When I was growing up, the people in my town talked about Goldman Sachs the way other towns talk about the local kid who made the majors. Getting in meant you had won. Nobody could say exactly what the job was, and nobody needed to.
Then I moved to the East Coast and met the people who did it. The work was presentations and schmoozing. Smart people, long hours, a lot of decks. The aura was real. The work under it was ordinary.
Then I moved to San Francisco, and we did the same thing with venture capitalists. Checks were few and far between, so whoever could write one was the most important person in any room. We idolized them. A few years later money was everywhere, and the median VC was just another person asking for a meeting.
Two different rooms, one lesson. What we worshipped wasn't the work. It was the gate.
Status is a gate
Status goes to whoever controls access to something scarce. It lasts exactly as long as that scarcity is defended. When technology, deregulation or a new competitor removes the barrier, the status drains away, usually with a lag, because people keep bowing to the gate after it opens.
Jobs whose whole value is access are the most exposed. Bankers, brokers and VCs often don't make what they sell. They stand between you and someone else's capital, markets or deals. That kind of value gets competed away.
How the gates were built
The Medici fortune came out of a bank. In 1397 Giovanni di Bicci de' Medici moved his bank to Florence. It became banker to the papacy, holding the post of depositary of the Apostolic Camera for 22 straight years from 1422. Being the pope's banker was the gate. The art, the politics and the dynasty came through it.
By the early 1900s the gate was a chair on Wall Street. To trade on the New York Stock Exchange you needed a seat, and seats were rationed. In 1929 a seat sold for a record $625,000, because rising volume had run into the cap on members. The exchange widened its own gate a crack, handing every member a quarter seat, about 275 new seats in all. Over the following decades the count was trimmed to 1,366, where it stayed from 1953 to 2006.
May Day and the $0 trade
For most of its history the NYSE also fixed what its members charged. Commissions had been fixed since 1792. Under the last fixed schedule, buying 100 shares of a $50 stock cost $71.50. Nobody could undercut it. That was the point.
On May 1, 1975, the SEC ended fixed commissions under Rule 19b-3. Wall Street called it May Day. While other brokers raised their commissions, Charles Schwab went the other way, set up as a discount broker and opened his first branch in Sacramento that September.
Then the gate fell in steps. In 1997 E*Trade charged $14.95 for an online market order. Robinhood opened to its waitlist with $0 trades in March 2015, while the big brokers were still charging. In March 2017 Schwab cut to $4.95 to match Fidelity. On October 1, 2019, Schwab announced $0 online stock trades, TD Ameritrade matched it the same day and E*Trade followed the next day.
01 / The price of a trade
A stock trade went from $71.50 to nothing
Representative commission on a retail stock trade, US dollars, nominal, 1973 to 2019.
Under the last fixed schedule, 100 shares of a $50 stock cost $71.50 to buy. E*Trade charged $14.95 online in 1997, Schwab charged $4.95 by March 2017, and $0 became standard in October 2019.
View data
| Broker and date | Commission |
|---|---|
| Fixed schedule, 1973–75 | $71.50 |
| E*Trade, 1997 | $14.95 |
| Schwab, Jan 2017 | $8.95 |
| Schwab, Mar 2017 | $4.95 |
| Schwab, Oct 2019 | $0 |
Here is the part people miss. The brokers didn't disappear. FINRA counted about 639,000 registered representatives in 2015 and about 640,000 in 2025, and more than half are now also registered as investment advisers. The jobs survived. The status went with the gate. The stockbroker used to be the person who could get you into the market. Now the market is an app, and the broker had to become an adviser to stay useful.
Venture after capital stopped being scarce
Venture never got a May Day. Its gate wore down over two decades. Active US VC firms more than doubled in five years, from 1,328 in 2019 to more than 3,000 in 2024. Founders got new ways around the gate too: Y Combinator's SAFE in 2013 and AngelList's rolling funds in 2020.
When the check stopped being scarce, the person who writes it stopped being special. Then money got tight again, and it didn't come back evenly. In 2025 total fundraising fell to a nine-year low and the count of firms fell for the first time. The top 10 funds took 32.9% of all the capital US venture funds raised, up from 13% in 2021.
02 / More gatekeepers
US venture firms more than doubled in five years
Active US venture capital firms, 2019 to 2025.
Active US VC firms grew from 1,328 in 2019 to 3,054 in 2024, then fell to 2,984 in 2025, the first decline on record.
View data
| Year | Firms |
|---|---|
| 2019 | 1,328 |
| 2024 | 3,054 |
| 2025 | 2,984 |
Scarcity came back, but only at the top. The handful of firms whose name opens doors for founders kept their gate, and in a tighter year the money went to them. The median VC didn't get the old aura back. Status returned exactly where the scarcity did.
Ferrari keeps the gate, Porsche loses it
Luxury runs on heritage. Heritage is a gate too, and it only holds if you defend it.
Ferrari defends it on purpose. Its stated policy is to deliver "one car less than the market demands." It shipped 13,752 cars in 2024 and 13,640 in 2025, with an EBIT margin of 29.5%. Volume held flat by choice.
_in_2018.01.jpg){kind=link}
Porsche leaned on its badge in China, and the slide started early. It delivered 95,671 cars there in 2021, 93,286 in 2022, 79,283 in 2023 and 56,887 in 2024, and 41,938 in 2025. Porsche blamed a tough luxury market and stiff competition in fully electric models in China. It now plans to cut its China dealer network from 144 outlets to about 100 by 2027.
The weak luxury market is real. The competition is the story. Chinese brands stopped selling only cheap cars. BYD put its Yangwang U8 on sale in September 2023 at RMB 1.098 million. Xiaomi is the sharpest example. It launched the SU7 in March 2024 at RMB 215,900, pitched it straight at the Taycan and delivered 136,854 of them that year, more than double what Porsche sold in all of China. Then it went after the top of the range. The SU7 Ultra lapped the Nürburgring in 7:04.957, beating the Taycan Turbo GT's 7:07.55, and sells for about RMB 529,000 against roughly RMB 1,998,000 for the Porsche in China. A lap record is marketing. So was the badge, and now it has competition.
The badge was a gate. Once Chinese brands started building faster electric cars next to it for a fraction of the price, the badge stopped carrying the volume.
Where the gates still hold
The obvious objection: doctors, Ferrari and Goldman partners still have enormous status after decades of competition. Maybe prestige is cultural, not economic.
The test is whether you can name the gate before the status moves. Doctors have licensing, and since the 1997 Balanced Budget Act capped Medicare-supported residency slots at each teaching hospital at 1996 levels, Congress has added slots only in 2021 and 2023. Ferrari has its production cap. Both gates are written down, and both still hold.
Goldman's gates are advice and underwriting, and they tell the story in miniature. Advice leaked first. Independent boutiques like Evercore and Moelis took 9% of US M&A advisory revenue in 2006 and nearly a fifth by 2013. Underwriting held longer. By the late 1990s, more than 90% of US IPOs raising about $20 million to $80 million paid a spread of exactly 7%. Nobody competed on price, which is what a gate looks like. Then even underwriting cracked. In 2018 Spotify listed on the NYSE with no underwriters at all, though it still paid three banks to advise.
Where a gate is written into law or rationed on purpose, status holds. Where it leaks, as with brokers and Goldman's advice desk, the money moves to whoever builds the next one. Status doesn't vanish evenly. It concentrates at the gates that remain.
The next gate to open
My forecast: the next gate to open is AI training data, and the data companies and founders everyone in tech envies right now lose their aura with it.
Right now it's one of the most envied gates in tech. Meta paid $14.3 billion for 49% of Scale AI, valuing a data-labeling company at $29 billion, and brought Scale's CEO in to run its superintelligence team. Part of that price bought a person. News Corp's deal with OpenAI is reportedly worth more than $250 million over five years, and Reddit's with Google about $60 million a year. Labs pay that because human-made data is scarce and they need more of it.
The best case against me is those deals themselves. A newspaper's archive or a forum's conversations can't be copied, and that kind of data keeps its gate. The generic kind doesn't, and the labs are already building around it. Microsoft's Phi-4 technical report says "synthetic data constitutes the bulk of the training data for phi-4", with web, book and code data used mainly as seeds. Once models can learn mostly from data other models made, anyone selling ordinary human data is guarding a door with no wall around it. If I'm right, by 2029 generic labeled data is a commodity, and the companies selling it are valued like contractors rather than gatekeepers. If they're still raising at higher prices then, I'm wrong.
Make the thing, and make it scarce
Every generation picks jobs to idolize, and a lot of them stand between people and something they want. A broker on the phone. A $71.50 trade. A check in a capital drought. Those jobs feel permanent while the gate holds. Then someone ships an app, or a regulator signs a rule, or a phone company builds a faster car, and the gate swings open.
Status from a gate lasts only as long as you can keep it narrow. Status from making the thing lasts only as long as the thing is better. Porsche learned that the badge isn't the car. Ferrari makes the car and rations it. Doctors do the work and guard the door.
If you want status that lasts, don't just guard a gate. Make the thing people are trying to get to, and make it scarce.