Databricks Raises at $188 Billion on a Simple Bet: Tokenmaxxing Is Dead

Databricks has signed a term sheet for a strategic funding round at a $188 billion valuation, the company announced Wednesday. Existing investor Coatue is leading, with new and existing investors joining, and Databricks expects to close the round later this summer.

To put the number in context: Databricks closed its $1 billion Series K at a valuation just above $100 billion in September 2025, a round co-led by Andreessen Horowitz, Insight Partners, MGX, Thrive Capital, and WCM. At that point the company had crossed a $4 billion revenue run-rate, with its AI products alone passing $1 billion. Less than a year later, investors are pricing the business nearly twice as high.

The valuation will get the headlines. The more interesting part is what Databricks says the money is for.

"Tokenmaxxing" is out

Co-founder and CEO Ali Ghodsi framed the raise around a shift in how enterprises are buying AI.

"Enterprises are moving from tokenmaxxing to valuemaxxing," Ghodsi said in the announcement. "They don't want to burn expensive tokens on the smartest model for every task. They want the best outcome per dollar. That means having the freedom to choose the right AI for the job."

He was blunter in a post on LinkedIn: "A year ago, everyone wanted to use the largest, most expensive model for every single task. Today, organizations realize they can't burn budget like that."

The data backs him up. Menlo Ventures' State of Generative AI in the Enterprise research found that organizations typically run three or more foundation models in their stacks, routing work to different models depending on the use case. Nobody is standardizing on a single frontier model for everything, and the companies footing the inference bills have noticed that most of that spend does not need the priciest model available.

If enterprises are going multi-model, someone has to own the routing, governance, and cost control layer that makes it workable. That is the position Databricks is spending to lock down.

Where the capital goes

Databricks named three product pillars for the new money.

The strategic centerpiece is Unity AI Gateway, its multi-AI governance product for controlling access and cost across models and vendors. If the valuemaxxing thesis holds, the gateway is the tollbooth: a control plane sitting between the enterprise and every model provider it uses.

The second is Genie, the company's AI coworker that turns business data into answers and actions. Ghodsi describes it as "AI coworkers that actually understand your business data," which is a pointed phrase. MIT's NANDA initiative reported last year that roughly 95% of enterprise generative AI pilots delivered no measurable P&L impact, and the researchers pinned the failures on poor integration with company data and workflows rather than on model quality. Genie is Databricks' answer to that specific problem, the gap between capable models and business context.

The third is Lakebase, a serverless Postgres database built for AI agents. Databricks launched Lakebase in 2025 and flagged it as a funding priority in the Series K as well, a sign the company sees agentic workloads becoming a major new consumer of transactional infrastructure.

Beyond the product roadmap, Databricks says the capital is expected to fund future AI acquisitions and deepen its research efforts. The company has form here. Its MosaicML deal ($1.3 billion) and Tabular deal (roughly $1 billion) both looked expensive at the time and both set up the AI push that investors are now paying up for.

Why this round matters

Databricks is making a bet that runs opposite to the vertically integrated model labs. Rather than winning by having the best model, it wants to win by being the neutral layer that lets 20,000-plus customers, including 70% of the Fortune 500 per the announcement, mix and match whatever models they want while keeping their data, governance, and costs in one place.

The MIT numbers explain why that pitch lands. Enterprises are not short on model capability; they are short on integration, context, and control. A platform that unifies data and AI with governance built in is a credible answer to why so many pilots stall before production.

And the round itself says something about where private-market conviction sits in mid-2026. Sentiment on pure model plays has cooled, but a data-plus-AI platform with real revenue, positive free cash flow, and a defensible position in the enterprise stack can still nearly double its valuation in ten months.

The round is expected to close later this summer. Watch where the acquisition dollars go. That will be the clearest signal of how Databricks plans to press its advantage.