—both are data platforms targeting enterprise customers. But Snowflake went public in 2020 and has faced market pressure ever since. The stock has been volatile, forcing the company to manage quarterly expectations rather than just executing long-term strategy.
Databricks is competing with cloud hyperscalers (AWS, Azure, Google Cloud) that have massive resources and integrated offerings. Staying private lets Databricks maintain partnerships with all three clouds simultaneously without picking sides or facing investor pressure to consolidate.
There's also OpenAI and Anthropic, which Databricks partners with to sell AI services to enterprise customers. Those relationships involve commercial terms that might be complicated if Databricks were public and required to disclose them.
The AI Infrastructure Thesis
Databricks' valuation is a bet that the "AI infrastructure layer" will be worth hundreds of billions. Data plumbing isn't sexy, but it's lucrative when every company is building AI products that need clean, accessible data.
The company's pitch is straightforward: you can't build good AI without good data. Most enterprise data is messy, siloed, and hard to access. Databricks provides the platform to clean, organize, and make that data AI-ready. As companies shift from experimenting with AI to deploying it in production, they need platforms like Databricks.
That thesis is playing out in revenue growth. AI products alone crossed $1 billion run-rate. Lakebase is growing faster than any previous Databricks product. Over 20,000 enterprises including AT&T, Block, Mastercard, and Rivian are building AI apps on the platform.
When Will They IPO?
Databricks filed confidentially for an IPO back in 2024, but that was more strategic optionality than actual intent. The company keeps raising massive private rounds, suggesting they're in no hurry.
Ghodsi hinted they want to avoid a situation where public markets force them to optimize for 30% EBITDA margins instead of investing in future products. That suggests the IPO happens when: (1) growth slows enough that public market multiples make sense, (2) private market funding becomes harder to access, or (3) employees need liquidity at scale.
None of those conditions exist yet. Growth is still strong. Investors are eager. The current funding round includes employee secondary sales, providing some liquidity without the complexity of an IPO.
What This Means for the Market
Databricks proves that IPOs aren't necessary for scaling to $100+ billion valuations anymore. The traditional path was: start, scale, raise venture capital, go public, mature as a public company. Now it's: start, scale, raise endless private capital, stay private indefinitely.
This only works in a market with abundant capital and investors willing to accept illiquidity. Both conditions are true right now for AI infrastructure companies. Whether they stay true is an open question.
For employees, staying private means stock options without the certainty of public market liquidity. For customers, it means Databricks has long-term runway without quarterly pressures. For competitors, it means facing a well-funded rival that doesn't have to report its strategy publicly.
My Take
I'm genuinely impressed by the execution. Growing 55% at nearly $5 billion revenue while generating positive cash flow is exceptional. The AI product traction validates that they're building things customers actually want.
But the valuation feels stretched. A 28x revenue multiple assumes: (1) continued hypergrowth, (2) successful expansion into new product categories, and (3) sustained enterprise AI spending. If any of those assumptions break, the valuation compresses fast.
The decision to stay private makes sense given current market conditions. Why go public when private investors will fund you indefinitely at premium valuations? But eventually, either growth slows, private funding dries up, or employees demand liquidity. At that point, the IPO becomes inevitable.
My guess: Databricks goes public in 2027-2028, probably at a valuation between $150-200 billion depending on market conditions. They'll keep raising private rounds until the window opens on favorable terms. And given their revenue trajectory, they'll probably pull it off successfully.
Until then, they're running one of the most interesting "we don't need public markets" experiments in tech history.