Uncapped with Jack AltmanFormer CAA Talent Agent Turned Investor with $70B in AUM on AI and Venture Strategy | Ep. 29
CHAPTERS
AI cycle check: from cash-funded buildout to leveraged, existential spending
Jack asks whether the market feels bubbly again and how to read the fall 2025 cycle. Thomas frames the moment through past tech inflections (iPhone, NVIDIA/ChatGPT) and highlights a newer shift: AI infrastructure spend moving from cash-rich hyperscalers to players willing to use leverage and invest while cash-flow negative.
- •Seminal inflection points: iPhone/Apple beats; NVIDIA data center surge alongside ChatGPT
- •Oracle’s recent announcement as a meaningful new signal in AI infrastructure
- •AI capex historically funded by hyperscalers’ free cash flow; now leverage/cash-flow-negative actors are investing too
- •Competitive intensity among cloud/infra players is rising (Oracle, CoreWeave, OpenAI, Anthropic)
- •Higher stakes imply increased investor vigilance and selectivity
Deploying capital from seed to IPO: thematic thinking across public and private
The conversation turns to how Coatue decides where to allocate capital across stages and markets. Thomas explains Coatue’s “wide aperture” approach—building conviction in big themes, then expressing them through the best vehicles across public and private investing.
- •Coatue invests across inception through IPO; bucket allocation is guided by themes, not labels
- •Wide-aperture lens: public + private, US + global, cross-pollinating research
- •Personal roots in semiconductors inform AI thesis: no AI without chips
- •Capital deployment follows conviction layers rather than rigid stage mandates
- •Public and private conversations both improve research quality
AI’s foundational layer: semis, data centers, and power as investable bottlenecks
Thomas lays out the “layer one” AI stack where he has the highest conviction: semiconductors, data centers, and power. He names both public winners and private innovators, arguing these constraints determine the pace and economics of AI progress.
- •Semis as unavoidable foundation (NVIDIA; Broadcom/Hock Tan; private innovators like Cerebras)
- •Data centers as scaling constraint tied to infrastructure buildout
- •Power becomes an enabling bottleneck alongside chips and compute
- •Opportunities exist in both public and private markets for foundational layers
- •High-conviction investing starts where dependency is non-negotiable
Power plays for AI: nuclear and gas turbines, plus ‘behind-the-meter’ deals
Jack asks specifically how to invest in power. Thomas focuses on practical, near-term sources—fission nuclear and gas generation—citing corporate power procurement and supply constraints as signs of durable demand.
- •Nuclear (fission) as an investable near-term solution; fusion still uncertain
- •Behind-the-meter arrangements linking hyperscalers directly to generation assets
- •Constellation Energy examples; reactor reactivation efforts
- •GE Vernova and gas turbines; US natural gas advantage
- •Generator capacity constraints (sold out for years) signal sustained demand
Up the AI stack: model concentration, then a fuzzier application layer
Thomas explains “layer two” conviction in foundational models, suggesting the market is converging on a handful of major model providers. He contrasts that with the application layer, where outcomes are less clear and require more exploration.
- •Foundational models viewed as core value creators (OpenAI, Anthropic, Google; Meta/Microsoft/Amazon TBD)
- •Models power transformative apps like ChatGPT and reset software expectations
- •Confidence declines moving from models to applications due to fragmentation
- •Market structure appears to be coalescing around a small set of leaders
- •Application investing requires more experimentation and differentiated insight
The data layer and the ‘system of record’ decline: Workday opens up
Thomas argues the era of SaaS platforms locking up enterprise data is ending, pointing to Workday’s integration posture (Snowflake/Databricks) as a key signal. The chapter explores how databases and data platforms become the hub while SaaS shifts toward agent-driven outcomes.
- •Workday’s move to plug into Snowflake/Databricks as a strategic inflection
- •Data portability and mergeable datasets become critical for AI and agents
- •SaaS differentiation shifts from storing data to delivering best-in-class agents (HR/finance)
- •Snowflake/Databricks positioned as foundational enterprise hubs
- •Debate: system of record ‘dead’ vs evolving into validation/coordination roles
Everything gets recorded: enterprise memory, compliance agents, and new norms
Thomas predicts default “record on” for enterprise interactions within a few years, enabling AI systems to extract knowledge automatically. He and Jack discuss the trade-offs—security, consistency, and privacy—then Thomas offers a compliance-driven case for proactive remediation.
- •Prediction: meetings and interactions become continuously recorded by default
- •Enterprise intelligence is embedded in Slack/email/Zoom/in-person interactions
- •AI can replace manual notes with searchable, queryable knowledge extraction
- •Compliance use case: agents flag and remediate issues early to prevent blowups
- •Cultural shift: assumption that anything said on Zoom may be recorded
From call recordings to performance: Gong and the rise of ‘interaction intelligence’
Using Gong as an example, Thomas explains how generative AI makes recorded interaction data far more valuable. The focus is on extracting what top performers do differently and compressing learning curves across teams.
- •Gong re-accelerates as genAI increases the value of call corpora
- •Extracting best-practice sales behaviors from top reps at scale
- •Reduced ramp time and faster training via AI-driven insight
- •Pattern recognition across customers, competitors, and objection handling
- •A preview of similar “interaction intelligence” across the enterprise
Winning in venture: be the traveling fisherman, not the fixed riverbank owner
Thomas describes Coatue’s venture strategy as mobility and breadth rather than owning a single “spot” in early-stage venture. He explains how cross-stage investing reduces the ‘one shot’ dynamic and enables repeated opportunities to earn into great companies over time.
- •Analogy: traditional top-tier VC firms occupy prime riverbank spots; Coatue travels to find opportunities
- •Cross-stage flexibility (seed to IPO) changes the competitive game
- •Best research comes from mixing public and private exposure
- •Zero-sum dynamics in venture can drive unhealthy behavior; public markets feel more shareable
- •Mindset: you may miss a round, not the company—later entries can still be great
Operating early-stage and navigating conflicts: disclosure, trust, and compliance
The discussion turns to conflicts created by thematic investing and owning multiple companies in a category. Thomas distinguishes true conflicts (board/large ownership) from manageable overlap, emphasizing disclosure to founders and rigorous information controls.
- •True conflict line: major ownership/board role in a direct competitor is avoided
- •Positioning: aim to be the founder’s “second call” with network and perspective
- •Disclosure first: tell founders about potential or perceived conflicts early
- •Trust/reputation plus strict compliance (SEC-regulated context) to prevent leakage
- •Benefits of broader portfolio: domain depth, network access, and field-level visibility
Wide-aperture curiosity: strengths, trade-offs, and where specialization wins (crypto)
Thomas defines “wide aperture” as curiosity unconstrained by narrow mandates, citing great investors’ curiosity as a common trait. He acknowledges the downside: some domains (like early crypto) reward deep specialization more than broad pattern-matching.
- •Wide aperture = pursue information first, decide relevance later
- •Curiosity as a key investor trait (examples: Druckenmiller, Andreessen)
- •Trade-off: broad lens can underperform where new domains need deep specialization
- •Crypto cited as an area where early rabbit-hole specialization mattered
- •No one formula: approach depends on market structure and novelty
Founder ‘star quality’: CAA lessons, magnetism, and polarizing dominance
Thomas draws from his CAA background to argue that “star quality” is real and investable in founders. He shares stories about presence and magnetism (Tom Cruise, Colin Farrell) and connects that to founder energy, narrative clarity, and competitive aura (e.g., Travis Kalanick).
- •Star quality: rare presence that changes room dynamics; you ‘know it when you see it’
- •Founder evaluation blends magnetism with a compelling market worldview (Snap example)
- •Polarizing founders can still be exceptional if their pull is strong (Uber/Travis)
- •Founders’ narrative can be prescient about broader societal/institutional shifts
- •People matter across stages, including late-stage/public investing
How Coatue decides: collaborative momentum over a single IC showdown
Thomas explains Coatue’s internal decision-making style: early and broad input, iterative debate, and “deal momentum” rather than a one-time investment committee verdict. The chapter highlights how collaboration improves ideas and how deals often fade rather than being explicitly killed.
- •IC exists but isn’t the decisive ‘smoke from the chimney’ moment
- •Early solicitation of views from public and private specialists
- •Deals build internal groundswell; kinks get worked out through repeated conversations
- •‘No’ often happens via loss of momentum rather than a dramatic meeting
- •Cultural expectation: input isn’t credit-taking; it’s how the firm improves
Firm model, meritocracy, and growth: middle ground between VC and hedge fund culture
They discuss organizational trade-offs between equal partnerships and more centralized structures, and the different feedback loops of hedge funds (annual scoring) versus venture (multi-year outcomes). Thomas argues Coatue sits between East Coast finance and West Coast VC, blending competitive metabolism with long-term company-building.
- •Faster impact and advancement possible without rigid GP timelines
- •Avoiding dead weight at both junior and senior levels; contributions must match economics
- •Hedge funds have annual performance clarity; venture outcomes can take 6–7 years
- •Crossover funds brought a competitive metabolism; entrepreneurs brought another kind of competitiveness
- •Running East + West Coast cultures is difficult; Coatue tries to balance both
Golf and presence: integrity, mentorship, and relationships off the phone
Thomas describes golf as life-changing primarily for its social and mentorship value, with integrity and self-competition as important subthemes. He connects golf (and surfing) to being present and building real relationships in an always-on, phone-driven world.
- •Golf as a relationship engine: four uninterrupted hours builds trust and depth
- •Integrity test: self-reported scoring and constant temptations to cheat
- •Competing against yourself; resilience under randomness and adversity
- •Handicap system enables competition across age, gender, and skill levels
- •Parallel ‘presence’ activities: surfing, vinyl, and hobbies that force focus
Working with family and mentoring juniors: trust, fairness, and the ‘gift wrap’ lesson
Thomas reflects on working with his brother Philippe: deep trust reduces politics but makes it hard to ever ‘turn work off.’ He closes with advice for young professionals: focus is a luxury early on—use it to develop craftsmanship, illustrated by his meticulous gift-wrapping story at CAA and how it created opportunity.
- •Family partnership advantage: motives are trusted; politics minimized
- •Economics discussions become rare because fairness is assumed
- •Downside: work can dominate family time; requires deliberate off-switch activities
- •Mentorship as a core value; evaluate leaders by how former colleagues describe them
- •Early-career advice: treat focus as a luxury; master small responsibilities (gift-wrapping/craftsmanship, concise models)
