How Databricks Won Enterprise Deals: A Co-Founder's Playbook for Startups
Arsalan Tavakoli-Shiraji, co-founder of Databricks, reveals that technology alone doesn't close enterprise contracts—strategy, partnerships, and understanding customer concerns do.
Startups pursuing rapid growth increasingly depend on landing enterprise customers with substantial budgets. Yet securing such deals remains notoriously difficult. A widespread misconception among younger companies is that superior technology will essentially market and sell itself, according to Arsalan Tavakoli-Shiraji, co-founder and senior vice president of field engineering at Databricks, the data storage and analytics platform.
While product quality matters, it is insufficient on its own, Tavakoli-Shiraji cautioned. Databricks, now valued at $190 billion with more than 70% of Fortune 500 companies as customers, was once a fledgling operation. Tavakoli-Shiraji played a key role in securing the company's initial major contracts, and he shared lessons from that experience at TechCrunch Disrupt last week.
Everybody when they're selling to enterprises at startups … we think it's all about the tech and having the best tech. But people and processes matter the most.
Arsalan Tavakoli-Shiraji
Tavakoli-Shiraji recounted how he once secured a meeting with a prospect only to learn the individual had already committed to a competing solution. "He has no interest in listening to me because he bet his career on that competitor," Tavakoli-Shiraji said. Early deals also collapsed when Databricks lacked readiness to address legal, contractual, and security requirements. These experiences shaped his recommendations for startups pursuing enterprise customers.
Leverage Partnerships With Established Technology Companies
Databricks attempted unsuccessfully to arrange a meeting with UBS, the major Swiss financial institution. The situation changed after the company formed "a partnership with Microsoft," positioning itself as an Azure cloud partner. This arrangement meant sales agreements would operate under Microsoft's contractual framework.
More significantly, Microsoft's sales organization actively scheduled meetings and championed the startup. "All of a sudden, that meeting that wasn't hard to get," Tavakoli-Shiraji recalled. "It was a huge boon to us that we were able to partner with somebody who had that access and distribution." With Microsoft's team supporting the effort, "we started getting pulled everywhere." While this expansion created new challenges—such as customers requesting features in markets Databricks had not yet entered—these represented favorable problems.
Databricks now occupies the position of the larger partner, collaborating with emerging companies to distribute their offerings under Databricks' own terms and security framework. Partners include Hightouch, an AI-driven marketing and customer-data platform in which Databricks has invested; Addepar, which supplies data tools for investment professionals; and Tecton, a machine learning company that Databricks subsequently acquired.
Address the Four Core Enterprise Concerns
Tavakoli-Shiraji categorizes the primary objections and concerns raised by enterprises into four categories he terms the four Cs: context, choice, cost, and control. Context refers to AI systems that can be trained "like bringing on a new employee," absorbing a company's terminology, workflows, priorities, and other organizational specifics.
Choice represents the ability to switch between different AI models rather than remaining dependent on a single vendor. Cost pertains to managing expenses associated with running the AI product. Control encompasses the mechanisms that regulate AI behavior and safeguard confidential data.
Target Modernization of Aging, Expensive Systems
Tavakoli-Shiraji uses a humorous scale when asking chief information officers about their AI maturity. A rating of one indicates employees merely have access to ChatGPT, while five represents a company that has integrated new AI capabilities into its core business operations.
"And everybody loves picking, 'I'm three leaning towards four.' But 99% of the ones I talk to are in bucket one and two," he noted. Most enterprises have not yet determined how to deploy AI agents to generate meaningful value beyond enabling engineering teams to write code. "It's going to take many, many years before people can actually properly deploy AI in a very meaningful way in the enterprise."
A significant opportunity for startups lies in modernizing the expensive legacy infrastructure that enterprises believe they cannot replace. "If you go to any bank today, they still have [an IBM] AS/400 and mainframes," he explained. "Nobody wanted to touch it because nobody knew what that code did."
Contemporary AI agents "can go in and understand and actually decompose what that code does." This capability unlocks possibilities ranging from creating modern interfaces for these obsolete systems to replacing them entirely. Tavakoli-Shiraji also discussed his outlook on enterprise AI adoption trends and shared an anecdote about posing as a lawyer to secure a major enterprise contract during Databricks' early stages.
Read more
Deep Tech Builders Move From Lab to Factory Floor, Controlling Bottlenecks
Startups across nuclear energy, defense, space and AI infrastructure are shifting focus from proving physics to manufacturing at scale. The winners will be those who control critical bottlenecks—whether power, factory capacity or domain expertise.
Anthropic Files for IPO, Targeting Nasdaq Listing Under Ticker ANT
The AI developer has submitted S-1 paperwork to the SEC, positioning itself for what could become the largest public market debut by an artificial intelligence company.
Finding Real Product-Market Fit in the Age of AI: What VCs and Founders Actually Look For
Venture capitalists and startup founders are sounding alarms about inflated metrics and fake stickiness as AI startups race to prove they've achieved product-market fit. The bar for distinguishing real traction from hype has never been higher.