Essay: The $9 billion bet on forward deployed engineers
VC Eze Vidra argues that deployment, not model access, is now the bottleneck in enterprise AI, and that FDE teams build a moat only when their fieldwork turns into reusable product.
In a long essay on VC Cafe, investor Eze Vidra looks at why forward deployed engineering has become one of the most sought-after functions in AI. He points to roughly $9 billion committed over a few months to organizations built to help customers implement AI at AWS, Microsoft, OpenAI and Anthropic, and to job-posting data showing steep growth in FDE listings.
His main argument is that capable models are becoming abundant while getting them working inside a bank, hospital or manufacturer is still hard: messy data, legacy systems, security controls, regulation and reluctant users. FDEs close that gap, and he frames the trade-off for startups as giving up some short-term margin in exchange for a long-term moat.
The essay is also a warning. If every engagement starts from scratch, a startup is running a consultancy by accident. Vidra's playbook includes deploying small pods instead of hunting for "unicorn" hires, qualifying customers for readiness, starting from a business outcome, designing every engagement to end, and turning fieldwork into product. For FDEs weighing offers, those questions are a good way to judge whether a company's FDE team is a product engine or a services shop.
Source: www.vccafe.com