The Twenty Minute VCAre Burn Multiples BS in an AI World? & Sam Altman Needs $1TRN of Energy
CHAPTERS
Burn multiples in 2025: why AI companies can look “more efficient” while burning more
The discussion opens with a key datapoint: AI-native software companies often have dramatically worse free-cash-flow margins than non‑AI peers, yet can show better burn multiples because growth is so fast. The group frames burn multiple as an ARR-per-dollar-burn heuristic—and immediately questions how well it translates to an AI cost structure with token/compute expenses.
Why burn multiple can mislead: hidden assumptions (ARR quality, churn visibility, margin shifts, CapEx)
Rory dissects the burn multiple’s embedded assumptions and explains why the metric got popular in the uniform SaaS era but becomes noisy today. The main risk: treating the ratio as a plug-and-play truth while ignoring revenue quality, churn masking, changing gross margins, and uncounted CapEx for model-heavy businesses.
Fundraising reality check: good metrics aren’t enough in a binary market of “breakouts” vs everyone else
Harry notes founders with strong metrics still struggling to raise; Rory and Jason describe a stark “haves and have‑nots” market. The conversation explains why sub‑scale companies (e.g., ~$15M ARR) can be unattractive to VCs when the perceived option value to IPO scale is low—even if fundamentals look healthy.
“There are no non‑AI deals anymore”: the redefinition of categories and the AI narrative requirement
Jason argues the market has moved past AI vs non‑AI as a binary; most companies must present an agent/AI story or risk being ignored. The chapter frames the new default expectation: software must assume AI capability as table stakes, even if the company isn’t “AI‑native.”
Kingmakers and “raise to deter”: how prestige + capital snowball reshapes competitive funding
Harry highlights the ‘kingmaker’ effect: investors hesitate to fund second/third entrants near a perceived category leader (Harvey, Abridge, etc.). Rory and Jason add nuance—this effect is stronger in Valley-centric markets—and explain how early brand leadership matters more in AI due to buyer uncertainty and compressed decision cycles.
Peak madness or early innings? Valuations, loss ratios, and whether venture has modeled risk correctly
The group debates whether today’s extreme AI valuations are justified by a coming labor-to-software spend shift—or whether a painful reversion is coming. Jason worries about whether venture has correctly modeled loss ratios and the probability that many unicorns won’t return venture-scale outcomes even if the tech wave is real.
Public markets and IPO gravity: Figma, Klarna, StubHub and what weak post-IPO trading implies
Using Figma’s post-IPO drawdown and weaker performance in other debuts, they discuss whether public markets are turning. Rory explains the IPO buyer’s logic: if public comps exist, new issues must price at a discount to compensate for uncertainty—potentially tightening future IPO pricing and slowing the pipeline.
The mega-buyout lens: EA going private and what it signals about growth, leverage, and the “AI revival” pitch
The episode briefly pivots to the largest LBO deal, focusing on what it means to pay substantial value for a business with low growth. Jason questions whether “more AI” can realistically restart growth, while Rory notes sophisticated sponsor history and the difference between leverage norms in recurring vs hit-driven businesses.
Sam Altman’s $1T energy/compute ambition: can the world fund ‘AI cities’ of GPUs?
They analyze OpenAI’s projected compute and energy requirements, the plausibility of 125× energy scaling, and the capital stack needed for trillion-dollar infrastructure. Jason paints a vivid picture of data-center “cities,” while Rory argues technology may be feasible but economics and adoption rates are likely to slow forecasts.
Vision vs finance: “willing it into existence,” and the danger of valuing suppliers as if the vision is a purchase order
Rory distinguishes between visionary CEOs setting direction and the financial system treating that direction as guaranteed. The key warning: markets may price Nvidia/Oracle and broader equities as if the full trillion-dollar build-out is inevitable, when a more realistic outcome is slower growth that still looks amazing but needs less immediate CapEx.
Commerce inside ChatGPT: monetization experiments, ads vs buying behavior, and the ‘must be billions’ constraint
They discuss buy-in-chat experiences and protocols emerging from both OpenAI and Google. Jason argues many integrations will be PR-heavy experiments unless they reach multi‑billion-dollar scale quickly—because OpenAI’s cost base demands massive monetization, and small wins won’t move the needle.
Fivetran + dbt and the coming wave of private-to-private consolidation to reach IPO scale
Rory frames the rumored acquisition as a strategically adjacent combo that can create a clearer ‘better together’ product story and, crucially, help companies reach critical mass for IPO readiness. They argue consolidation is structurally necessary with hundreds of unicorns and a limited IPO throughput.
Does tech private equity’s playbook break in AI? Static products, seat-model pressure, and new tech risk
Harry asks whether PE is being paid for the new AI-driven displacement risk when buying mature SaaS assets. Jason and Rory argue the last 15 years were unusually stable—products barely changed—whereas AI compresses cycles, threatens seat-based pricing, and forces continuous reinvention, making both PE underwriting and VC forecasting harder.
Founders, politics, and corporate restraint: free speech vs fiduciary impact in the attention economy
The closing segment debates whether CEOs should moderate political expression for the company’s benefit. Rory supports keeping corporations out of culture wars while struggling with limiting personal speech; Jason shares real-world anecdotes about advising execs to delete posts and the frequent response: leaders knowingly accept the business cost.
