The traditional VC model of a small, generalist partnership is an artifact of a time when the tech market was nascent. The massive expansion of technology has necessitated a shift towards larger, specialized firms with platform services to effectively deploy capital and support founders.
Traditional media has become increasingly critical of the tech industry, making it an unreliable channel for founders. Consequently, VCs must build their own direct-to-audience media platforms to shape the narrative, support portfolio companies, and establish thought leadership.
The most significant constraint in the current AI boom is the scarcity of talent with hands-on experience in training large-scale models. This has made the competition for talent far more fierce than the competition for customers, driving high salaries and 'acqui-hire' transactions.
Certain AI applications have clear, working economic models, particularly those using diffusion models to reduce the marginal cost of content creation (images, music, code) to near zero. In contrast, the economic viability of more complex, agentic automation in enterprise workflows is still being proven.
In rapidly expanding markets like AI, traditional metrics such as TAM and valuation are less important than identifying the winning team and company. The core investment philosophy is that the only unforgivable error is a sin of omission—missing the category-defining company.
Keep pulling the thread on Martin Casado.