Data reveals an extreme power law in venture capital, with less than 1% of firms consistently achieving top-tier 3x net returns over two decades. This highlights the intense concentration of returns and the necessity for LPs to gain access to a select few managers to succeed.
Unlike traditional startups where excess capital creates bloat, frontier AI companies can convert dollars directly into compute. This immediately improves the product and reinforces the market leader's competitive advantage, fundamentally altering the scaling dynamics for venture-backed companies.
Framing AI as the next evolution of software dramatically underestimates its market size. AI's true TAM is the vast, multi-trillion-dollar labor economy it is augmenting and automating. This market is orders of magnitude larger than traditional SaaS, justifying massive valuations and investment.
A core incentive misalignment exists between VCs (GPs) and their investors (LPs). GPs are punished for errors of omission (missing a generational company), driving risk-taking. LPs are punished for errors of commission (backing a public failure), encouraging risk-aversion and conservatism.
To deploy large, late-stage checks into the best companies, firms need the information and founder relationships built by an integrated early-stage practice. Standalone growth funds struggle to compete, as it's nearly impossible to enter a hot deal cold and secure a meaningful allocation.
Merely getting 'logo' access to a top company is no longer sufficient for late-stage funds. To generate fund-returning multiples from multi-billion dollar outcomes, firms must have the conviction and ability to concentrate 5-10%+ of their fund into a single winner, a new dynamic in growth investing.
The key constraint holding back AI development is not a lack of demand or total energy capacity, but the 'speed to power.' This refers to the regulatory and transmission hurdles that delay new data centers from coming online, creating a massive supply-side investment opportunity in next-gen infrastructure.
The extreme power law in venture means a tiny fraction of firms drive nearly all returns. An LP who over-diversifies across 50-70 managers is essentially buying the index and guaranteeing average performance. To achieve top-tier returns, LPs must concentrate capital in the few proven, top-performing firms.
AI startups can show explosive growth, like reaching millions in ARR in months, before any customers have renewed. This makes traditional traction analysis difficult for VCs, who must underwrite deals at high valuations based on uncertain, potentially ephemeral, customer signals.