THESIS
//
//
//
//
NECESSARY CONDITION
Regulatory frameworks must remain permissive to innovation (avoiding the 'European' model) and open source development must remain unencumbered by downstream liability.
04:52
RISK
Steel Man Counter-Thesis
Lead Edge's consistent return profile masks a structural vulnerability to regime change. The strategy optimizes for 2-2.25x net returns through disciplined entry multiples, active selling, and loss avoidance in a specific market environment: one where software multiples mean-revert, incumbents maintain pricing power, and private market liquidity enables creative transactions. However, this entire framework could fail simultaneously under three plausible scenarios. First, if AI genuinely commoditizes enterprise software delivery, the 70% gross margin businesses Lead Edge favors become 40% margin businesses, compressing both growth rates and exit multiples. Second, if the LP network's value depreciates as AI-enabled sales and diligence tools democratize access to decision-makers and market intelligence, the sourcing and win-rate advantages disappear. Third, if the current AI investment bubble bursts violently as Green predicts, the 70% secondary and CV exposure could face severe mark-to-market losses with limited liquidity, precisely when the 800 LP base of wealthy individuals might seek redemptions. The doubles-and-triples strategy that worked in a rising-multiples environment may produce sub-1x outcomes in a prolonged multiple compression regime, particularly for a fund that explicitly avoids leverage and high-growth moonshots that could provide upside optionality. Cal Ripken's consistency was valuable in a stable game; it provides no edge when the rules change.
//
THESIS
DEFENSE
//
THESIS
DEFENSE
//
THESIS
DEFENSE
//
ASYMMETRIC SKEW
Downside appears bounded at 0.8-1.0x through loss avoidance discipline and preferred structures, but upside is explicitly capped at 2-2.5x net by design philosophy. This creates a roughly 1:2 to 1:2.5 asymmetry ratio favoring upside in normal markets, but the ratio could invert to 1:1 or worse if multiple compression persists for a decade and the secondary-heavy portfolio proves less liquid than assumed.
ALPHA
NOISE
The Consensus
The market believes that software companies face existential threat from AI disruption, that the best investment returns come from backing high-growth AI-native companies at premium valuations, and that traditional software business models are structurally impaired. The consensus view holds that AI capex spending is justified by transformational productivity gains, that frontier AI model companies represent the most attractive investment opportunities, and that late-stage growth investing requires concentration in category-defining rocket ship companies with exponential growth trajectories.
The market's logic holds that AI fundamentally changes the cost structure of software development, enabling new entrants to build competitive products at a fraction of historical cost through vibe coding and AI-assisted development. This view assumes R&D capability was the primary moat for software companies, so democratized coding ability eliminates barriers to entry. The market also believes that the unprecedented capital flowing into AI infrastructure and model companies reflects genuine demand signals and that the companies capturing this capex will generate commensurate returns.
SIGNAL
The Variant
Green believes the best risk-adjusted returns currently exist in public software names precisely because 'people hate software.' He argues incumbent software companies hold structural advantages through existing customer relationships, distribution, and implementation lock-in that AI cannot easily disrupt. He views the AI capex cycle as a bubble analogous to the telecom bubble, predicting AI models will commoditize and that hyperscalers like Google, Amazon, and Microsoft have insurmountable data and cost advantages over standalone AI model companies. His core thesis is that software competitive advantage has never been about R&D but rather distribution, sales, and customer relationships—meaning incumbents have the most to gain from AI productivity enhancements rather than being displaced by them.
Green's logic inverts the consensus causality chain. He argues software moats were never about R&D capability—Microsoft could replicate any portfolio company's product in a month with 500 engineers but simply does not care about niche markets like chamber of commerce software or manufacturing price optimization. The true moat is distribution, customer success, and implementation stickiness. Workday's 98-99% gross dollar retention proves this: enterprises that spent three to five years implementing HR software will not rebuild internally regardless of AI capabilities. Therefore AI actually reinforces incumbent advantage by making existing engineering teams exponentially more productive while preserving distribution moats. On AI capex, he applies basic unit economics: the capital required implies earnings assumptions that require power generation capacity that does not exist, while Chinese and European alternatives demonstrate the same capabilities at a fraction of the cost, suggesting model commoditization is inevitable.
SOURCE OF THE EDGE
Green's claimed edge rests on pattern recognition from direct operational experience: cold-calling approximately 10,000 companies over his career, which he asserts provides intuitive judgment about business quality that theory cannot replicate. His second claimed edge is the LP network of 800 world-class executives who provide proprietary customer diligence, backchannel references, and warm introductions that surface information unavailable through standard processes. The third claimed edge is disciplined selling—the disposition committee meets one to two times monthly specifically to evaluate exits, which he argues most venture and growth firms neglect. Assessment of credibility: The cold-calling pattern recognition claim is plausible but unfalsifiable—it is a narrative about accumulated judgment that cannot be independently verified. The LP network edge is structurally real and verifiable: having a former GM CEO call an automotive software company or a former Pfizer CEO backchannel a pharma software customer relationship represents genuinely differentiated information access. The selling discipline claim is partially verifiable through stated outcomes like Toast secondary sales at $40-50 versus current price of $30. The weakest part of the edge claim is the eight-criteria framework itself: Green explicitly states there is no correlation between eight-criteria deals and five-criteria deals in terms of performance outcomes, which undermines the claim that the framework provides predictive investment edge. This admission suggests the framework is primarily an operational filter to manage time allocation rather than a genuine alpha source. The real edge appears to be the LP network for diligence and the selling discipline for exit timing—not the sourcing criteria themselves.
//
CONVICTION DETECTED
• I'm convinced that people invest in all these AI companies, all these VCs, like have to portray the view that software is dying, is going to be dead because they have to justify how much money they're going to spend • I believe this AI capex bubble will end badly • my fundamental belief is that the models will commoditize • Our belief for right or wrong is that the competitive advantage of software company has never been about R&D • We believe that it is the incumbent's game to lose in software today • If you think they're going to like start building their own HR software, you're on your mind • a lot of the incumbents will win • I think the best risk-adjusted returns right now are in public software names • 100% [in response to whether he still cares after making money]
//
HEDGE DETECTED
• for right or wrong • I don't know if it'll be like electricity but like it'll be pretty damn close • I have no clue when this will like stop • It will probably go longer than people think • I I for one strongly believe [the qualifier 'for one' introduces personal limitation] • by the way, we could be wrong like some of our companies have been and you look like an idiot • I think investing in Open AI 800 billion is a little insane personally but like I don't know if it goes on to do like a trillion dollars of earnings, I was going to be very wrong • What am I the most fearful for what I don't know and just like AI is going to change the world and it's going to do it in ways that nobody can think about The ratio reveals genuine high conviction on core thesis elements—software moats, AI model commoditization, capex bubble—combined with intellectual honesty about timing uncertainty and the limits of his knowledge on transformational technology outcomes. This pattern suggests authentic conviction rather than performance: Green hedges on prediction timing and unknown-unknowns while maintaining categorical certainty on structural dynamics he has observed directly. The hedging on Open AI's outcome specifically demonstrates he distinguishes between his investment framework (which says 800 billion is insane) and acknowledging he could be wrong about unprecedented outcomes. This calibration increases credibility of his core convictions.