AI Companies' Business Models Under Scrutiny: Services Hidden Behind Software
Critics are questioning the true business models of large AI firms. A model where "autonomous" systems require significant human labor is facing growing criticism.
A debate and criticism have emerged within Silicon Valley's corporate landscape regarding the business models of certain AI companies. Behind many so-called "autonomous" or "agentic" AI solutions lies a significant amount of human labor, challenging their software-based valuation.
According to critics, many AI companies that have raised substantial funding rounds are selling customers "AI platforms" that do not operate independently. Instead, these companies are compelled to hire a large number of engineers – known as "forward deployed engineers" (FDE) – to embed within customer offices to customize and maintain the systems. These engineers spend months manually building software workflows, despite product marketing promising autonomous functionality.
This business model raises questions about the valuation principles for software companies. When customer integration and ongoing support necessitate significant and costly personnel, the scalability and profitability of the business as a software company are called into question. This "people-market fit," rather than true "product-market fit," can lead to a situation where the company's growth is directly tied to the hiring of expensive staff, thereby limiting margins and slowing down scaling.
Based on industry publications and job postings, the demand for "forward deployed engineer" roles has surged dramatically, increasing by as much as over 1,000% year over year. Salary levels in these roles can reach hundreds of thousands of dollars, further increasing companies' costs and diminishing the high gross margins typical of software businesses.
The core question is whether the delivery of customer benefits can occur without the company's personnel. If the answer is "no," it is not a genuine software-based solution but rather a service business disguised as software. This model may function in the short term, driven by high growth and financial market momentum, but in the long term, it could lead to unexpected costs and challenges as investors and customers begin to demand a more detailed explanation of the actual costs and benefits.