Databricks co-founder Arsalan Tavakoli-Shiraji has outlined how startups can secure large enterprise customers, emphasizing that strong technology alone is insufficient to win major corporate contracts. Speaking at TechCrunch Disrupt 2026, he identified strategic partnerships, enterprise purchasing processes and customer concerns about artificial intelligence as critical factors in closing deals.
Tavakoli-Shiraji, who serves as senior vice president of field engineering at Databricks, drew on his experience helping the company secure its earliest enterprise customers. Databricks is now valued at $190 billion and serves more than 70% of Fortune 500 companies.
Strategic Partnerships Can Help Startups Reach Enterprise Buyers
Tavakoli-Shiraji said startups frequently underestimate the importance of relationships, purchasing procedures and internal decision-making when approaching large organizations. He recalled losing potential deals because Databricks was unprepared to address contractual, legal and security requirements.
The company initially struggled to secure a meeting with Swiss banking group UBS before establishing a partnership with Microsoft Azure. Microsoft’s existing customer relationships, sales representatives and contractual arrangements subsequently helped Databricks reach prospective enterprise clients.
The partnership also allowed customers to purchase Databricks services through established Microsoft agreements, reducing procurement barriers. Microsoft and Databricks have since expanded their strategic partnership, extending their collaboration into the 2030s.
Databricks now provides similar distribution opportunities for smaller technology companies, including Hightouch and Addepar. The company also previously partnered with machine learning startup Tecton before acquiring it.
Enterprise AI Buyers Focus on Four Key Requirements
Tavakoli-Shiraji identified four primary considerations for companies purchasing AI products: context, choice, cost and control. Context refers to adapting AI systems to an organization’s terminology, procedures and business priorities, while choice allows companies to switch between AI models without depending on one provider.
Cost involves managing the operational expense of AI applications, particularly as usage increases across departments. Control addresses security safeguards, data protection and mechanisms for governing how AI systems operate within enterprise environments.
Legacy Systems Present Opportunities for AI Startups
Tavakoli-Shiraji also questioned how extensively enterprises have adopted AI beyond individual productivity applications. He estimated that 99% of chief information officers he speaks with are still at the earliest stages of adoption, despite many describing their organizations as more advanced.
He argued that meaningful enterprise-wide adoption of AI agents could take several years, particularly outside software development. Many businesses have yet to integrate AI into operational processes that generate measurable value.
One opportunity he identified involves modernizing expensive legacy systems, including IBM AS/400 infrastructure and mainframes still used by banks. AI agents could analyze older programming code, identify its functions and support the development of modern interfaces or replacement systems.
Tavakoli-Shiraji said this capability could make previously difficult modernization projects more accessible to startups developing enterprise software.
Featured image credits: Wikimedia Commons
For more stories like it, click the +Follow button at the top of this page to follow us.