Oct 15, 2025, 10:41 AM ETFireside Chat, Interview, Conference Presentation

Ben Horowitz and Ali Ghodsi: How to Run a $100 Billion Business

Databricks faced a critical inflection point in 2016 due to the "open source paradox," where widespread adoption of Apache Spark threatened the company's ability to monetize, as cloud vendors (including AWS) offered the software for free.
Ali Ghodsi was recruited as CEO in 2016 after Janos Szita (co-founder) and Ben Horowitz recognized the need for a leader with commercial experience to execute an aggressive internal pivot away from a pure open-source ethos.
The core strategic challenge was shifting from a product-market fit in open-source downloads to a B2B enterprise sales model to secure proprietary revenue.
Ali Ghodsi hired Ron Gabrisco, a non-PhD sales veteran from the company Axway, as the first sales executive, a decision Ghodsi notes was "uncomfortable" for the founding team but "transformational" for the company's growth trajectory.
The leadership transition was precipitated by a near-fatal funding freeze during the Series C round in 2015, where existing investors stopped responding, leaving the company reliant on its founders (A16Z and NEA) to co-lead the Series D.

Ben Horowitz rates Ali Ghodsi's CEO capability as superior to his own in most dimensions, citing his unique ability to combine deep technical product knowledge with rapid commercial learning.
Ghodsi's key leadership trait is "non-hesitation"; he trusts his instinct on high-risk strategic moves, such as the development of a data warehouse, even when the idea seems "quixotic" or lacks immediate market precedent.
Ghodsi's approach to professional development involves admitting ignorance, actively studying industry best practices (e.g., reading books, meeting top product managers for 30-minute calls), and synthesizing conflicting advice to build a custom playbook.
The company emphasizes "radical candor" in feedback delivery, favoring high-frequency, low-context daily corrections over annual reviews to prevent employees from being blindsided by performance issues.
Ben Horowitz advocates for a "high-intensity culture" but warns against burnout, noting that leadership must set the tone by working extreme hours while simultaneously monitoring team well-being scores to ensure sustainability.
A critical failure mode for engineering-led startups is hiring successors based on technical similarity rather than role-specific competency (e.g., hiring an engineer to lead sales); Databricks corrected this by hiring non-engineers for non-engineering roles.

A pivotal 2017 partnership with Microsoft was secured after a multi-year effort that involved overcoming internal resistance, with the deal ultimately requiring Ali Ghodsi to physically fly to Redmond and engage directly with blocking engineers to resolve last-minute vetoes.
The Microsoft deal was structured around a massive pre-commit forecast that created "skin in the game" for Microsoft executives, ensuring they would lose their jobs if the partnership failed, thereby securing long-term execution.
The deal succeeded because of a perfect strategic alignment: Microsoft needed a product to compete with AWS and fill a gap in its portfolio, while Databricks needed Microsoft's 60,000-seller distribution channel.
Ali Ghodsi and Ben Horowitz emphasize that successful enterprise deals require a reciprocal "give and get" that remains beneficial to both parties post-closing, noting that most partnerships fail because the smaller vendor has no unique value to offer the giant.
The partnership required overcoming 10+ internal rejections at Microsoft, driven by a "nerd bird" tactic where the Databricks team spent excessive time on the ground at Microsoft locations to influence the organization from within.

Databricks' acquisition philosophy explicitly rejects "buying revenue," prioritizing instead the integration of founding teams and product culture to avoid the "death spiral" where acquired companies lose their CEOs and key talent immediately post-deal.
The acquisition vetting process is strictly sequenced: 1) Assess founder chemistry and cultural fit, 2) Evaluate product integration and engineering compatibility, and 3) Finally, review financials and revenue multiples.
Ali Ghodsi vetoed a potential acquisition of a company that made financial sense on paper but possessed a talent base that was "mediocre," fearing the acquisition would dilute Databricks' engineering culture and employee quality.
The company targets acquiring individuals who have "boomerang" experience—those who have scaled at large tech firms (e.g., Google, Amazon) and subsequently failed at their own startups, as these candidates possess the grit and perspective to appreciate Databricks' processes.
Acquisitions are evaluated on their ability to maintain a unified customer experience; Databricks avoids deals that would introduce fragmented architectures requiring customers to learn new access control models or support systems.

During the Series D pitch, A16Z proposed a "FangDB" narrative (adding "Databricks" to FANG), prompting the team to calculate that their market cap per employee allowed them to pay engineering talent at the P95 percentile, effectively outcompeting Google and Meta on compensation.
Ali Ghodsi rejects the Silicon Valley narrative of "$100 million offers" as a manipulation tactic by CEOs to poach talent, advising that the "FOMO" is largely exaggerated and that founders have decades to build their own companies later.
The company retains top talent by emphasizing mentorship and the potential for impact, with the CEO personally mentoring early-career employees who aspire to become CEOs in the future.
Retention strategy involves "calming" employees during market hype cycles (e.g., the AI boom) by reinforcing the long-term career value of joining a stable, high-growth organization versus the extreme risk of early-stage entrepreneurship.
Ben Horowitz and Ali Ghodsi agree that the decision to keep the company private rather than sell at a 6x valuation offer (which was larger than the company's then-current valuation) was driven by the rarity of having both a massive market opportunity and a founder capable of executing it.

The success of Databricks is attributed significantly to timing; the company's trajectory was perfectly aligned with the maturation of cloud infrastructure and the rise of AI/machine learning, whereas a one-year shift in founding (2013 vs. 2014) could have led to failure due to lack of market readiness.
The founding team remains uniquely cohesive, with seven co-founders continuing to contribute to the company's leadership and strategy years after inception, a rarity in the industry.
The company's initial failure of Product-Led Growth (PLG) in 2015 was a critical learning moment that forced the pivot to enterprise sales, as the "credit card swipe" model proved insufficient for complex enterprise data needs.
The decision to hire a non-PhD sales leader was a calculated risk that broke the "echo chamber" of the all-PhD founding team, introducing necessary external discipline and customer focus.
Ben Horowitz advised against selling the company during a peak acquisition offer by framing the choice as a long-term trade-off: immediate wealth versus the chance to discover the full potential of the business ("take it all the way").