A former Goldman Sachs analyst has raised $22 million for a new AI venture, Bloomberg reported, and the name of the company barely matters next to what it represents: the smartest juniors on Wall Street would rather build the machine than keep feeding it.
Bloomberg's deals desk, in a report from reporter Manuel Baigorri on June 30, said the founder spent time as an analyst at Goldman Sachs before striking out to build an artificial intelligence company, and that the round closed at $22 million. Bloomberg has not yet named the founder, the startup, or the lead investors, and until it does, treat the specifics as developing. What is not developing is the pattern this fits into, and that pattern is now impossible to miss.
Start with Mo Achour. He spent eight years in M&A at Goldman Sachs in London, building and reviewing what he has described as thousands of financial models, before leaving to found Axe AI, a startup that promises to build institutional grade financial models in minutes instead of weeks, according to eFinancialCareers. Achour's pitch is blunt: he already knows exactly what a good model looks like because he built so many of them, so his company can skip the years most AI founders spend learning the domain from scratch. That is not a generic efficiency story. It is a specific bet that the fastest way to automate a job is to have actually done it.
OpenAI is not ceding this ground. The company has quietly built a team of more than 100 former investment bankers from Goldman Sachs, JPMorgan Chase and Morgan Stanley under an internal effort known as Project Mercury, paying contractors $150 an hour to produce financial models that train its systems, Bloomberg reported in October. Applicants go through an AI-run interview and then have to submit one Excel model a week, built to the same standard a real deal team would demand. So the same population, ex-Goldman analysts who know exactly how a leveraged buyout model should look, is being split two ways right now. Some are taking OpenAI's hourly rate to train someone else's model. Others are raising venture money to build their own.
That is not a coincidence. It is a labor market repricing itself in real time.
Gabe Stengel took a third path. He left Lazard, where he worked as a banker for two years after a Princeton computer science degree, and cofounded Rogo in 2022 with John Willett and Tumas Rackaitis. Rogo builds generative AI tools that automate financial analysis for investment bankers, and it raised an $18 million Series A from Khosla Ventures on the premise of being, in its own words, Wall Street's first AI analyst. Stengel's edge was never that he coded better than a typical engineer. It was that he had sat in the seat the software was meant to replace.
None of this happens in a market that is pricing AI seed rounds the way it priced them three years ago. Marlon Nichols of MaC Ventures told TechCrunch in March that his average entry check has grown from $1 million in 2019 to $2.5 million now, capping out around $5 million, because
Also read: Bhavin Turakhia is spending $30 million of his own money to rebuild Office from scratch • Switch chases a $19 billion valuation as private money floods into AI data centers • Anthropic's Fable 5 Pricing Mess Is Handing Chinese Rivals an Opening
