Interview, Fireside Chat
Saam Motamedi: Why Series B Won’t Make Money & Why $1M ARR is a BS Milestone for Series A | E1177
- Sam Altman (partner at Greylock) characterizes the current AI investment environment as "even crazier" than the 2021 peak, noting that seed rounds for pre-revenue companies are frequently priced at $100 million+ post-money.
- The Series B market is described as "even frothier" than 2021, driven by an influx of dry powder from growth platforms (e.g., Sequoia, Tiger) that are not constrained by public market multiples.
- Valuation multiples for AI seed companies currently range from 100x to 200x revenue, creating a massive dislocation compared to public SaaS leaders trading at 15x to 20x revenue.
- Market distortion is attributed to corporate strategic investors (Amazon, Google, cloud providers) prioritizing partnership and cloud credits over financial returns, thereby removing price sensitivity.
- Greylock's investment thesis prioritizes backing "iconic founders" in large markets over strict valuation adherence, accepting that paying a premium for top-tier teams is rational if the outcome is a $100 million+ ARR company.
- Seed stage pricing for high-quality teams ranges from $20 million to $40 million post-money, with some leading rounds reaching $26 million (e.g., Upwind) due to founder track records (e.g., NetApp exit).
- Series A pricing for companies with early product-market fit ranges from $80 million to $150–$200 million post-money.
- Greylock rejects the notion that signaling is a primary threat, arguing that brand strength facilitates subsequent fundraising and that investors prioritize company merit over previous investor participation.
- The firm maintains high reserve ratios to play both offense and defense, having previously led inside rounds for companies like Wiz and Abnormal Security when they hit growth "air pockets."
- Sam Altman disagrees with the "end of SaaS" narrative, predicting a new wave of SaaS giants will emerge because the data model, delivery model, and interface of existing systems (like Salesforce) are being fundamentally disrupted by AI agents.
- AI is expected to shift pricing models from pure per-seat to "seat plus work" or "work-only," enabling companies to charge more by directly replacing specific labor tasks (e.g., replacing a BDR role).
- Greylock believes the foundational model layer is less attractive for early-stage VCs unless the model is tightly integrated into a specific application (e.g., personal agents, code generation) where differentiation is required.
- The firm has moved away from the strategy of "playing the game on the field" at expensive growth stages, arguing that it has historically resulted in poor returns unless one has independent conviction in long-term growth durability.
- Mistakes cited include passing on Glean (due to skepticism of the enterprise search market) and Codium (due to pitching the wrong product initially), leading to a revised strategy of backing iconic founders even in seemingly narrow markets.
- 11 Labs is identified as a prime example of a focused model that successfully moved up the stack to the application layer, creating defensibility that large general-purpose models like OpenAI may struggle to replicate.
- Young investors are advised to focus on sourcing and building unique lanes of expertise rather than treating a VC role as a stepping stone to personal brand building or a future fund.
- Greylock's approach to incubation is selective; they attribute success in past incubations (e.g., Palo Alto Networks, Workday, Abnormal Security) to having exceptional founders, large markets, and deep collaboration on recruiting and customer development.
- The firm recently changed its view on the model layer, now believing that startups should start with focused models and rapidly move to the application layer rather than waiting for general-purpose models to mature.
- Sam Altman advises founders to check if a VC adds unique value by asking what would be different about the company without the VC's involvement, noting that referencing checks with VCs is not a reliable predictor of partnership success.
- Greylock typically makes 1-2 investments per partner annually, often starting with small checks ($5M-$6.5M) after extended due diligence periods, emphasizing that time and founder alignment are the primary constraints rather than capital.
- The firm believes that while many venture investors are unhelpful, a small list of 5-10 high-impact partners (e.g., Pat Grady, Eric Schmidt, Ravi Rajwani) is sufficient for board composition, and they prefer to let founders choose new investors if the incremental value is low.
- Most successful SaaS companies historically emerge from disruptions where the data model, delivery model, or interface changes; AI is currently driving all three shifts simultaneously.
- Greylock's portfolio includes companies like Figma, Discord, Wiz, and Abnormal Security, with a focus on backing teams that can execute in massive, secular markets regardless of early traction.