The Twenty Minute VCEventbrite Sold for $500M, Databricks $5B Raise at $134B Valuation & Why SaaS is Like Japan
CHAPTERS
- 0:00 – 2:01
OpenAI–Thrive partnership, and OpenAI’s sudden “code red” focus
The conversation opens on Thrive’s partnership around OpenAI, but quickly reframes it as yesterday’s story compared to OpenAI’s new “code red” posture. They debate what the partnership really signals and who benefits most from the halo effect.
- •Thrive/OpenAI partnership matters less than OpenAI’s abrupt shift to focusing on the core
- •“Code red” framing: OpenAI responding to competitive pressure from Google
- •Partnership as brand/attention arbitrage for Thrive
- •Unclear strategic upside for OpenAI beyond data access/vertical insights
- 2:01 – 4:07
Power laws inside power laws: why VCs double down on the one or two winners
Jason and Rory use the Thrive situation to discuss how venture outcomes concentrate in a tiny number of breakouts. They argue that once you have a winner, the rational move is to go as deep as possible with that founder/company.
- •Only a handful of deals truly move a VC’s fund outcomes
- •Board/relationship time allocation is itself power-law distributed
- •Turning funds into holding-company structures to concentrate exposure
- •“Go deeper with your winner” as a repeatable strategy
- 4:07 – 9:28
Databricks raising $5B at $134B: paying for growth vs. paying for certainty
They compare Databricks’ rumored valuation to public comps (especially Snowflake) and frame the core question: how much multiple is justified by incremental growth. The key nuance becomes Databricks’ ability to re-accelerate at massive scale.
- •Databricks vs Snowflake as a clean comp: ~$4B revenue, very different growth rates
- •The valuation question: extra multiple for extra growth (and duration of that growth)
- •Public-market scarcity of high-growth comps (Palantir as an outlier)
- •Re-acceleration at scale breaks traditional de-acceleration valuation models
- 9:28 – 15:59
Snowflake vs Databricks: oligopoly dynamics and the “agentic data” unknowns
The discussion shifts from valuation to market structure and product trajectories. They predict a long, bruising oligopoly fight—and note agents accessing enterprise data could reshape how analytics, CRM, and workflows are built.
- •Different origins: Snowflake (cloud warehouse) vs Databricks (AI/data movement roots)
- •Enterprise software tends toward oligopolies that “slug it out” for a decade
- •Agents as a new interface to enterprise data (rapid capability shifts in weeks/months)
- •Open questions: CRM as database + agents vs CRM platforms building their own agents
- 15:59 – 20:41
Security backlash: platform lockouts, data residency, and incumbents’ advantage
Jason argues the industry is underestimating security and data governance risks as agents proliferate. Examples of vendors being cut off highlight how incumbents may use security both as necessity and as strategic leverage.
- •Examples: vendor lockouts (Gainsight/Salesforce), Drift permanently removed, OpenAI security breach fallout
- •Ransom/data exfiltration raises stakes for platform ecosystems
- •Risk that enterprise buyers become less forgiving and more restrictive
- •Security as “revenge of the enterprise” and a rationale to push first-party solutions
- 20:41 – 25:21
Eventbrite at ~$500M and PagerDuty at ~2x revenue: the harsh math of low growth
They react to Eventbrite’s acquisition multiple and PagerDuty’s compressed valuation, interpreting it as the vulnerability of slow-growing public companies. Rory highlights that acquirers may see “fixable” assets, while Jason questions how often AI bolt-ons truly revive growth.
- •Public-company vulnerability: low valuation + 50% premium can force boards’ hands
- •Optimistic view: savvy buyers believe they can restore growth and create value
- •Skeptical view: few proven examples of legacy+AI mashups producing step-change outcomes
- •PagerDuty’s missed opportunity: distribution advantage but slow product expansion
- 25:21 – 32:09
The TAM Trap and “SaaS is like Japan”: saturated markets and shrinking seat growth
They develop the “TAM Trap” thesis: many SaaS markets are saturated, and adjacent expansions are crowded by other venture-backed vendors. Jason’s “SaaS is like Japan” analogy captures a world where seat-based growth slows structurally.
- •TAM Trap: public SaaS growth slowing suggests many companies hit market ceilings
- •Saturation vs CEO failure: too many companies crowded into adjacent categories
- •Overpayment only works with huge TAMs; finite TAMs require disciplined pricing
- •Seat-based growth pressure: fewer net new seats as efficiency rises
- 32:09 – 38:10
AI pricing and the seat-model debate: value-based, usage-based, and commoditization risk
They debate whether AI enables radically higher pricing via labor substitution—or whether competition rapidly erodes that pricing power. Workday’s warning and Twilio’s usage model become case studies for where pricing models may go.
- •AI can tap labor budgets and justify higher ARPU—if value is provable
- •Risk: multiple equivalent AI providers compress price from $500 to $100 quickly
- •Seat pricing is easy to count; value pricing is harder to measure and defend
- •Usage-based models (e.g., Twilio) may be structurally better positioned
- 38:10 – 40:18
What VCs really want: growth vs efficiency, and why ARR per employee keeps rising
Harry challenges the mixed message founders hear: “grow fast” and “be efficient.” Jason and Rory reconcile it: elite outliers can ignore profitability, but the whole system is shifting toward higher revenue per employee, especially in mature publics.
- •Top performers can raise regardless; efficiency matters less when growth is undeniable
- •Public SaaS: slower growth must be offset by FCF and operational efficiency
- •AI startups: often can’t hire fast enough to match demand; efficiency emerges naturally
- •Hiring bloat becomes a red flag; “grow 100% with ~50% headcount growth” as a new bar
- 40:18 – 43:38
The labor market future: three AI-era company archetypes that ‘don’t need people’
Rory outlines a sobering near-term labor implication across three buckets: public incumbents cutting for efficiency, foundation model companies spending on GPUs not headcount, and fast-growing AI apps whose traction outruns hiring. They remain long-term optimistic but acknowledge the short-term squeeze.
- •Mature companies reduce headcount to improve ARR/employee and protect margins
- •Model builders deploy capital into compute; headcount stays relatively small
- •AI apps can scale revenue faster than they can scale teams
- •Near-term labor softness in tech: capital intensity rises relative to labor
- 43:38 – 46:17
Google enters vibe-coding: cloning speed, defensibility, and where models will compete
They assess Google’s new Lovable/Replit-like product and what it signals about competitive timelines. Rory argues big-company prioritization constraints still matter, and model providers won’t necessarily invade every vertical app category.
- •Google’s clone shipped quickly but lacked key features (DB/OAuth) at launch
- •Founders no longer get years of runway before incumbents copy—sometimes months
- •Big companies can test many things, but sustained priority is the real constraint
- •Model providers likely focus on core wars; vertical app defensibility can still exist
- 46:17 – 55:10
AI in wealth management as a ‘real’ disruption bet—and how TAM discipline still applies
A spirited debate: Harry is skeptical of wealth-management outcomes and long build cycles, while Rory and Jason argue AI can finally unify taxes, trusts, and planning for the mass affluent. They stress segment selection, realistic TAM math, and patient compounding as valid venture paths.
- •Thesis: AI can automate fragmented, human-heavy financial workflows (tax, estate, trust)
- •TAM caution: don’t assume everyone pays $10k; segment realistically to avoid a new TAM Trap
- •Wealthfront/Schwab as examples of long-duration compounding businesses
- •Core idea: deliver ‘rich-person’ capabilities to broader markets via automation
- 55:10 – 1:08:05
Compounding vs the ‘relevance game’ in venture: momentum rounds, media, and conviction
Harry frames modern VC as a relevance/momentum game where fast follow-ons and hype matter. Rory counters that the ultimate filter is building a truly big company—momentum helps, but conviction on end-state matters more than short-term markups.
- •AI markets provide rapid validation and frequent follow-on rounds (momentum investing)
- •Relevance dynamics: brand, competitive rounds, and signaling matter more than before
- •Rory’s rule: prioritize high certainty of building a big outcome over short-term hype
- •A balanced approach: win hot deals when possible, but don’t ignore durable compounding plays
- 1:08:05 – 1:11:41
Quick-fire debate: Supabase at $5B vs Lovable at $6B—picks, risk, and defensibility
They close with a rapid comparison between infrastructure (Supabase) and front-end distribution (Lovable). Rory prefers the higher-upside category winner, while Harry prefers the harder, stickier database layer for stability and defensibility.
- •Rory: if vibe coding wins, the front-end captures more value; take the bigger upside
- •Harry: databases are harder to replace; switching costs and reliability create durability
- •Both acknowledge cloning/competition risk across the stack
- •Wrap-up includes gratitude and the live-show close
