The startup "excellence gap" is widening

🎉 I loved hosting the annual DECODE conference!

Wonderful catching up with everyone in-person at the conference!

• DECODE is the largest founder community co-hosted across UC Berkeley and Stanford. The DECODE annual conference focuses on helping founders in their earliest stages of starting a startup. I’ve had the honor of serving on the DECODE board since 2016.

• To quote Keith, my fellow DECODE board member, from LinkedIn: “what consistently makes this event different … is that it has maintained an intimate and casual vibe, while providing an opportunity for early-stage entrepreneurs to engage with an astonishing cast of tech, business, academic, and finance visionaries and executives.”

• You can find my top takeaways from the conference here.

3 trends in startups/tech/venture

📈 The startup "excellence gap" is widening

• The latest data shows that the gap between “good” startups (50th percentile) and “great” startups (75th percentile) is widening — as defined by growth rate, net revenue retention (NRR), gross margins, and ARR per full-time employee (FTE).

• What does “good” vs. “great” look like for early stage startups with <$1M annual recurring revenue (ARR)?

  • For year-on-year growth, good = 100%; great = 300%

  • For NRR, good = 100%; great = 116%

  • For gross margin, good = 74%; great = 80%

  • For ARR per FTE, good = $55k; great = $100k

• The biggest gap in metrics between good and great startups are in growth rates and ARR per FTE. Great startups grow efficiently, achieving high growth rates while maintaining high ARR per FTE.

🤖 Human-agent teaming is going mainstream

• The latest research from Stanford and Carnegie Mellon compared humans vs AI agents across speed, cost and quality of work.

  • Speed: AI agents finished tasks 88% faster than humans.

  • Cost: AI agents cost 90-96% less than human workers.

  • Quality: Humans achieved significantly higher success rates across all task types.

  • Human-agent teaming maintained human-level quality while improving efficiency by 69%.

• The takeaway? Delegate programmable, repetitive work to AI agents and instead give humans the tasks that require visual judgment and creativity.

💰 Investors’ appetite for private tech is growing

• The latest data shows that private tech companies valued above $1B (~1,300 companies) now represent roughly $4.7T in aggregate value and make up ~15% of the entire NASDAQ market cap (and closer to ~40% if you exclude the Magnificent 7 of Apple, Microsoft, Amazon, Alphabet (Google), Meta, Nvidia, and Tesla).

• By count, there are ~6x more private unicorns than public companies with a $1B+ market cap.

• The takeaway? Exposure to public equities alone is no longer sufficient for effective, diversified investing — and investors are taking notice, with more funds flowing into private tech markets than before.

2 theses on what’s next

🤖 AI is more powerful when it can simulate how the world works (“world models”)

• The most powerful AI can integrate data from both the digital and physical worlds. A growing number of startups and frontier labs are developing world models to help AI understand and learn from the real world.

• A key challenge for startups is compute cost, where their Cost of Goods Sold (COGS) = Customer Acquisition Cost (CAC). The most successful startups leveraging world models will build products with high virality and retention while keeping margins high.

🤔 The best founders optimize for making many small mistakes on the way to a giant win

• It is better to make small mistakes (and to end up making big wins) than to only have small wins (only to end up making a big mistake).

• More small mistakes = more learning = more surface area for big wins.

1 tool I love

💊 AI to optimize healthcare referrals

• Floe Health helps primary care physicians make specialist referrals more efficiently. They’re currently formalizing a partnership with Mayo Clinic.

• You can learn more here.

More of my work 👇🏼

🎤 Decode videos | For top Berkeley and Stanford founders

• DECODE is the largest founder community co-hosted across UC Berkeley and Stanford. The DECODE annual conference focuses on helping founders in their earliest stages of starting a startup. I’ve had the honor of serving on the DECODE board since 2016.

• You can find the latest conference highlights video and playlist here.

🤖 Founders' prompt library

• My team and I have compiled the largest AI prompt library optimized for founders – especially fractional founders. This is not just another collection of prompts. It’s the largest and most comprehensive prompt library for founders.

• You can access it here.

👥 Pop up board

• I regularly host pop-up board, which is a unique 45-minute session where founders can present to a handpicked group of seasoned operators and executives, who have led teams as C-Suite at some of the world's top companies. This is founders’ chance to ask their toughest strategic questions, get tailored advice, and learn what it’s like to engage with a real board.

• You can find the application here.