Databricks is raising at a $188 billion valuation because investors are still paying up for the unglamorous layer of AI: the data systems companies need before any model can do useful work.
Databricks has signed a term sheet for a funding round that values the company at $188 billion, Reuters reported on July 16. The Wall Street Journal put the size of the investment at $3 billion and said Coatue Management is leading it. The deal hasn't closed. Databricks expects that to happen later this summer, with new and existing investors joining the round.
You can see why the number grabbed attention. In February, Databricks completed about $5 billion of equity financing at a $134 billion valuation, alongside about $2 billion of additional debt capacity. Five months later, the private-market price has jumped by about $54 billion. That is not small. It puts Databricks in the same private-company conversation as OpenAI and Anthropic, without the same consumer-facing brand.
That is the point.
Databricks doesn't sell a chatbot you play with for ten minutes and forget. It sells the machinery companies use to ingest, clean, govern and query their data, then build AI applications on top of it. That pitch is less glamorous than a frontier model demo, but it is a cleaner enterprise story. If a bank, retailer or drug company wants AI to work on its own records, it first needs the data layer to behave. Databricks sits there.
The company's own numbers explain the investor appetite better than any slogan would. Databricks said in February that it had crossed a $5.4 billion revenue run rate, up more than 65% year over year. The Wall Street Journal reported that CEO Ali Ghodsi said revenue from AI products reached a $1.7 billion run rate by June, up from $1 billion in September. Those are the figures investors are underwriting, not a vague belief that every office worker will soon chat with an agent all day.
The infrastructure trade is widening
Databricks is not the only AI infrastructure bet getting repriced. SambaNova announced on July 8 that it had completed the first close of a $1 billion Series F led by General Atlantic, valuing the company at $11 billion post money. The same announcement said JPMorganChase had selected SambaNova as an inference infrastructure partner and would deploy its SN40 and SN50 systems for secure on-premises AI inference.
That detail matters more than the valuation headline. A bank choosing on-premises inference wants control over where models run, how fast they answer and who touches the data along the way. That's not a popularity contest. You don't have to love every chip startup's valuation to understand why that customer profile is attractive.
Fireworks AI shows the same shift from a different layer of the stack. In May, Bloomberg reported that Fireworks was in talks to raise money at a $15 billion valuation, with Index Ventures set to co-lead. That story has already moved on. Fireworks announced on July 15 that it had raised a $1.505 billion Series D at a $17.5 billion valuation, led by Atreides Management, Index Ventures and TCV. The company also said it had passed $1 billion in annualized revenue run rate and was serving more than 40 trillion tokens a day.
Frankly, that is the cleaner version of the AI boom for investors. Databricks sells the data layer. Fireworks sells model-serving infrastructure. SambaNova is selling systems for inference. None of them needs to own the winning consumer app. They need enterprises to keep building AI into existing workflows, and for now that spending is real enough to support very large private rounds.
The boring bet still has a public-market problem
The risk is not that Databricks has no business. It plainly does. The harder question is whether a $188 billion private valuation leaves enough room for public investors later. Snowflake, its most obvious public-market comparison, gives investors a daily mark on what data-cloud growth is worth when sentiment changes. Databricks does not have to face that mark yet.
Staying private gives Ghodsi more control over timing and more room to invest through public-market swings. It also lets late-stage investors price the company on the belief that AI infrastructure spending will keep compounding before Databricks has to publish full public-company numbers. That can work. It also means the next test is not another term sheet. It is whether the business can grow into a valuation that is now larger than most public software companies.
For customers, the round changes very little tomorrow morning. Databricks will still compete with Snowflake and the data platforms run by Amazon, Microsoft and Google. It will still have to convince enterprises that their messy internal data can be turned into working AI systems, not just expensive experiments. The new money gives it more room to buy and hire, and to bundle more aggressively. It doesn't make the work easier.
That is why this raise is useful to read as a signal, not a coronation. Investors are still willing to pay huge prices for the companies underneath AI, especially the ones with revenue already attached. Databricks has earned that attention more than most. Now it has to prove that a $188 billion private valuation is not just the market rewarding plumbing because the flashier parts of AI have become harder to price.
Also read: Gemini 3.5 Pro's Delay Just Cost Alphabet Nearly $200 Billion in a Day • Visa and Mastercard Just Joined a New Standard for AI Agent Payments • The White House Now Decides Who Gets Early Access to Claude and GPT
