The Twenty Minute VCAre Burn Multiples BS in an AI World? & Sam Altman Needs $1TRN of Energy
CHAPTERS
- 0:00 – 4:09
Burn multiples in 2025: why AI companies look ‘efficient’ while burning more cash
The discussion opens by unpacking Iconiq’s software report: AI-native companies often have far worse cash flow margins than non-AI peers, yet show better burn multiples because growth is so fast. Jason explains what burn multiple measures and why it can make heavy spend look “capital efficient” when ARR is compounding quickly.
- •AI-native sub-$100M ARR firms show extremely negative free cash flow margins but strong burn multiples due to rapid growth
- •Burn multiple = dollars burned per net new ARR; fast ARR expansion can offset high burn
- •Metric can mislead: capital efficiency ≠ profitability or runway safety
- •Examples used to illustrate how large raises can still look ‘efficient’ if ARR ramps fast
- 4:09 – 12:58
When burn multiples break: hidden assumptions about ARR quality, churn, margins, and CapEx
Rory argues burn multiple remains useful, but only if you remember its embedded assumptions—many of which no longer hold in an AI era. He details how overstated ARR, masked churn, changing gross margins, and ignored CapEx can all distort the metric.
- •Burn multiple assumes ARR is ‘real’ and durable; in practice, ARR quality can be questionable
- •Hypergrowth can hide churn because churned cohorts are small relative to the new base
- •Changing gross margins (e.g., token costs) and shifting economics reduce comparability
- •CapEx—critical for model/data-center businesses—often isn’t reflected
- •Better validation: cross-check with GAAP revenue run-rate changes and other ‘honesty’ checks
- 12:58 – 16:57
Fundraising reality check: good metrics, no funding, and the new ‘haves vs have-nots’ market
Harry shares founder frustration: companies with strong burn multiples and growth are still being ignored by VCs. Rory and Jason explain why: sub-scale businesses can be unattractive to venture when investors price on fundamentals rather than hope, and when the market demands clearer path to large outcomes.
- •VCs increasingly split the market into breakout winners vs everyone else
- •At smaller revenue scale, many deals lose option value for VC-style outcomes
- •Shift from ‘pricing on hope’ to ‘pricing on multiples/fundamentals’ shrinks VC appetite
- •Advice: prioritize runway and fundraising certainty over optimizing price
- •Absolute burn and cash balance matter as much as burn multiple
- 16:57 – 21:00
‘Take the deal’: why 2021-style fundraising advice has become dangerous
Jason criticizes board and investor guidance that assumes great growth guarantees financing. Both guests stress that in 2025, founders should close reasonable rounds when available—especially if they’re not the category-defining breakout—and plan as if capital will remain scarce.
- •“Triple-triple/double-double means you’re golden” is framed as bad 2025 advice
- •Some investors are ‘living in the past,’ encouraging founders to wait and optimize price
- •Founders should de-risk by raising when terms are decent, not timing the market
- •Operate with a conservative cash mindset even with strong metrics
- •Non-breakout companies should not spend like #1 contenders
- 21:00 – 28:01
Kingmakers, ‘Postmates effect,’ and the wall-of-money dynamic in AI categories
The conversation turns to “kingmaker” firms and how their backing deters investment in close competitors. They compare prior cycles (where #2/#3 could still win) with AI’s current urgency, where buyers want a trusted default and capital piles into perceived leaders.
- •Investors hesitate to fund #2/#3 when a kingmaker-backed leader exists (Harvey/Abridge examples)
- •2021’s ‘Postmates effect’ made #2/#3 investing more acceptable due to abundant capital
- •AI buying is urgent and brand-led; customers want an obvious default quickly
- •Follow-on capital concentrates behind leaders, creating a self-reinforcing ‘wall of money’
- •Counterpoint: deterrence varies by customer base; Valley-centric markets amplify the effect
- 28:01 – 33:54
Peak AI valuation madness vs genuine labor-to-software budget transfer
Harry questions whether current pricing is irrational exuberance or justified by a massive shift from human labor budgets to software/AI spend. Rory frames two possible futures: either explosive productivity and huge AI revenues validate prices, or a painful reset occurs if the transfer disappoints.
- •Examples of extreme valuation: multi-billion prices on minimal revenue/progress
- •Thesis: valuations require a major labor-to-software spend transfer to be right
- •Rory’s bifurcation: either AI leaders reach massive revenues fast, or markets reprice harshly
- •Jason: B2B AI adoption is early; direction likely correct even if pricing is risky
- •Risk focus: whether VC loss ratios and underwriting assumptions are accurate at these entry prices
- 33:54 – 38:58
Public markets temperature check: Figma, Klarna, StubHub, and IPO pricing pressure
They assess what recent post-IPO drawdowns signal: are markets turning and will IPO appetite weaken? Rory argues the lesson is investors demand a bigger discount for new listings, while Jason notes multiples are still surprisingly generous for modest growth compared to historical baselines.
- •Figma’s post-IPO decline is framed as normal after a large first-day pop; still high multiple
- •Public SaaS ‘bedrock’ multiples appear elevated vs long-term averages, raising reversion risk
- •Klarna/StubHub outcomes may widen IPO discounts demanded by public buyers
- •Potential impact: founders may delay IPOs if pricing doesn’t meet expectations
- •Macro worry: if public comps reset lower, everything priced off them drifts down
- 38:58 – 42:34
EA’s record take-private and what it says about exits, leverage, and ‘AI pixie dust’ narratives
The discussion digresses into the largest LBO in history (EA), why sophisticated PE may pursue it despite low growth, and skepticism that “more AI” alone can reignite shrinking franchises. The segment highlights how attractive a clean exit can look even for stagnating assets.
- •EA take-private: scale of leverage vs business profile (hits-driven gaming risk)
- •Silver Lake’s track record cited as evidence of smart sponsorship
- •Jason’s skepticism: ‘AI’ as a growth revival story is not automatically credible
- •Investor takeaway: many would happily sell slow/negative growth assets at healthy multiples
- •Illustrates exit optionality outside IPOs when public narratives and private leverage align
- 42:34 – 57:18
Sam Altman’s $1T energy problem: data centers, ‘AI cities,’ and the limits of exponential forecasts
They tackle OpenAI’s implied compute and energy trajectory—potentially requiring more power than major nations. Jason imagines GPU ‘cities’ and Rory argues technology progress may be exponential, but economic, logistical, and adoption realities will likely slow the ramp vs today’s projections.
- •Energy demand is a proxy for compute demand and model ambition
- •Jason: single mega-buildouts can rival the power demand of major cities; ‘AI cities’ concept
- •Rory: practical constraints (power, buildout, adoption, selling) likely reduce forecasted growth rates
- •Harry: asks where a trillion dollars of capex funding can realistically come from
- •Key distinction: visionary ambition vs financial plans being taken too literally by markets
- 57:18 – 1:00:44
Monetizing ChatGPT: commerce integrations vs the reality that OpenAI must think in billions
The hosts debate OpenAI’s move toward shopping and commerce inside chat interfaces. Rory views commerce/ads as inevitable monetization paths, while Jason frames many of these moves as experiments unless they can generate multi-billion revenue quickly enough to matter at OpenAI’s scale.
- •Only a few monetization paths for massive free-user bases: commerce and advertising
- •Skepticism that shopping behavior naturally shifts to chat, based on past social platform attempts
- •Need for ‘billions’ in revenue for any initiative to be material at OpenAI’s scale
- •Parallel to Google’s early Cloud era: new businesses start as rounding errors until they aren’t
- •Trillion-dollar capex implies an enormous revenue engine is required to cover the ‘nut’
- 1:00:44 – 1:09:04
Fivetran + dbt and the coming wave of private-to-private consolidation among unicorns
Fivetran’s reported interest in acquiring dbt becomes a lens on a broader reality: there are too many unicorns and too few IPO slots. Rory argues portfolio “whipping into shape” via M&A will become common to reach IPO-scale critical mass, while Jason notes shared lead investors can make the mechanics easier.
- •Strategic adjacency: data ingestion + transformation can create a coherent ‘better together’ product story
- •Macro math: too many private unicorns, too few IPOs—M&A becomes a release valve
- •Consolidation can turn mid-sized companies into IPO-viable scale faster
- •Shared cap tables/lead investors simplify negotiation and alignment (in theory)
- •Primary risk: integration failure can turn dilution into owning less of a worse outcome
- 1:09:04 – 1:16:54
Does tech private equity’s playbook still work when AI accelerates product cycles and seat economics?
They examine whether tech PE’s classic model—buy stable SaaS, optimize, pay down debt—faces new AI-era risk. The key change: products used to evolve slowly, but now must move fast, and AI agents may reduce seat counts, undermining predictability in revenue and retention assumptions.
- •Old world: 2008–2023 products changed slowly; PE relied on stability + high NRR
- •New world: faster innovation cycles force continuous reinvestment and raise tech risk for PE-owned assets
- •Agents can reduce human headcount and seat-based revenue, changing unit economics for incumbents
- •Even startups face ‘PMF instability’ as model capabilities shift quickly
- •Implication: underwriting must incorporate AI displacement and accelerated competitive threat
- 1:16:54 – 1:24:57
Founders, politics, and corporate speech: balancing personal expression with fiduciary duty and brand risk
The closing segment addresses whether leaders should constrain political expression for the company’s sake. Rory argues companies should avoid culture-war positions, but struggles with the idea that CEOs lose personal speech rights; Jason shares tactical experience privately flagging risky posts and the reality that many executives accept the trade-offs.
- •Consensus trend: companies speaking less politically is often strategically wise
- •Hard question: CEO personal speech vs corporate repercussions (customers, recruiting, retention)
- •Jason: private ‘DM feedback’ can help, but most executives choose to post anyway, knowingly accepting fallout
- •Rory: only extreme business harm might justify leadership change; otherwise, personal opinion should stand
- •Harry: attention cycles move fast—many controversies fade quickly, reducing long-term impact
