The Twenty Minute VCBenchmark's GP, Everett Randle on Why Mega Funds Will Not Produce Good Returns
At a glance
WHAT IT’S REALLY ABOUT
Benchmark’s Everett Randle Dismantles Mega-Funds, AI Metrics, and Moats
- Everett Randle, newly minted GP at Benchmark and former investor at Founders Fund and Kleiner Perkins, explains how fund structure, incentives, and culture shape venture outcomes far more than most people admit. He argues that mega-funds have effectively chosen capital velocity and billion‑dollar checks as their “main product,” which structurally caps their money‑on‑money returns even if they make huge absolute dollars. Randle also lays out why AI companies need a new analytical framework: margins matter less than absolute gross profit per customer, labor-budget displacement, and differentiated product usage relative to the labs. Throughout, he defends Benchmark’s small-fund, high-conviction, high-involvement model, discusses misses like OpenAI at $32B, and reflects on governance, firing founders, and what will really threaten Benchmark over the next decade.
IDEAS WORTH REMEMBERING
5 ideasAI applications shouldn’t be judged by classic SaaS margin benchmarks.
Randle argues that AI app companies may have lower gross margins due to inference costs, but can generate far higher absolute gross profit per customer by tapping labor budgets and delivering 24/7 capabilities—so investors should focus on gross profit dollars and terminal margin structure, not just today’s percentage margins.
Fund size and org design inexorably dictate investment strategy and behavior.
Invoking a ‘Conway’s Law for VC,’ he says mega-funds with $7–10B vehicles and large teams must prioritize capital velocity and billion‑dollar checks, while small firms like Benchmark are structurally optimized for high-conviction, high-ROI, concentrated positions that can still return 20–60x.
Mega-funds will make huge absolute profits but struggle to deliver classic VC multiples.
Randle believes firms writing billion‑dollar checks into companies like OpenAI and Databricks will earn a lot of money, yet can’t credibly promise 5x+ net across massive, pari‑passu fund complexes—whereas Benchmark can still target and demonstrate those higher money-on-money returns.
True AI moats remain largely technological and talent-based, not just about distribution.
Contrary to the narrative that moats have shifted to distribution and data access, he emphasizes how few teams can actually build world-class AI products, integrate LLMs into workflows, and out-innovate the labs—making product quality and elite talent the core defensible advantage.
Governance requires real spine: boards can’t trade fiduciary duty for founder NPS.
Randle defends occasionally replacing founders when laws or ethics are breached and criticizes both the ‘never fire founders’ posture as a convenient abdication of responsibility and the current tendency of some boards to avoid hard decisions to stay popular with CEOs.
WORDS WORTH SAVING
5 quotesWe need a new taxonomy for AI companies.
— Everett Randle
If your AI app has high gross margins right now, it probably means no one is actually using your AI features.
— Everett Randle
You ship your fund size. For a $7 billion fund, billion‑dollar checks are your main product.
— Everett Randle
I don’t think they can go to LPs and say, ‘We’re going to get you 5x net on that,’ and say it with a straight face.
— Everett Randle
The only thing that’s ever made me less of a capitalist is realizing what happens when capitalism focuses on our minds instead of on physical goods.
— Everett Randle
High quality AI-generated summary created from speaker-labeled transcript.
Get more out of YouTube videos.
High quality summaries for YouTube videos. Accurate transcripts to search & find moments. Powered by ChatGPT & Claude AI.
Add to Chrome