Anu Hariharan – Lessons in Growth Investing - [Invest Like the Best, EP.198] — Summary & Takeaways · Colossus (Invest Like the Best / Business Breakdowns)

Anu Hariharan – Lessons in Growth Investing

Quick Overview

Anu Hariharan, partner at Y-Combinator Continuity Fund and former investment partner at Andreessen Horowitz, systematically outlines the core framework of growth investing in this interview. The most weighty judgment of the entire episode: she believes that global digital transformation is currently only about 20% complete, from the $8 trillion internet economy to an expected $60 trillion scale, the opportunity is far greater than market consensus—especially in B2B wholesale e-commerce (only 8% online) and fintech.

1. Use Business Models Instead of Industries to Classify Investment Targets

Anu Hariharan believes that in the growth stage, companies should first be classified by business model rather than by industry. She points out that technology has permeated all industries, but the metrics for the same business model are highly consistent: for marketplaces, look at GMV (Gross Merchandise Volume), take rate, and contribution margin; for advertising businesses, look at revenue per user; for SaaS, look at LTV/CAC. She summarizes roughly 9–10 common business models. The essence of growth-stage investing is to verify: after product-market fit, can this business model continue to generate leverage at scale?

> "Whether B2B or B2C, if your business model is a marketplace, your metrics are similar." — Anu Hariharan

II. Winner-Takes-All Is Not Absolute: DoorDash Breaks the Consensus

Anu believes that the internet e-commerce sector is not naturally winner-takes-all, and each market can accommodate at least 2–4 players — the DoorDash case is a key proof point.

2012 YC Application: DoorDash was originally called PaloAltodelivery.com. The founders observed that: (1) most competitors chased urban areas, but suburban orders have higher average value and a more abundant supply of delivery drivers; (2) drivers in the suburbs earned higher hourly income due to order aggregation, and the low-cost customer acquisition efficiency was 10 times better than Uber’s.
2016 Uber Eats launched, and the market widely viewed it as a zero-sum game, but four years later DoorDash became the #1 player in U.S. market share.
Selection strategy: 70% of customers stay because of selection; merchant workflow integration (order queuing, accounting, invoice processing) creates a lock-in effect.
Conclusion: Economies of scale exist, but it is not a monopoly — the market can accommodate 2–4 players.

3. Market Priority: Who Is the Hardest? — Stakeholder Priority in Network Effects

Anu cites Chris Dixon’s framework: Market platforms must continuously determine which side is "the hardest" at the moment and subsidize it accordingly.

Airbnb early stage: The guest side was easy to acquire (driven by events), but the host side was extremely difficult, so all subsidies went to hosts, with customer acquisition cost being zero.
Later stage: Once supply was locked in, the demand side became harder, and subsidies shifted to guests.
General rule: At each stage, ask "which side is the hardest today," and adjust the subsidy strategy accordingly.

4. Growth-Stage Founder Assessment: Three Dimensions

Anu compares the differences in founder assessment between early and growth stages, with the growth stage placing greater emphasis on execution evidence rather than potential.

Dimension	Early Stage (Series A)	Growth Stage (Series B/C)
Team	Founder's unique insight and speed of execution	Mental clarity: ability to foresee the company's path to at least $3B–$10B
Learning and Scaling	Whether MVP is iterated quickly	Team-building ability: when scaling from 5–10 to 30+ people, can the founder transition from a DIY-style CEO to a company builder
Perseverance and Resilience	Persistence in the early days of entrepreneurship	Has experienced at least 2–3 near-death experiences (e.g., DoorDash, Rappi)

Brex case: Founders Enrique and Pedro took an accounting course in the second week of their YC batch, just to validate whether Brex could become a $10B company—this kind of mental clarity is once in a decade.
FAIR case: CEO Max Rode was initially told by feedback that "the team is afraid of him." He proactively hired a coach and engaged in continuous self-reflection, successfully transforming.

5. LATAM: An Underappreciated Growth Frontier, but Risks Cannot Be Ignored

Anu believes Latin America is the most underappreciated growth market today, arguing the case through the example of Rappi.

Data chain: LATAM has a population of 650 million, 70% internet penetration, and a GDP per capita of $9,700 (comparable to China), yet e-commerce penetration is only 4% (China: 22%).
Counter‑intuitive advantage: The average order value in LATAM is about $20 (vs. $30–$35 in the US), but delivery costs are less than 10% of the order value (vs. 50% in the US), due to low minimum wages and a supportive socioeconomic structure.
Risks: Fragmented regulation (operations across 9 countries), difficulties in managing talent across time zones, and a scarcity of early‑stage investors.
Rappi case: A 2016 YC company, it expanded to 9 countries within four years. Anu describes its perseverance and resilience as "unmatched."

6. Four-Dimensional Model for Hard-Tech Investing: The Case of Boom Supersonic

Anu uses the four-dimensional model of scientific risk, engineering risk, market/commercialization risk, and funding risk to evaluate hard tech.

Boom Supersonic: Scientific risk low (Concorde has proven feasibility, but poor economics); engineering risk high (relies on software simulation and manufacturing iteration); market risk low (has secured contracts from 20+ airlines, including Japan Airlines); funding risk moderate (can leverage supplier financing, such as GE's $500 million engine project).
Core belief: Founder Blake's focus and ability to attract talent are the most critical variables within engineering risk.

7. Valuation: Art, Not Science, But Must Avoid the 'Next Round Valuation Trap'

Series A/B: Valuation ≈ Investment Amount ÷ Lead Investor's Expected Ownership (20–25% or 7–10%).
Post-Series C: Focus on dollar returns rather than multiples—$100M invested in a $3B valuation company, if future value is $15B, return is $500M.
Biggest risk: Overvaluation leads to difficulty in the next round (down round), affecting hiring and morale. Anu emphasizes that she sets a maximum valuation threshold, but will not give up on truly upside-potential companies because of valuation—"Your secret weapon is upside probability, not downside model." (citing Marc Andreessen's teaching)
Stripe case: 3–4 years ago, the market viewed its $20B valuation, but if 2 new products achieve 5% market share, upside could reach $50–$100B.

8. Global Digital Transformation: Only 20% Complete, B2B Wholesale E-commerce is the Biggest Gold Mine

Total: The share of the global market capitalization attributable to the internet economy is expected to grow from less than 10% ($8T) to over 15% ($60T).
B2B Wholesale E-commerce: In the $16T U.S. market, only 8% is online, and 49% still transacts via phone/fax. FAIR covers only the retail sub-market ($670B), leaving large verticals such as aerospace, chemicals, and industrial goods untouched.
FinTech: It is expected that over the next decade, 3-5 companies will reach $100B+ valuation, 10 companies will reach $50B+, and 2,000 companies will reach $10B+.
EdTech: Lambda School addresses the software developer shortage (the U.S. faces a shortfall of 200,000-500,000 annually, while only 50,000 computer science graduates are produced); OutSchool leverages Zoom to disrupt extracurricular tutoring.

9. Growth-Stage Investor Ecosystem: Many Participants, but Few Understand Entrepreneurship

Surface phenomenon: YC Demo Day has 2,000+ investors, but no more than 10–15 investors at the growth stage can write $100M+ checks.
Core difference: New entrants (e.g., public market funds, PE) do not understand the "difficult years" that startups inevitably go through — nearly every company valued at over $10B has experienced 1–2 tough years. Pattern recognition (such as zero-sum games) is a dangerous tool that can mislead investors into exiting.
New role of SPACs: SPACs led by sponsors who understand entrepreneurship (such as Reid Hoffman, Miki Malka) may be useful for regulated fintech companies, but whether they can replicate a Snowflake-level IPO remains to be seen.

Mentioned Companies

Company	Guest Attitude	Key Data
DoorDash	Bullish (Case Study)	2012 YC company, #1 US market share in 2020; higher suburban order value, customer acquisition cost 10x better than Uber
Airbnb	Reference Case	Subsidy strategy changes with market stage: early subsidies for hosts, later subsidies for guests
Instacart	Reference Case	No specific data given
Brex	Bullish (Case Study)	Founder anticipated company path to $10B+; precise product roadmap
Rappi	Bullish (Case Study)	2016 YC company, covered 9 countries in 4 years; LATAM delivery cost < 10% of order value
Boom Supersonic	Risk/Bullish Balanced	Obtained 20+ airline contracts; high engineering risk, but low scientific risk
Stripe	Bullish (Upside Underestimated)	Market once valued at $20B, actual could be $50-100B
Monzo	Bullish (Case Study)	4.5M customers, NPS > 70 (rare in banking), fully product-driven growth
FAIR	Bullish (Case Study)	On the brand side, average each brand brings in 5 retailers, zero customer acquisition cost
Lambda School	Bullish (Trend)	Addresses the US annual shortfall of 200-500k developers, graduate salaries rise from $15-25K to $75K
OutSchool	Bullish (Trend)	Zoom tutoring exploded during the pandemic, expected to continue post-pandemic
Robinhood	Bullish (Trend)	Global replicas (India, Indonesia, LATAM)
Uber	Reference Case	Brazil/Mexico are the most important non-developed markets for ride-sharing

Judgments Worth Remembering

1. Business model classification is more effective than industry classification—Anu Hariharan

Market, advertising, SaaS, and 9–10 other models each have a quantifiable set of metrics that are universal across industries.

2. Winner-take-all is not a natural law of the internet—Anu Hariharan

The DoorDash case shows that scale effects can accommodate 2–4 players; consensus is dangerous.

3. "Mental clarity" in founders during the growth stage is a rare trait—Anu Hariharan

Brex's founder took accounting classes during YC batch just to validate the business model—such foresight appears once in a decade.

4. A four-dimensional model for evaluating hard tech: science, engineering, market, capital—Anu Hariharan

Boom: low science risk, high engineering risk, low market risk, medium capital risk—depends on the founder to solve engineering and capital.

5. 70% of a partner's time should be spent on upside probability, not downside modeling—Anu Hariharan, paraphrasing Marc Andreessen

"You will lose 1x of your money, but if you miss a 10x opportunity, you lose the entire fund."

6. The hidden advantage of LATAM e-commerce: $20 order value, delivery cost < $1—Anu Hariharan

LATAM order values are close to the US ($30–$35), but delivery costs are extremely low; the socioeconomic structure supports high profitability.

7. B2B wholesale e-commerce is the largest untapped gold mine in the internet economy—Anu Hariharan

The US $16T market is only 8% online, 49% still uses phone/fax; verticals (aviation, chemicals, industrial goods) are wide open.

8. No more than 10–15 investors can write $100M+ checks at the growth stage—Anu Hariharan

On the surface competition is fierce, but investors who truly understand entrepreneurship are scarce; founders should choose carefully.

Anu Hariharan – Lessons in Growth Investing - [Invest Like the Best, EP.198]

In plain words