Nikhil KamathInside Silicon Valley’s VC Playbook | WTF is Venture Capital? - 2025 Edition | Ep. 24
CHAPTERS
- 0:00 – 1:09
Setting the agenda: where the next decade’s tailwinds will be
Nikhil frames the episode as a forward-looking exercise for investors and young professionals: which sectors will benefit from compounding tailwinds over the next 10 years. The guests agree to keep the conversation practical, but acknowledge that venture, technology, and macro trends inevitably shape their viewpoints.
- •Goal: identify sectors with the strongest 10-year tailwinds for careers and capital allocation
- •Tension between ‘venture lens’ vs broader investing/career lens
- •How to think in scenarios rather than point predictions
- •Why “rate of change” feels higher now than in prior decades
- 1:09 – 5:11
Deedy Das: from engineering to Menlo Ventures, plus the Anthropic backstory
Deedy shares his journey from India to Cornell, then Meta/Google, to founding-team experience at Glean, and finally venture investing at Menlo. He also offers a concise origin story for Anthropic and how it differentiates itself in the foundation-model landscape.
- •Deedy’s background: CS at Cornell, Meta/Google, early team at Glean
- •Glean’s evolution: enterprise search → enterprise assistant/knowledge discovery
- •Menlo’s focus: seed/Series A across AI, SaaS, infrastructure; broader fund spans later stages too
- •Anthropic overview: OpenAI alumni, emphasis on text-to-text, strong coding/reasoning, enterprise-first business
- 5:11 – 11:38
Nikunj Kothari: operator-to-investor path via LinkedIn, startups, Opendoor, Meter, and FPV
Nikunj explains his shift from operator roles to investing, detailing product and operations lessons from multiple startups. He breaks down Opendoor’s iBuying model and Meter’s “networking-as-a-service,” then explains why he prefers concentrated seed/Series A investing at FPV Ventures.
- •Career arc: LinkedIn → multiple startups → Opendoor → Meter → Khosla (short stint) → FPV
- •Opendoor model: reduce 6% transaction friction with data-driven instant offers and inventory ownership
- •Meter model: replace CapEx-heavy networking installs with subscription pricing and full-service maintenance
- •FPV thesis: generalist fund focused on founder quality (“Founder’s Point of View”)
- 11:38 – 14:42
Niko Bonatsos: founder-first investing and why ‘too hot’ is a red flag
Niko introduces himself as a long-time early-stage investor who prioritizes exceptional technical founders over sector theses. He argues that the biggest risk at pre-seed/seed is chasing overcrowded, fashionable ideas—where differentiation collapses and prices rise.
- •Background: Greece/UK → Silicon Valley since 2009 → General Catalyst
- •Core philosophy: founders matter more than sector at pre-seed/seed
- •Avoid “super hot” themes where dozens of similar startups pile in
- •Examples of hot patterns: AI code gen, app builders, AI receptionists, RL tooling; inbound pitches often sound identical
- 14:42 – 27:31
What to avoid vs what becomes investable: CapEx, ‘non-tech,’ and AI-driven unlocks
The group debates whether any sector should be avoided outright. They converge on a nuanced view: every industry can be interesting, but venture-style returns often clash with high CapEx and slow adoption—until AI shifts ROI, efficiency, and buyer urgency.
- •Deedy: ‘avoid’ is hard; all sectors can generate opportunities depending on problem/approach
- •Nikunj: rising rates + efficiency pressure made many “uninvestable” legacy sectors suddenly buyers of AI
- •AI unlock mechanism: data becomes actionable; fine-tuning + automation converts latent data into value
- •Niko: as an early investor, the bigger danger is crowdedness rather than the sector label
- 27:31 – 38:50
Emerging AI shifts: data scarcity, reinforcement learning, evals, China’s model progress
They highlight underappreciated AI trends: exhaustion of public data, reinforcement learning environments, and the growing importance of evaluation (finding corner cases and exception handling). Deedy emphasizes China’s ability to produce strong models under hardware and capital constraints, challenging US-centric narratives.
- •“We’ve run out of public data”: model improvement shifts toward RL and other feedback-based methods
- •Evals as a new market: experts define edge cases; systems get refined beyond generic training
- •China factor: strong models despite limited access to cutting-edge GPUs; frontier capability is broadening
- •Embodied intelligence requires new physical-world datasets and data pipelines
- 38:50 – 53:19
Birth-rate decline, phones, and the ‘abundance’ debate: capitalism, inequality, meaning
The conversation pivots from AI capabilities to social consequences: declining fertility, digital dopamine replacing offline intimacy, and whether abundance leads to leisure or deeper inequality. They discuss how productivity gains can coexist with rising wealth concentration and status competition.
- •Birth-rate decline as a slow-moving macro force affecting growth, consumption, and social structure
- •Phones/social platforms as a plausible inflection (post-2009) reducing offline interaction and intimacy
- •Abundance scenario: shorter workdays, more leisure/arts—but inequality may widen (Piketty reference)
- •Meaning-seeking: religion and new “AI religion” possibilities as social cohesion mechanisms
- 53:19 – 1:00:55
Living in public: surveillance, kids’ digital identities, and the rise of ‘no-tech’ escapes
They argue privacy is eroding structurally—cameras, phones, sensor-rich cities—making “public life” the default. In response, they foresee growth in detox apps, dumb-phone hardware, retreats (Vipassana/silent retreats), and ‘offline-first’ products as people seek control over attention.
- •Assumption shift: behave as if life is public and trackable; digital footprints start in infancy
- •Kids’ interfaces trend: touch/voice-first; keyboard-centric computing feels alien to younger generations
- •Countertrend: silent retreats, offline enclaves, detox apps (Clear Space/Opal), and utility-focused dumb phones
- •Addiction vs mitigation: detox grows, but total screen time (e.g., YouTube share of waking hours) still rises
- 1:00:55 – 1:16:00
Dating apps and digital twins: from swipes to curated intros and offline experiences
They predict dating shifts away from swipe-based dopamine loops toward AI-curated matching, digital twins, and more structured offline meetups. They also discuss algorithmic inequality on dating platforms and the likely premium placed on real-world social experiences.
- •Future dating: AI handles early filtering/intro conversations; users arrive at higher-quality first dates
- •Digital twins concept: deeper matching using personal history/memory layers vs static profiles
- •Current dysfunction: ‘match dopamine’ > actual dating; distribution skew (small % of men get most dates)
- •Offline resurgence: curated events (e.g., dinner/cooking formats) as scarcity-driven alternatives to swiping
- 1:16:00 – 1:32:51
Investing scorecard: education, secondaries, aging, senior living—what compounds by 2035
Nikhil runs a structured “rate the thesis” exercise across sectors. They largely agree on premium/outcome-focused education in India, expect aging to expand healthcare demand, and strongly favor senior living as community-driven longevity infrastructure—while debating how big private secondary markets become.
- •Outcome-based premium education: strong tailwind in India and education-forward cultures
- •Private company secondaries: disagreement—structural tailwinds vs regulatory cycles and access gating
- •Older world: debate ‘older and sicker’ vs ‘older and healthier,’ but healthcare spend likely rises
- •Senior living thesis: community improves quality of life and longevity; stigma fades as economics and demographics change
- 1:32:51 – 1:47:28
Brand building in the AI era: content vs product, and David vs Goliath brands
They argue distribution is being re-written as search and performance marketing get disrupted; content becomes a primary discovery engine. A counterpoint is raised: attention without product quality produces churn; durable brands require both retention and discoverability, enabling smaller “David” brands to beat incumbents.
- •If traditional search ads weaken, content becomes the durable discovery surface for brands
- •Storytelling as a competitive advantage; young founders exploit low-cost distribution tactics
- •Counterargument: too much ‘attention on attention’—quality/retention is the compounding moat
- •David vs Goliath: incumbents lose brand equity with youth; artisanal/offline/bespoke brands gain appeal
- 1:47:28 – 2:06:13
Mobility and physical infrastructure: EV adoption, beauty/luxury tailwinds, live events
They examine where consumer spend concentrates: transportation, vanity, and experiences. The group expects EV share to grow (especially for new cars), but notes convenience, politics, and charging/battery constraints; they’re extremely bullish on beauty/luxury and moderately bullish on premium live experiences as ‘offline scarcity’ goods.
- •EV adoption: cost + charging convenience are bottlenecks; China’s EV economics highlight what free markets could accelerate
- •Prediction framing: new-car EV share by 2035 likely high; installed-base transition slower
- •Beauty/luxury: viewed as a top-performing category; male grooming and anti-aging as expanding segments
- •Live events: historically tough businesses, but scarcity and superfan monetization (e.g., Taylor Swift) create premium upside
- 2:06:13 – 2:52:16
Climate/energy, data centers, vices, and legal AI: where capital will cluster
They treat ‘climate tech’ as too broad, but agree energy abundance is central because AI is compute-hungry. They discuss data centers as a geopolitically strategic asset with operator-selection risk, then endorse prediction markets as durable “vice + entertainment” businesses—before closing with thoughts on Harvey/legal AI and India’s role in the AI race.
- •Climate/energy: energy production and density (nuclear, geothermal, fusion) matter more than labels; incentives and project finance complexity remain
- •Data centers: demand tailwind is strong, but execution depends on energy, real estate, hardware refresh cycles, and consolidation strategy
- •Vices/prediction markets: speculation is persistent; profitability strong if regulation aligns (Kalshi/Polymarket)
- •Legal AI: market is crowded; distribution matters now, but better products may disrupt incumbents; India’s AI challenge is talent/capital/GPU-stack depth vs strength in applications
