Anyone see the news out about OpenAI hiring ex investment bankers to help train their models to do analyst grunt work? Curious what u guys think this means for the industry, especially with younger talent looking at these jobs
| +326 | Which Offer to Accept: UBS Analyst Role or Part Time Cashier at the Local Taco Bell | 39 | 1h |
| +249 | AI is insane | 30 | 1d |
| +202 | 2026-27 IB Energy Top Banks | 22 | 2h |
| +186 | Useless associates | 28 | 1h |
| +174 | How do you manage nerves in IB technical interviews? | 15 | 41m |
| +174 | I Snuck Into UBS Office And Took A Shit On The Floor | 20 | 44m |
| +167 | Differences in Restructuring Teams? | 31 | 5h |
| +148 | TPH V.S. Mizuho [HOU] | 22 | 4h |
| +146 | GF is an IB Analyst at a top BB, hates it, what exit options does she have? | 26 | 1h |
| +141 | Roku-Fox Deal - UBS Misses Out on yet another deals after years chasing this deal | 23 | 1h |
Career Resources
Get instant access to lessons taught by experienced private equity pros and bulge bracket investment bankers including financial statement modeling, DCF, M&A, LBO, Comps and Excel Modeling.
I would be very very surprised if they automate a structured finance model because I've tried using chatGPT during my internship and holy fucking shit was it BAAAAAD and I mean that: bad.
Typically it'll be M&A models maybe TMT or Healthcare but maybe a renewables or a challenging sector would be the real metric to base their automation efforts on.
Where was the tech 1 year ago?
Where will be in 1 year?
Where will it be in 3 years?
Where will it be in 5 years?
IMO, a lot of entry level work is very toast, including finance.
Its not going to leave entry level roles as a whole "toast." It will be a tool used for more efficiency. Demand for financial services specifically from humans will remain, and such humans will still "enter" the industry through some degree of an entry role. How it will impact working hours, comp, etc i dont know, but the idea the whole bullpen will be outsourced to ai is restarted
I find it really funny when people say stuff like this because it really just shows how ahistorical their reference point is
gpt 5 can barely spit out a coherent line of code it's going to be years before a viable product replacement rolls around
Keep throwing monkey shit at me Sam Altman
Bro is coping and hasn't heard of mosaic [ dot ] pe. Your AI analyst that you email and it sends back decks within 5 minutes.
now you got rogo doing the analyst job almost to the point of not needed analysts anymore
and before you say "its not perfect", neither are analysts. if you compare rogo to a year 1 analyst, it blows them out of the water
Genuine question - why would investment banks ever hire summer/off cycle analysts then? Why not just all MBA associates. I’m out of the industry now but I can still see my university people getting summers and FT analyst positions
I’d be very interested to see if this will cut hours for analysts as that would be nice, but I’d assume that would come with a pay cut and people would be upset over that
No. It just means we are going to be doing more and more complex analysis in a much more compressed time frame to spit out the same number that the MD had in mind the whole time.
I don't think people understand how LLMs work. It will never be good enough for the more esoteric models/companies we deal with.
they are going after banking for marketing purposes bc they are failing to close their funding round. why dont they build a gpt for accounting which is a way bigger industry?
I had a similar thought. Kind of odd to be targeting this market specifically, who exactly is it supposed be helping, the investment banks? Where is the AI that is supposed to make peoples lives easier and more efficient… what does “no more IB analysts” achieve for broader economy?
Im sorry, but I am at the point where this whole topic is just pissing me off. Everyone with the exception of AI CEOs and some newly created AI bros actually believes these LLMs will do any challenging work. They are only inflating revenues with their "AI deals" and talk dreams.
Timeline is difficult to comprehend, but once trust and accuracy is fine-tuned, I think this will especially reek havoc on the boutique and middle market firms in the next 5-10+ years.
Employment: The hierarchy may become leaner "Associate-VP-Director" teams with advanced degrees and/or have industry specialists (e.g. ex-SW engineers for Tech, Ex-lab techs for bio/life sciences) as a way for firms to differentiate. Simultaneously, firms may also have to charge lower fees and retainers, and the profession may become lower paid. I could also see law firms simply spinning up "financial advisory" offerings to clients given that lawyers build a lot of early relationships with companies on IP and other needs.
Deal Activity: Small companies hire banks largely due to not having strong internal corp dev teams to run a process. With this technology, why would a company hire a banker if they can build out a Teaser, CIM, Valuation, and buyer research with proposed terms and waterfall distributions in minutes? We may see still an execution of larger, more sophisticated deals with select firms, but a complete disintegration of middle market banking. Deal timelines will also likely go from months to weeks.
The argument that this will "free up more time" to work on other things is a pleasant facade management teams like to give. If this was true, the invention of the internet would still have mailroom workers at companies "having more free time" to work on other things.
Your first sentence is doing a lot of work for you
I'm making $150 an hour about 20-30 hours a week doing this rn
how do I sign up for this?
good, as Hazlitt would say, the bottom 20% of lost IB jobs will fuel the 20% rise in gardeners tending Altman's lawn
gotta love free markets
Maybe they are training the models
Omg - That’s so gross. Where?
How do I apply for this job 😂
We are sorta fucked. Chat cannot do anything in excel well enough, but if you ask for a bottom-up framework to analyze a company for revenue towers or like the more "thinking" parts of the modeling, it does fairly well.
But who knows, the role of a banker/investor is not just modeling anymore.
I still don’t get how AI will help with the workstreams - yes it will build the financial model, public comps table, precedent transactions with PPT output. But we already have pre-formatted 3 statement models to use and full table with CapIQ formulas to pull the data.
And yet when I was in IB we will literally handspread comps one-by-one to be “bespoke” for their clients’ discussion materials, or go line by line on FDD VDR report to see if there are any adjustment to QoE.
CapIQ already does all this so unless OpenAI can offer full flexibility with zero halluciation and 100% accuracy (or your MD will scream at you), not sure how this will help. Either way BB/EBs have their own proprietary AIs but last time I checked they were pretty bad. Yes, AI will get better with time, but IB is already semi-automated with all the tools we have.
Oh and don’t forget the cost of each tokens running AI - hopefully we can build data centers and energy grid quick enough to cover all the demand in an cost efficient manner
This is right.
For the more esoteric models with 50 adjustments to EBITDA, I don't think AI will be able to come up with shit like "pro-forma post merger seasonality adjusted EBITDA" or "debt adjusted for debt-like items and future FX movements as guided by selected brokers".
Will also be hilarious when the MD tells the AI "please increase EV to forward EV/EBITDA 20x" and AI just increases Revenue growth by 20% in the model.
Unfortunately, there'll definitely be less juniors and it'll probably be a junior running AI and a VP checking the model/output.
Finding information is easy. But discerning information is not.
Even now I can ask Gemini (google AI) to give me a valuation estimate given EBIT margin, growth rate, revenue, CapEx, and assets.
I've never been a banker but I do understand that the most important thing - either you are a sell-side or a buy-side person - that people assessing a business needs to be able to do is to connect the dots btw product experience, operational excellence, and accounting statements.
If making models is literally all junior bankers do, they are legit getting paid way too much.
If they can bring insights from people experienced at sniffing BS and finding where interesting things in accounting statement are, that would be wonderful. That's a very realistic applied research work they can do. If they launch a product around it, I'd love to try it out so I could spend a week instead of months learning.
A year on from this thread, the hiring side of it can be measured. I work on AITraining Jobs, which reads the AI training platforms' job boards every day, so here is what they show on 26 Oct 2026.
255 roles for finance professionals are open across 25 platforms, typically $60 to $150 an hour, median $80. Banking, PE and equity research are a small slice of that, and most of those roles ask for time at a bank or a fund. So it is not the bullpen being hired. It is people who already sat in it, paid by the hour to build the model or the memo and to write the rubric the AI gets graded against.
On whether the money is holding up: Mercor's median advertised rate across all its roles was $85 an hour in the first week of August and $80 last week. These are contracts that can be paused or cut at short notice, so they work as side income, not as a path.
For anyone here who did one of these projects: did the rubric work look like it would replace an analyst, or like it would make a better tool for one?
2nd year at a MM tech coverage group. Firm is very tech forward so I have Claude, Chat, and Rogo. IDK if you guys are just bad at prompting, but both Chat and Claude native in excel can do like 90% of my modeling work. Everything from a customer cube to offer analysis can be pretty much automated. I haven't updated a single file manually in months (minus a few errors in data that are not the fault of the AI). Building an operating model from scratch is difficult for it to do, but if you already have a base built out it can perfectly do everything else.
Granted this has made working with my first years super difficult - they just one shot everything in Chat, and can't explain any of the wonky formulas that is outputed. So you need to have built a decent model from scratch to properly be able to wield AI for modeling, but I have automated pretty much everything I do to a high degree
I feel like it is basically garbage in garbage out. People who are already great (modelling, ppt etc) can automate alot of workflow (especially one off requests) but junior analysts who dont know best practices or never built a model would not know what to catch etc
Earum sed aliquid velit quia. Repellendus nesciunt sint molestias repudiandae repudiandae sed repellat ex. Sunt reprehenderit asperiores sint porro laboriosam. Voluptatem aspernatur repellat eaque voluptate consectetur labore culpa eaque.
See All Comments - 100% Free
WSO depends on everyone being able to pitch in when they know something. Unlock with your email and get bonus: 6 financial modeling lessons free ($199 value)
or Unlock with your social account...
Libero ut animi facilis ullam commodi quibusdam eum. Eveniet voluptas voluptate sint. Alias temporibus voluptas aut quaerat eum autem dolore. Deleniti ea ab dicta optio occaecati.