The Twenty Minute VCPeter Singlehurst: Lessons from Turning Down Stripe, Coinbase and Losing Money on Northvalt
CHAPTERS
- 0:31 – 4:08
Accidentally Building Baillie Gifford’s Private Investing Capability (2014 origin story)
Peter recounts how Baillie Gifford’s move into late-stage private investing started almost by accident: a public-markets team seeing Airbnb/Spotify-scale companies staying private. He explains why he volunteered, what he underestimated, and why he’d advise his younger self to be less of a “purist” about fund structures.
- •Moved from a $50B public strategy into evaluating scaled private companies
- •Early private opportunity set included Airbnb/Spotify-like businesses
- •Volunteering to “own” privates internally created a new capability
- •Lesson: distinguish true edge from non-essential “differences”
- •Regret: being too rigid on permanent-capital vehicles vs traditional funds
- 4:08 – 7:04
When Losses Aren’t Mistakes vs When They Are: Intarcia and Northvolt
Peter separates bad outcomes caused by known uncertainty from true analytical mistakes. He uses Intarcia (biotech/FDA risk) as a painful-but-expected risk and Northvolt as an execution and cap-table misalignment mistake they should have judged better.
- •Not all losing investments are mistakes; uncertainty sometimes resolves against you
- •Intarcia: FDA rejection/bankruptcy as known risk in biotech
- •Northvolt: error was over-weighting “must exist” narrative vs execution capability
- •Hindsight reveals signs; they reduced follow-on capital as issues emerged
- •Financing round structure can create misalignment and worsen outcomes
- 7:04 – 8:07
Narrowing the Mandate: From Product Risk to Growth-Stage Business Model Risk
Peter explains Baillie Gifford’s evolution toward true growth-stage investing—companies with product risk largely removed. The focus becomes scalability, business quality, and eventual high returns on equity rather than early technical/regulatory uncertainty.
- •Today they’d avoid Intarcia-like product/regulatory risk
- •Target: growth-stage companies where product works and customers pay
- •Core underwriting: scalability and business-model quality
- •Portfolio focus has narrowed over 5–6 years to where they have edge
- •Shift from “can it work?” to “can it become exceptional at scale?”
- 8:07 – 14:00
What “Scalability Risk” Looks Like: Wise and Return on Equity as a North Star
Using Wise as the concrete example, Peter describes how they assess whether a business can become many times larger and highly profitable relative to its equity base. He argues venture markets often under-emphasize return on equity, leading to overcapitalized companies.
- •Median entry profile: ~$200M revenue, ~70% YoY growth, modest losses
- •Wise example: grew from ~$50–60M revenue to multi-billion revenue
- •Key metric: high return on equity at scale (public-market mindset)
- •Venture’s low ROE focus can distort company formation
- •Overcapitalization metaphor: “foie gras-ing” startups with too much cash
- 14:00 – 17:40
AI Investing: Where Value Accrues, Commoditization, and When to Pay Up
Peter and Harry debate where durable advantage will sit in AI. Peter notes they own infrastructure/distribution-adjacent AI exposure (e.g., Databricks, Tenstorrent) but avoid big LLM model bets due to uncertainty about defensibility, while emphasizing disciplined selectivity rather than rigid valuation rules.
- •They avoid big LLM investments пока competitive advantage is unclear
- •Comfort higher in infrastructure/distribution layers than model layer
- •Open-source pressure (e.g., DeepSeek) accelerates LLM commoditization
- •Discipline ≠ refusing high multiples; discipline = choosing when to pay up
- •Concern: AI-era growth rates may reset expectations for non-AI companies
- 17:40 – 19:42
Baillie Gifford’s 10 Questions Framework: Growth, Moats, Culture, Financials, Valuation
Peter outlines their internal diligence framework: 10 questions across four buckets. The approach emphasizes long time horizons, evolving competitive advantage, culture/execution fit, rigorous financial analysis (including ROE), and public-market-like valuation discipline.
- •Framework splits into: long-term growth, enduring success factors, financials, valuation
- •Moats must be assessed as evolving with time and scale
- •Culture and execution are central, not a checkbox
- •Financial analysis targets high ROE/returns on capital over time
- •Valuation is anchored to long-term intrinsic value vs price today
- 19:42 – 22:35
Defensibility Beyond Product: Founder-Led Bias and Non-Founder Turnarounds (Vinted)
The conversation turns to what makes advantages enduring when products can be copied or AI can replicate features. Peter argues the best moats often live in systems, strategy, and culture; he notes their biggest holdings skew founder-led, while acknowledging exceptional “re-founding” by non-founders like Vinted’s CEO.
- •Enduring moats often sit in strategy/process/culture, not just product features
- •Bending Spoons as example: integration playbook and shared tooling
- •Portfolio skew: ~9 of top 10 positions are founder-led
- •Founder-led bias partly due to survivorship by growth stage
- •Non-founder excellence exists (e.g., Vinted transformation)
- 22:35 – 28:10
Modeling Upside: 5x Scenarios, Probability Thinking, and Patience on Deployment
Peter describes how they model every deal to a 5x outcome and then stress-test the assumptions and likelihood. He explains why 30–50% odds of 5x is attractive, how fund structure affects recycling, and why they largely sat out 2022–2023 due to valuation games.
- •Standard underwriting: model to a 5x upside for comparability
- •Acceptable probabilities are modest; 30–50% for 5x is compelling
- •Random 5x odds in public markets ~5% (their reference point)
- •Capital recycling depends on vehicle; always maintain a high “buy bar”
- •They deployed little in 2022–2023 amid convert/valuation distortions; more in 2024
- 28:10 – 31:59
Global Opportunity Set and Macro Risk: Investing Off the Beaten Path
Peter defends being global generalists with a large universe (2–3k companies) and investing across multiple countries. He addresses concerns about liquidity pathways in places like Brazil/India by emphasizing pricing risk appropriately and maintaining a long time horizon.
- •Broad universe and global remit expands opportunity set
- •They invested across six countries in a year (not for exoticism)
- •Macro/liquidity risk exists; investors must be compensated via price
- •Planning reserves and dilution matters, especially for cash-hungry models
- •Profitability reduces dilution risk and changes capital needs profile
- 31:59 – 34:45
The ByteDance Case: China Business Strength, TikTok Risk, and Liquidity Paths
Peter argues ByteDance is misunderstood in the West and that their thesis is primarily based on China’s massive profit engine (Douyin/Toutiao), not TikTok US. He explains how they think about a TikTok ban scenario and potential liquidity through IPO or buybacks.
- •ByteDance viewed as exceptionally profitable in China (ads + growing e-comm)
- •Toutiao and Douyin are core products powering monetization
- •Base case: even without TikTok US, they still see 5x path
- •Invested around 2019; growth since then has been dramatic
- •Liquidity could come via IPO (more likely Hong Kong) and share buybacks
- 34:45 – 39:05
Why Go Public at All? Focus, Disclosure Burdens, and the Rise of Large Secondaries
Peter explains why companies increasingly stay private longer: operational focus and avoiding public-market distractions. He outlines reasons to eventually go public (employee liquidity, acquisition currency, regulation) and predicts company-facilitated secondaries (and possibly private dividends) will grow as an alternative liquidity mechanism.
- •Public-company life is hard: reporting, misaligned owners, competitor visibility
- •Staying private can improve focus and business-building
- •Employee liquidity increasingly served by large secondary rounds
- •Public currency helps acquisitive companies; public status can aid regulators
- •Private dividends and structured secondaries may become more common
- 39:05 – 48:05
From ‘Opportunists’ to Institutional Owners: Growth-Stage Market Structure and Anduril
Peter describes how late-stage private ownership shifted from public-market firms to a mixed set of hedge funds and venture players during the ‘stay-private-longer’ era, with shakeouts post-2021. He uses Anduril to illustrate why companies seek investors experienced owning at scale and transitioning to public markets.
- •Shift: scaled companies stayed private, pushing out traditional public owners
- •Vacuum filled by hedge funds and early-stage investors spinning up growth funds
- •Post-2021: consolidation and professionalism increased in growth stage
- •Anduril round favored public-market-style long-term owners
- •Ownership quality matters for rational pricing and company support over time
- 48:05 – 50:50
How Baillie Gifford Actually Makes Decisions: Funnel Metrics, 10-Q Essays, IC Rhythm
Peter shares their internal operating cadence—from meeting 1,000 companies to making 11 new investments. He details how diligence becomes an essay-style ‘10-Q,’ debated weekly, then decided by a small investment committee, and explains check-size ranges and reinvestment rules.
- •Annual funnel example: 1,000 meetings → 600 rounds → 65 first cuts → 30 deep dives → 11 investments
- •Team of ~10, supported by broader ~170 public growth investors
- •Diligence output is essay-based (no PowerPoint) using 10 questions
- •Weekly cadence: Thursday team debate; Friday small IC (four people) decides
- •Checks range ~$10M–$150M; reinvestments may require 5x if doubling down
- 50:50 – 1:00:45
Missed Bets and Over-Modeling: Stripe Follow-On and Turning Down Coinbase
Peter candidly revisits two regrets: not joining Stripe’s down-round follow-on and passing on Coinbase. He attributes misses to doubts about product expansion progress (Stripe) and an over-engineered model that missed the bigger outcome (Coinbase).
- •Stripe: invested at ~$30B; skipped later round around ~$50B—now seen as mistake
- •Reason: questions about broader software suite progress; chose to ‘wait and see’
- •Databricks: did pro rata at higher valuations; distinction vs “doubling down”
- •Coinbase miss: overly elaborate modeling led to wrong conclusion
- •Meta-lesson: over-intellectualizing can obscure obvious breakout potential
- 1:00:45 – 1:12:05
Quick-Fire and Closing: Convictions, Bending Spoons, LLM Valuations, and Long-Term Optimism
In quick-fire, Peter argues for being a global generalist, names Bending Spoons as his one 10-year hold, and declines to buy any top LLM companies due to unclear durable advantage. He closes with optimism about the growth-stage pipeline: many experiments, strong human capital, and a more balanced capital environment.
- •Contrarian belief: you can be a global generalist yet specialist in growth equity
- •One-stock pick: Bending Spoons (tools + culture + M&A integration playbook)
- •If couldn’t fail: would take extreme risk, e.g., early-stage biotech
- •LLM choices: buy none; advantage likely in distribution rather than models
- •Optimism: more venture experiments reaching growth stage + better operators + healthier capital balance
