Databricks co-founder: partner sales channels, not tech alone, win enterprise deals

Databricks co-founder Arsalan Tavakoli-Shiraji argues that startups tend to lose large enterprise contracts for a reason that has little to do with engineering: they expect superior technology to sell itself. Speaking at TechCrunch Disrupt, in an interview published by TechCrunch on October 26, 2026, he said the lesson from Databricks' early years is that people and processes decide enterprise deals, and that a big partner's sales force is the fastest way through the door.

The technology will not close the deal

According to TechCrunch, Databricks is now valued at $190 billion and counts more than 70% of the Fortune 500 among its customers, but Tavakoli-Shiraji was there when it was a small company chasing its first large contracts as senior vice president of field engineering. He described two early mistakes. In one, Databricks finally secured a meeting with a prospect, only to learn the person had signed a mega-deal with Databricks' main competitor and had staked his career on it, so he had no interest in listening. In other cases, deals fizzled because the startup was unprepared for the legal, contractual and security questions enterprise buyers raise as a matter of course.

Selling under Microsoft's umbrella

The move that changed Databricks' trajectory was partnering with Microsoft and joining the Azure cloud program. Sales agreements were then covered under Microsoft's terms, and, crucially, Microsoft's salespeople booked the meetings and advocated for the startup. Tavakoli-Shiraji recalled that a meeting Databricks had repeatedly failed to get with UBS, the large Swiss bank, suddenly became easy to arrange. Demand followed: the company was pulled into deals it had not chased, including requests tied to countries where it had not yet set up operations, which he characterized as the good kind of problem.

A secondary write-up on dev.to summarized the interview as advice to piggyback on a large vendor's sales force and claimed no specific partners were named; TechCrunch's own report names Microsoft explicitly, which is the more precise account for anyone tracing the story back.

Databricks is now the giant

The roles have since reversed. Databricks now sells younger companies' products under its own terms and cybersecurity setup. TechCrunch cites Hightouch, which builds AI-powered marketing and customer-data tools and has also received Databricks investment; Addepar, which offers data tools for investment professionals; and Tecton, a machine learning startup that Databricks later acquired.

The four Cs of selling AI

For startups selling AI specifically, Tavakoli-Shiraji groups enterprise concerns into four categories:

• Context: the AI must be trainable like a new employee, absorbing the company's terminology, processes and priorities.
• Choice: buyers want the freedom to switch models rather than be tied to a single provider.
• Cost: they want control over how much the system costs to run.
• Control: they want safeguards on behavior and protection for sensitive information.

Legacy systems as the opening

He also described a maturity question he puts to CIOs, asking them to rate their AI adoption from one to five. Most answer that they are a three heading toward four, he said, but in his view nearly all sit in the bottom two buckets, and meaningful enterprise AI deployment will take many more years. The near-term opportunity he sees is legacy modernization: banks still run IBM AS/400 midrange machines and mainframes that no one dared touch because no one understood what the code did. AI agents can now decompose that code, he argues, opening the door to products that add modern interfaces to those systems or replace them entirely.

Why it matters

For startups, the core lesson is that distribution is as much a product decision as architecture. A larger partner's sales channel compresses sales cycles, absorbs the legal and security scrutiny that sinks young vendors, and lowers the perceived risk that enterprise buyers actually price into their decisions. The Databricks story also shows how the ecosystem repeats itself: a startup that once rode Microsoft's channel is now the channel, choosing which younger companies get access to Fortune 500 buyers. And for incumbents, the legacy-modernization angle suggests where near-term enterprise AI value may concentrate, not in headline-grabbing agents but in finally decoding systems that have resisted change for decades. TechCrunch notes that the full interview also covers his predictions for enterprise AI adoption and an anecdote about posing as a lawyer to land an early contract.