At TechCrunch Disrupt 2026: Databricks’ co-founder on what kills enterprise AI deals

Enterprise organizations are not rejecting AI. They are rejecting operational instability.

That is the shift many founders still misunderstand — and it is becoming one of the defining realities separating enterprise AI companies that scale from the ones that stall after early momentum.

For the last several years, AI startups benefited from a market driven by experimentation. A strong demo, an impressive model, and a powerful vision were often enough to generate enterprise interest, pilot programs, and investor enthusiasm.

But enterprise AI is entering a different phase now, one where enterprises are no longer evaluating whether AI is exciting. They are evaluating whether it is safe to deploy broadly.

At TechCrunch Disrupt 2026, taking place October 13–15 at Moscone West in San Francisco, Arsalan Tavakoli-Shiraji, co-founder and SVP of field engineering at Databricks, will unpack that shift during his AI Stage session, “The Enterprise Isn’t Broken. Your Assumptions About It Are.”

Disrupt will bring together 10,000+ founders, investors, and operators to explore the technologies and operational pressures changing how companies are built and scaled. The three-day event will feature 250+ sessions across six stages, led by tech leaders directing the industry today.

Explore the sessions appearing on the Disrupt AI Stage. Ticket savings of up to $410 end on May 29 at 11:59 p.m. PT. Register here.

The pilot was never the hard part

The enterprise AI market is full of successful pilots that never became real deployments. Not because the technology failed. But because the organization could not absorb the operational consequences of adopting it.

Now the reality founders need to face is that startup AI deals rarely die because the model underperformed. They die because the enterprise lost confidence in what the deployment would require.

That is the gap Tavakoli-Shiraji’s session is designed to explore. Most enterprises are not simply evaluating whether an AI product works. They are evaluating:

• Implementation risk.

• Governance complexity.

• Workflow disruption.

• Infrastructure strain.

• Compliance exposure.

• Organizational trust.

An AI product can perform exceptionally well in a controlled environment and still fail commercially if its deployment creates instability within the business.

That distinction is important to founders because many AI startups are still optimizing for the wrong outcome. They are building for initial excitement rather than long-term operational adoption. And enterprises are becoming far more disciplined about recognizing the difference.

Register for Disrupt to hear how enterprise AI leaders evaluate what actually survives beyond the pilot phase. Lock in your ticket savings of up to $410 when you register by May 29 at 11:59 p.m. PT.

Enterprise AI is becoming an operational trust problem

The AI startups gaining traction inside large organizations increasingly share one thing in common: They reduce uncertainty.

They integrate more cleanly into existing systems. They create less workflow friction. They are easier to govern, easier to explain internally, and easier for organizations to trust over time.

That sounds less exciting than breakthrough demos or model benchmarks. But it is quickly becoming the difference between AI startups that generate attention and those that generate durable revenue.

The market is maturing. Enterprise buyers are asking different questions now:

• What happens after deployment?

• How much operational change is required?

• How does this affect governance?

• Can teams realistically adopt this at scale?

• What happens when the model fails?

Those concerns are no longer secondary. In many organizations, they have become core to the buying decision itself. For AI founders selling into the enterprise, this session breaks down what actually drives adoption after the pilot phase ends. Check out the session details and get your $410 ticket savings to learn what to prioritize to gain traction with enterprise AI deals.

Why Tavakoli-Shiraji sees the market differently

Tavakoli-Shiraji brings an unusually relevant perspective to this conversation because his background spans both enterprise strategy and deeply technical systems architecture.

Before joining Databricks, he was an associate principal at McKinsey & Company, advising enterprises, technology vendors, and public-sector organizations on cloud computing, next-generation IT, and enterprise transformation strategy. He also earned a PhD in computer science from UC Berkeley, focused on networking and distributed systems.

That lens is valuable to startups because enterprise AI success increasingly depends on more than strong engineering alone. Founders now need to understand how technical systems interact with organizational behavior, infrastructure realities, procurement processes, governance concerns, and operational risk.

The startups that succeed in enterprise AI over the next several years may not necessarily be the ones with the most advanced models. They may be the ones that best understand how enterprises actually absorb change.

That is the kind of operational pressure that Tavakoli-Shiraji and other speakers on the AI Stage at Disrupt will explore. Presented by Google Cloud, the stage examines how AI agents and generative AI are reshaping SaaS, enterprise adoption, software economics, security, and operational infrastructure — including Tavakoli-Shiraji’s session on why enterprise AI success increasingly depends on operational trust rather than simply technical performance.

Across the stage, founders will learn how and why the focus is shifting away from AI novelty and toward the real-world challenges of deploying, governing, and scaling AI systems inside real organizations.

Two days left to save on enterprise AI insight

Explore the Disrupt agenda and learn how founders, investors, and enterprise operators are managing the next phase of AI adoption. Register by May 29 at 11:59 p.m. PT to save up to $410 on your passes.