a16zHow AI Is Rewriting the Power Law of Venture Capital
At a glance
WHAT IT’S REALLY ABOUT
AI turns capital into compounding compute, intensifying venture’s power law
- The conversation argues that AI is amplifying the venture power law because capital can be converted into compute that directly improves products, accelerating scale advantages and concentrating value creation.
- AI is framed as a GDP-wide platform attacking transportation, labor, services, and coordination, making its total addressable market potentially far larger than traditional software categories.
- From an allocator perspective, consistent top-quartile venture results are extremely rare (20 of ~3,000 firms with consistent 3x net outcomes), so access, selection, and concentrated sizing drive performance more than broad diversification.
- The panel describes structural changes in venture: companies stay private longer, late-stage can now produce venture-like fund-returning outcomes, and “the middle” of the VC market is being squeezed between niche seed specialists and scaled multi-stage platforms.
- They highlight emerging portfolio and market risks: noisy AI traction metrics, LP timing fears and incentive misalignment, and a looming legacy SaaS/private credit overhang where AI-resiliency and growth acceleration increasingly determine terminal value.
IDEAS WORTH REMEMBERING
5 ideasAI makes the venture power law more extreme by letting capital directly compound competitive advantage.
The speakers argue AI uniquely turns cash into compounding advantage because dollars can be converted directly into compute, which then improves models/products, unlike past startups where too much capital often created coordination drag. This increases economies of scale and concentrates outcomes in the best-capitalized and best-executing players.
AI’s TAM is better measured by labor/task value across GDP than by traditional software spend.
They cite AI reaching ~$100B in revenue in ~4 years versus SaaS taking ~15, and claim AI’s addressable market is tied to the economic value of tasks (especially labor), not software budgets. This reframes TAM from “IT spend” to “share of GDP/task value,” implying much larger potential winners.
Consistent top-tier venture performance is rare—access to repeat winners dominates outcomes.
Verdiyan notes an analysis of ~3,000 US VC firms found only 20 achieving consistent 3x net returns over two decades (defined as 3–4 funds delivering 3x net TVPI). The implication is that allocator outcomes depend heavily on access to a small set of firms and their ability to repeatedly invest in category-defining winners.
Sizing and concentration matter more than ever—especially in late-stage where fund-returning positions are now possible.
They emphasize that “logo exposure” is insufficient: early-stage requires owning enough, while late-stage requires concentrated position sizing (e.g., 5–10%+) to enable single-company fund-returning math. This concentration is presented as increasingly necessary as outcomes scale (top decile moving from ~$10B to ~$40B+ and potentially higher).
Venture is polarizing into niche early funds and scaled multi-stage platforms—“the middle” is getting squeezed.
They describe a barbell: highly specialized pre-seed/seed funds can win earlier when uncertainty is high, while platform firms spanning seed-to-public win by de-risking founders with resources and lifecycle support. The “messy middle” struggles because it lacks either extreme specialization or scaled capabilities/brand pull.
WORDS WORTH SAVING
5 quotesRight now, especially with the labs, for the first time, you know, in my career, you can take capital and throw it at a company, and it compounds their advantage.
— David George
AI is attacking every facet of the GDP, transportation, labor, services, capital, coordination. There hasn't been a technology paradigm that hits on 30 trillion in GDP at the same time.
— Aram Verdiyan
We've looked at the data of three thousand venture capital firms in the US. Only twenty have achieved consistent 3X net returns over the last two decades.
— Aram Verdiyan
If you, as an allocator, have not had access to the top five to ten companies over the last five to ten years, you're significantly behind in terms of returns.
— Aram Verdiyan
If you are a software company that is in somewhat not resilient to AI, I think you're challenged both in terms of your equity position, also credit as well.
— Aram Verdiyan
High quality AI-generated summary created from speaker-labeled transcript.
