FirstMark Partner Matt Turck: Software is Moving Towards 'Headlessness' Rendering Due Diligence Ineffective
Matt Turck, partner at FirstMark Capital, pointed out that while venture capital should strengthen due diligence, the form of software has evolved from 'serverless' to 'headless', reducing the visible structure and reviewable objects of enterprise products, making traditional due diligence methods increasingly ineffective.
This judgment aligns with the current trend in AI applications: more and more products are built on APIs, model calls, and third-party infrastructure, shifting core value from 'self-built systems' to 'combinatorial capabilities'; ecosystems including Stripe, Vercel, and OpenAI API allow startups to launch products with almost no proprietary infrastructure in the early stages.
Several English investment and technology communities (such as discussions among a16z and Y Combinator partners) have also noted that in the AI era, startups are seeing a decrease in 'code proportion' and an increase in 'dependency', making it difficult for investors to assess moats through traditional metrics like code quality and technical barriers.
Source: Public Information
ABAB AI Insight
The essence of 'headlessness' is the migration of software value from 'internal systems' to 'system combinations'. Previously, software companies built moats through code, architecture, and deployment capabilities, but under the serverless and API economy, these capabilities have been abstracted and commoditized by platforms, with startups increasingly stitching together existing modules. This directly weakens the effectiveness of technical due diligence, as the reviewable 'internal complexity' is disappearing.
This will change the pricing logic of venture capital. Traditional VCs rely on assessments of technical depth and team engineering capabilities to evaluate long-term competitiveness, but as products increasingly depend on external models and infrastructure, moats shift towards distribution, data, brand, and user relationships. This means that due diligence focus shifts from 'code quality' to 'growth structure' and 'user lock-in mechanisms'.
Furthermore, this structural change is compressing the barriers to entry for startups while increasing competitive density. As building costs decrease and supply expands rapidly, the market is more likely to see homogenized products, making it difficult to maintain a leading advantage through technical accumulation, which now needs to be amplified through scale, channels, or capital. This is also one of the important reasons why the current AI application layer is 'exploding but hard to profit'.
From a longer-term perspective, 'headlessness' is a continuation of the industrialization of software. The infrastructure layer continues to concentrate (cloud vendors, model providers), while the application layer tends to be fragmented and rapidly iterated. The distribution of power and profits in the industry chain will further concentrate towards the underlying platforms, while upper-layer applications need to seek differentiation in high dependency, which will long-term reshape VC's investment methodology and return structure.