Discover your dream Career
For Recruiters

Morning Coffee: HSBC’s top bankers have fewer places to hide. Misery of the junior bankers who are stuck in the past

“Co-heads will roll”, as the saying goes when there’s trouble in the financial world.  There are many reasons why investment banks love to have more than one person in charge of the same business line, some of them good and some less so.  But according to HSBC CEO Georges Elhedery, there is one thing that all co-head structures have in common – each of the co-heads has someone else to blame if targets aren’t met.

Since taking over the top job, Elhedery has made it his mission to reduce management layers in general, but to be particularly harsh on shared ownership of important jobs.  He now claims that “about 60% of our revenue is generated under single accountability”.  He also says that previously, there was “zero percent single accountability”, and that some business lines had not just double, but “multiple” leadership.

That might be slightly exaggerated.  HSBC certainly did have more co-heads of things than many of its competitors, but this was at least partly because of its unique geographical structure.  And within that structure, the geographic business units were often very clear indeed; if there was one head in Hong Kong and one in London, there might have been at least as many problems arising from a lack of shared ownership of the global P&L as from not enough individual accountability.

And there’s more to accountability than an org chart, in any case.  Big banks – particularly ones like HSBC which have cross-selling between corporate lending and ancillary services as part of their business model – really do have to be managed across internal reporting lines, and they tend to do quite badly when they get divided up into silos with divisional heads ruthlessly guarding their own turf. 

That might be why, even after a radical restructuring aimed at “ruthlessly killing the complexity”, there is still 40% of the group’s revenue that has more than one owner.  To get it any lower than that would be dangerous. After all, the whole point of a bank the size of HSBC existing is that it’s meant to be more than the sum of its parts.

In any case, Elhedery is now rolling out AI tools to all 200,000 workers, saying that “utilization is not optional” and that “I will be monitoring usage, I’ll be monitoring super-usage.  Those will be the most influencing among their colleagues to drive the utilisation”. And, of course, once a decision has been delegated to the computer, it’s much more difficult to hold an individual human being accountable for it.  Ironically, after getting rid of the co-heads, HSBC might be on the way to giving every single manager the perfect partner, one that can’t argue back.

Elsewhere, boutique investment banks are apparently really optimistic about being able to sell new AI-driven services to their clients. Navid Mahmoodzadegan, the CEO of Moelis & Co. has told a conference that “we have a project going on working out how to integrate a lot of this data that we’ve collected over the last 18 years and making sure that our data is usable and can plug into these AI tools”.  For Houlihan Lokey, Scott Adelson says that already, “having statistically significant datasets really allows us to analyse things differently, and on a basis where we can have an information set that others don’t”.

All great stuff, no doubt. But passive-voice constructions like “we have a project going on” and “we can have an information set” rather invite the question – who is actually doing this work?

Because it actually sounds like it might be a nightmare job.  M&A transactions are not like bond trades, there is no standardized reporting format.  Each deal is bespoke, and the differences can be very important indeed.  Agreeing how the data is going to be compiled is a project in itself, and the specifications for inputting it are likely to be very demanding indeed.

When you put it that way, it’s obvious who is going to be doing it – junior bankers, doing a “pls fix” on the entire history of their firm.  Who else can you rely upon to handle reams of minute detail and boring repetitive work?  This might be the only banking job in the world where the people doing it wish they were checking fonts in a PowerPoint deck.

Meanwhile …

Lazard’s Peter Orszag is taking the opposite view from Georges Elhedery; he’s just appointed Raymond McGuire and Tim Donahue to be co-heads of North American financial advisory. There’s also going to be “a new group of senior bankers” immediately below Mark McMaster, the global head of M&A, charged with “more complex deals”. (Financial News)

JPMorgan have saved a headhunter’s fee on the appointment of Todd Combs to run their Security and Resiliency Investment Fund.  Jamie Dimon apparently personally caught up with him at a company event and immediately said “we’re all in”, the moment he expressed any interest in the job. (FT)

Senator Elizabeth Warren has demanded that the US Treasury publish any details of discussions they might have had with UBS over its possible relocation to the USA. (Bloomberg)

For anyone who might be thinking about some major lifestyle changes in the New Year, a somewhat cautionary long read about being a stay-at-home dad in New York.  It’s not an easy life. (GQ)

“Many people are quite excited about next year”, according to one senior trader in Europe.  It looks like it isn’t just M&A bankers who might be heading for a hiring spree. (Financial News)

There isn’t any scandalous entertainment, and the HR guidelines are presumably still pretty tight. But after a long period of austerity through the financial crisis and the pandemic, big Christmas parties are back for bankers in Australia at least.  And the budget which used to go on international pop stars now appears to be spent on some of the very best food and wine. (AFR)

If casinos are licensed in New York City, hedge fund geniuses will no longer rub shoulders with firefighters and wise guys at the city’s borderline-legal poker games. (FT)

Have a confidential story, tip, or comment you’d like to share? Contact: +44 7537 182250 (SMS, Whatsapp or voicemail). Telegram: @SarahButcher. Signal: sarahbutcher.22  Click here to fill in our anonymous form, or email editortips@efinancialcareers.com. 

Bear with us if you leave a comment at the bottom of this article: comments are moderated intermittently by human beings. Sometimes these humans might be asleep, or away from their desks, so it may take a while for your comment to appear. You must take sole responsibility for comments you post on this site. We will take reasonable steps to weed out anything that we consider to be offensive or inappropriate.

author-card-avatar
AUTHORDaniel Davies Insider Comment
  • Ro
    RonaldKunenborg
    12 December 2025
    "M&A Transactions are not like bond trades, there is no standardized reporting format. Each deal is bespoke, and the differences can be very important indeed. Agreeing how the data is going to be compiled is a project in itself, and the specifications for inputting it are likely to be very demanding indeed." As a data modeller, information analyst and enterprise data architect I've been responsible for quite a few of these projects. They're always considered easy and every data entry item is obvious, until you start asking questions like "what does this mean for your business if I read it like this?" in a room with colleagues from other business lines. At the tax authorities I had to discuss Fiscal Year. As it turns out, that isn't nearly as clear cut a concept as some people thought. And the one even *I* thought would be simple (document number) turned out to be a major pain a few months later when we got into a debate over whether it means "the file you entered into the archive" or "the message that was put into the archive in several different formats, each a separate file". That little miscommunication nearly forced a year of rework. So anyone busy with AI that ignores that detail is just doing marketing in order to say to the stock market "we're doing it too! Don't sell our shares please!" - Which is fine I suppose, but also the reason why currently no AI project is going to be profitable. Unless you do the hard work that should have been done in the 80's and 90's already (force your business architecture to be current and modelled, force your business information to be modelled for every new system and slowly do it for existing ones) and spend real money on that effort not by hiring Accenture but actual professionals who know what they're doing, AI will never have a solid base to operate from. Why are LLMs succesful? Because dictionaries for languages, and words, aren't up for debate. They're just words with a chance of appearing. The second you want actual reasoning, semantics takes over. And most people are horrible at that. So good luck with AI, dear banks. There are VERY few people qualified in this space.

Sign up to Morning Coffee!

Coffee mug

The essential daily roundup of news and analysis read by everyone from senior bankers and traders to new recruits.

Sign up to Morning Coffee!

Coffee mug

The essential daily roundup of news and analysis read by everyone from senior bankers and traders to new recruits.