Four engines, and the boring one is winning#
M&A advice. When one company buys another, someone has to value the target, structure the deal so it doesn’t detonate on tax or regulatory grounds, and hold the client’s hand through months of negotiation. The fee is a percentage of deal size that shrinks as deals grow — 5–10% on something under $10M, roughly 0.5–1.5% on a multi-billion deal. One percent of $1B is still $10M for a single transaction.
It’s also a winner-keeps-winning business. Advise one landmark deal and you get the call for the next, which is why the same few names appear on everything.
Underwriting. Taking a company public or issuing its bonds. IPO fees run 3–7% of the amount raised; bonds are safer and more standardised, so usually under 2%. The under-appreciated detail is that the bank frequently buys the shares from the company first at an agreed price and then resells them. Good outcome, they keep the spread. Bad outcome, they’re holding stock nobody wants, with their own balance sheet absorbing the loss.
Market making. Not directional betting — the airport currency booth. Always ready to buy or sell, earning a sliver of the bid-ask spread, thousands of times a day. When a pension fund needs to offload $500M of stock without collapsing the price, a desk absorbs it. Tiny margin, colossal volume.
Asset and wealth management. The quiet giant. An annual fee of roughly 0.5–1% of assets, charged whether markets rise or fall. Goldman’s asset and wealth arm brought in $16.14 billion in 2024, up 16%, on assets under supervision of $3.14 trillion. Management fees alone cleared $10 billion.
Note the structural difference. An M&A fee is a one-off — the deal closes and you start hunting again. A wealth management fee arrives every year, from the same client, forever. It’s a subscription, and subscriptions are worth far more than transactions.
The trading myth has a date attached#
Banks mostly stopped betting their own capital on market direction after 2008, when regulators — in the US via the Volcker Rule — restricted proprietary speculation. Today, when a bank trades, it’s overwhelmingly executing for a client: a pension fund, a corporate treasury, an asset manager.
This matters for reading criticism. A lot of the anger aimed at investment banks describes either hedge funds — which genuinely do take institutional money and bet it on markets — or pre-2008 practices that the rules since dismantled. Three different businesses get blurred:
- A commercial bank takes deposits and lends: mortgages, car loans.
- A hedge fund takes wealthy and institutional money and actively bets it.
- An investment bank is a facilitator — deals, fundraising, trading flow, managing client assets.
Firms like JP Morgan run commercial and investment banking under one roof, which is why the distinction blurs in public argument. It’s still worth keeping straight, because criticism aimed at the wrong entity is easy to dismiss.
Why nothing disrupts this#
Fees this high should attract competition. They mostly haven’t, and the reasons are four things software genuinely cannot manufacture.
Relationships. A CEO hires the banker who has known the company for fifteen years and can be trusted with information that would move the share price. That’s not a search-results decision.
Distribution. Selling $2B of new stock means finding institutions willing to write enormous cheques within days. Big banks have those lines open already.
Balance sheet. Underwriting sometimes requires putting billions of your own capital at risk to absorb the deal. A startup cannot fake this.
Trust. Built one transaction at a time over decades, and destroyable in a single leak or botched listing.
The deal that shows all four at once#
June 12, 2026: SpaceX listed on the Nasdaq, selling 555.6 million shares at $135 to raise about $75 billion — the largest IPO in history by a distance. It broke Saudi Aramco’s $25.6B record from 2019 by roughly threefold, and closed its first day valued near $2.1 trillion. Demand reportedly exceeded $250 billion, around four times oversubscribed.
Ten banks shared the underwriting, led by Goldman Sachs and Morgan Stanley, with JP Morgan, Citigroup, Barclays and UBS in the syndicate.
Look at what that required. Relationships to be in the room at all. Distribution to place $75B of stock. Balance sheet to underwrite a slice each. Trust to be handed the largest listing ever attempted. And the reason it was split ten ways is that the risk was too large for anyone to carry alone — which is itself the moat, visible.
The aftermath is a useful corrective on IPO pricing, incidentally. The stock peaked at $225.64 intraday on June 16, then fell for three straight sessions and was trading near $153 by late June. Above the offer price, well below the frenzy.
So what to actually do#
- Reset the mental model — the money is in advice, access, and recurring fees, not in traders shouting.
- Notice which business model you’re looking at when you evaluate any financial firm: one-off transactions or recurring fees? The second is worth far more.
- Check which institution a criticism actually applies to before accepting it. Bank, hedge fund, and investment bank are not synonyms.
- When you pay 1% a year on assets, understand you’re on the other side of the single best business in finance.
Sources & further reading#
- Wealth and asset management revenues rose at Goldman Sachs in 2024 — WealthBriefing — the $16.14B segment revenue, $3.14 trillion in assets under supervision, and the $53.5B firm total.
- Goldman Sachs 2024 Annual Report — the primary source for the segment breakdown.
- SpaceX IPO: the largest in history — CNBC — the June 12, 2026 listing, $135 share price and $75B raise.
- SpaceX IPO raises $75B in largest public market debut in history — Fox Business — the comparison against Saudi Aramco’s 2019 record.
- SpaceX IPO 2026: the largest IPO in history — Intellectia — share count, oversubscription, and post-listing price action.
