The idea that venture is splitting into giant platforms and tiny boutiques is flawed. A16z, the largest platform, is structured as a collection of specialized, boutique-sized funds. This model proves that focused, sector-specific funds are the effective unit, even within a mega-firm.
Startups like Cursor that are built on foundation models face existential platform risk. Their supplier (e.g., Anthropic) could limit access, degrade service, or copy their product, effectively killing their business, much like the scorpion stinging the frog mid-river.
Contrary to assumptions about user stickiness, consumers of AI models will quickly switch to a better-performing or cheaper alternative. The 22% drop in ChatGPT usage after new Gemini models were released demonstrates that brand loyalty is low when model performance is the key value proposition.
With efficient discovery from accelerators like YC, the main opportunity for smaller VCs is to invest when a promising company stumbles or its re-acceleration is non-obvious. These "glitches in the matrix," where progress is non-linear, are moments where mega-funds might look away, creating an opening.
Despite its massive price tag, Anthropic's valuation is justifiable on a forward revenue multiple basis. If they achieve another year of hypergrowth, their NTM revenue multiple would be lower than public tech companies like Palantir, making the current round look inexpensive.
While AI companies with usage-based APIs like ElevenLabs can grow incredibly fast, their easy-to-implement nature is a double-edged sword. As costs scale for developers, the same simplicity that drives adoption also makes it trivial to swap them out for a cheaper alternative, creating underlying fragility.
Efficiency gains from AI will create a new normal where B2B companies target $1-2 million in revenue per employee. This is a dramatic increase from the previous SaaS benchmark and means startups will operate with significantly smaller teams, exacerbating job displacement and wealth disparity.
A large, multi-stage VC firm's growth fund serves as a risk mitigation tool. The ability to concentrate capital into late-stage winners covers losses from a higher volume of early-stage mistakes, allowing the firm to be more "promiscuous" and take more shots at Series A.
The proposed California "entrepreneur's tax" is not a one-time levy on billionaires. It's viewed as the first step toward an annual tax on paper wealth, with thresholds planned to drop to $25M. This would impact founders with illiquid equity post-Series B, forcing a mass exodus before an IPO.
OpenAI could go to zero if a macro disruption cuts off its access to massive capital infusions. With the short shelf-life of LLMs, an inability to fund the next model would render it obsolete while better-capitalized competitors like Google and Anthropic continue to innovate, causing a rapid death spiral.
For a megafund like Andreessen Horowitz's $15B vehicle to generate venture returns, it must consistently capture a significant market share—roughly 10%—of all successful outcomes. This transforms their investment strategy into a game of market share acquisition across all stages, not just picking individual winners.
The fundamental risk profile shifts dramatically between venture stages. Early-stage investors bet against business failure, an idiosyncratic risk unique to each company. Late-stage investors are primarily betting on public market multiples and macro sentiment holding up—a systematic risk affecting all late-stage assets simultaneously.