OpenAI was raising funds for a new venture focused on enterprise deployment, with $4 billion raised from 19 investors against a $10 billion valuation, when a competitor announced an equivalent enterprise AI services venture hours later.
The proximity of the announcements suggests both identified the same bottleneck, and it isn't technical. Selling models through an API reaches people who already know what they want to build; large companies need someone to come in, understand the process and make it work inside it.
It's the same diagnosis that would appear weeks later in the contest for forward-deployed engineers, named as the sector's most sought-after hire.
Creating a separate venture rather than an internal team has logic in scale and risk. Deployment services are a lower-margin, slower-growth business than selling access, and mixing the two in one structure usually worsens both sets of numbers.
For corporate customers, another sales channel changes little. What changes is the expectation: if the manufacturers themselves admit the product doesn't deploy itself, project costs become part of the budget from the start.