Anthropic, the company hailed as the fastest-growing software entity of all time, has captivated the tech world with its meteoric ascent. This remarkable trajectory was a central theme when Deedy Das, a Partner at Menlo Ventures, joined the Latent Space podcast. He spoke with hosts Alessio Fanelli and Swyx about his journey from operator at Glean to venture capitalist, the explosive rise of Anthropic, and the fundamental shifts reshaping enterprise software and coding. The discussion offered a rare glimpse into the strategic underpinnings of AI success, emphasizing that enduring value is often forged in the crucible of persistent, unglamorous effort rather than fleeting hype.
Menlo Ventures, having invested in Anthropic when it had no revenue, recognized the immense potential in its foundational generative AI layer. This audacious bet on a research-driven company, now projected to reach $7 billion in revenue by 2028, underscores a venture thesis that prioritizes deep technological innovation and the caliber of talent driving it. Yet, the path to such valuations is rarely straightforward, demanding more than just cutting-edge models.
The genesis of elite talent, crucial for these ventures, often springs from unexpected places. Das highlighted the intensely competitive Indian education system, which, despite its rigorous filtering, serves as a powerful engine for social mobility. He noted that "1,000,000 people take the core engineering exam and the top 10,000 get in and the top 200 get into computer science. That's how hard it is." This arduous journey cultivates a resilient mindset, exemplified by individuals like Rahul Patil, Anthropic's CTO, who climbed the ranks without attending a top-tier Indian university. His success is a testament to the fact that "even though you didn't have the opportunities early... If you work hard enough for a long time on things you care about, anything can happen."
Das’s prior experience at Glean, an enterprise search company, offered a nuanced perspective on building enduring value in the enterprise AI space. He candidly recalled the pre-AI era, stating, "From 2019, I remember going to parties in the Bay Area and I would say enterprise search and it's shutting down the conversation right there." Back then, enterprise search was seen as a "boring, unsexy company."
The advent of large language models, particularly the "ChatGPT moment" of late 2022, dramatically shifted this perception, making enterprise search an "interesting" topic. However, Das was quick to clarify that Glean's subsequent $7 billion valuation was not merely a stroke of AI-driven luck. Instead, he asserted, "I really think we did all the hard work to build search and AI happened to accelerate our go-to-market motion at the right time." Glean’s competitive moat was built on years of foundational work, top-down enterprise sales, a difficult rip-and-replace barrier for competitors, and an expansive total addressable market encompassing every knowledge worker. These seemingly unglamorous efforts, solving critical infrastructure problems without shortcuts, proved to be the bedrock that allowed Glean to capitalize on the AI wave.
Related Reading
- Claude Redefines Private Equity Deal Flow with AI-Powered Speed and Diligence
- Anthropic Reveals AI-Led Hack, Reshaping Cybersecurity Landscape
- Top investors gauge AI opportunity: Here's what to know
The entry of major model labs like Anthropic into the enterprise search arena presents a new set of challenges, particularly concerning API access and data ownership. Das noted Anthropic's recent launch of an enterprise search product, but highlighted the inherent difficulties these large model companies face when competing in deeply integrated enterprise solutions. The core issue, he argued, is that "you joined a big AI lab to work on models, not to build Google Drive connectors." Top AI talent is naturally drawn to frontier research, not the painstaking "last mile" integrations required for robust enterprise deployments. Furthermore, unlike consumer search, enterprise search often lacks the high volume of user feedback necessary to rapidly improve ranking algorithms, and the freshness of data is paramount.
Das also critiqued the trend of data providers imposing rate limits on API access, arguing it defies sound business logic. He explained that a third-party tool like Glean, by showing relevant Slack results only to authorized users, actually enhances the value proposition for the data provider, potentially leading to increased seat sales rather than cannibalization. He emphasized that customers own their data and expect the flexibility to utilize it across platforms. In his view, the true moat for companies like Glean lies in their willingness and capability to tackle these complex, multi-faceted problems that larger, model-focused entities are less incentivized to pursue.
