Artificial intelligence will make many services much cheaper, but not every good wants to be cheap. In economics, products whose demand rises with price are called Veblen goods; canonical examples include handmade luxury items such as Hermès bags. AI is unlikely to substantially lower the production costs of those goods, but the Veblen concept also helps explain the dynamics in high-end professional services — including the investment-banking advice sold by Wall Street.
Many parts of Wall Street’s knowledge work are vulnerable to AI: background research, data processing and preliminary modelling can be automated and offered more cheaply. Yet historically there has been surprisingly little price competition in investment banking: there was no “Great Cola War” over IPO underwriting fees because such head-to-head price cutting simply did not occur. Corporate clients do not generally buy more Goldman Sachs M&A work just because it’s cheaper.
There are several reasons for this. Partly, corporate executives often spend other people’s money (shareholders’ capital), which dulls sensitivity to direct price moves. More importantly, paying for top-tier advice is itself a signal — similar to dropping $10,000 on a handbag — and the stakes in corporate deals are much larger.
When the previous generation of rainmakers left big firms to set up boutiques in the late 2000s and early 2010s, they did not win business by undercutting incumbents. In fact, major houses such as Goldman Sachs often welcomed the arrival of Centerview-style boutiques because the newcomers charged high fees and thereby propped up pricing in the industry’s upper ranks.
Asked in April whether his firm’s fees would shrink as AI automated grunt work, Lazard’s chief executive Peter Orszag replied, “I hope not.” That response reflects a common industry instinct: white-glove, bespoke service is rarely marketed as a cut-price offering, even to cash-strapped clients — as a Bloomberg report about Lazard trimming fees to win a lucrative Venezuela assignment from a rival illustrated.
If AI makes “good-enough” advice cheap and ubiquitous, the premium on the prestige alternative may rise rather than fall. Fast fashion’s emergence made Hermès more valuable as a status signal precisely because the baseline became cheaper. Commoditization at the bottom of a market can illuminate the halo at the top.
The question for investment banks is whether they can actively manage their businesses like luxury houses — resisting the temptation to move downmarket and keeping a velvet rope around elite services. An alternative strategy is to expand the pool of companies willing to pay for the Goldman Sachs or Morgan Stanley imprimatur by offering, say, an 80% AI-generated service augmented by humans at a lower price point.
Still, few confuse Harvard Extension School courses with the full Harvard degree. I do not expect Wall Street’s elite to stoop. They will likely cut costs where AI allows, and funnel those savings largely to the bottom line rather than producing major fee reductions at the top end.
Note
Business Insider reported that Warren Buffett has long been distrustful of investment bankers, a stance visible in Berkshire Hathaway’s 2022 takeover of Alleghany.
