新智元报道
去华尔街、大厂还是去车库，这代美国大学生做出了自己的选择。
去SpaceX，或者Google，拿一份安稳、体面、年薪六位数的实习offer；还是一头扎进旧金山拥挤的「黑客屋」（Hacker House），和一群同样疯狂的年轻人挤上下铺，前途未卜。
如果是你，你会怎么选？
普林斯顿大学的高材生查尔斯·穆尔伯格（Charles Muehlberger）毫不犹豫地撕掉了第一张门票，跳上了独木舟。
仅仅四周后，他不仅创办了自己的AI初创公司，还一路杀到了巴塞罗那，向潜在客户推销他那个极具野心的构想——将开源AI模型离线部署到本地设备上。
这种被时代在后面疯狂追赶的紧迫感，让他甚至做出了一个违背传统精英教条的决定：休学一年。
「现在就动手建设的人，才能在未来的图景中拥有话语权。」查尔斯说。
这绝不仅是一个关于「大学生叛逆创业」的老套故事。
穆尔伯格不是个例。
今年夏天的旧金山，出现了一种反常的景象：常春藤联盟、斯坦福、MIT的学生，成批地拒掉大型科技公司和华尔街投行的实习offer，涌入一种叫「黑客之家」的地方。
这些学生里，许多人甚至不确定夏天结束后还要不要回学校。
几十年来，精英大学生的路径清晰得像一条高速公路。
拿到顶尖公司的实习，毕业后转正，在职场阶梯上一路往上爬。金融、咨询、大厂，三条赛道，条条通罗马。
这条路径有到它成了身份标签。
「我在高盛实习」「我在谷歌实习」，这些句子在名校社交圈的含金量，有时候比GPA还高。哈佛和沃顿的校友晚宴上，自我介绍的第一句话永远是在哪家机构。
名校加名企，这套组合拳打了几十年，从来没失过手。
但AI正在让这条曾经稳如磐石的高速公路变成一条没有目的地的路。
最直接的冲击来自软件工程。
曾经，CS专业是理工科学生的避风港。学好了进大厂，年薪六位数美元起步，稳得不能再稳。
硅谷的码农文化甚至催生了一整套生活方式：湾区的公寓、特斯拉、周末去Napa Valley喝酒。这套生活方式的入口就是一份大厂offer。
现在完全变了。Cursor、Copilot、Claude Code一轮接一轮迭代，一个人加上AI就能干以前三个初级工程师的活。
所以穆尔伯格们做了一个看起来疯狂的选择：与其在大厂里做一颗随时可能被替代的螺丝钉，不如自己去造机器。
旧金山的这些黑客之家和孵化器，遍布Mission District和SoMa区。
门槛出奇地低，配置出奇地全：免费住宿，导师指导，创业者社交网络，全套配齐。
有些孵化器甚至提供小额启动资金和法务支持，恨不得把创业的所有后勤障碍一键清零。
唯一的代价：全身心投入一个AI创业项目。
没有薪水。没有大厂title。没有写在LinkedIn上好看的品牌背书。
有的只是一张床、一台电脑、和一群跟你一样焦虑又兴奋的同龄人。
跟大学宿舍最大的区别是：没有人是被分配来的。所有人都主动放弃了更舒服的选择，跑来挤上下铺。
凌晨两点，隔壁床的人还在调模型参数。早上七点，公共区域已经有人在白板前画架构图了。厨房里堆着吃了一半的外卖盒，没有人有时间做饭。
这里的节奏跟大厂截然不同。早上9点讨论产品方向，中午找真实用户做测试，下午写代码调模型，晚上复盘迭代，连夜改方案。
一周七天，没有周末。一个月后交出demo，两个月后见投资人。节奏快到你根本来不及怀疑自己。
「在这里待一个月学到的东西，比在阶梯教室里听一整学期课还多。」这话在黑客之家里已经成了共识。
一位创业者说，大学教的是知识体系，黑客之家教的是生存本能。两者的差距，就是模拟器和实弹的差距。
有些项目甚至得到了顶尖大学的官方支持。
普林斯顿AI Lab搞了创业者路演会，给学生和投资人搭桥。斯坦福的创业学期项目允许学生全职创业一个学期而不用退学。
大学也意识到了一件尴尬的事：如果不主动拥抱这股潮流，学生会用脚投票。直接退学。
去年已经有先例了。两个21岁的常春藤辍学生在湾区创办了一家AI公司，几个月内就拿到了数百万美元的融资。
这种故事在校园里传得飞快，比任何招生宣讲都有说服力。每一个成功案例都在削弱大学的核心叙事：「你需要先念完书，才能去闯世界。」
话说回来，风险同样是真的。
绝大多数AI创业会在12个月内死掉。
没有大厂实习经历的简历，一旦创业失败，想回到传统轨道会更难。休学一年的机会成本，不是一句「学到了很多」就能抵消的。
家长那边更不好交代——你告诉他们你要放弃谷歌的实习去住上下铺，换来的多半是一阵沉默，然后是一通视频电话。
更残酷的地方在于：黑客之家里挤满了同样聪明、同样拼命的人。
你以为自己是最有想法的那个？隔壁床位的斯坦福PhD可能已经做出了比你更好的原型。
竞争烈度不比任何大厂面试低，甚至更高。因为在这里，没有HR帮你筛选对手，没有结构化面试保护你。
市场就是面试官，产品就是答卷。
一位孵化器的导师私下说了一句大实话：「这里100个人里，可能只有3个能真正做出东西。但那3个人的成就，会让另外97个人觉得自己也应该来。」
3%的成功率。跟早期风投的命中率差不多。
每个人都觉得自己会是那3%。
回到穆尔伯格。
他赌的是一件事：在AI的窗口期里，速度比学历值钱。
这场精英学生的集体出逃，最耐人寻味的地方在于它的方向。
不是逃离竞争——旧金山的竞争烈度不比华尔街低。不是逃离压力——黑客之家的工作强度不比投行低。
他们逃离的是确定性。
穆尔伯格们的选择，说到底是用一种不确定性去对冲另一种不确定性。这笔账算不算得过来，没人知道。
但有一件事已经很清楚了：精英人才的流向，正在发生结构性的转移。从大厂到创业公司，从稳定薪水到股权赌注，从职场阶梯到创业荒野。
熟悉科技史的人会立刻想起2007年。iPhone发布后，一批最聪明的年轻人涌入移动互联网创业。那一次，赢的人定义了下一个十年。输的人简历上多了一段「联合创始人」。
这一次，AI的窗口可能更短，赌注更大。
参考资料：
编辑：大卫
不喜欢
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
New Ji-won.
To Wall Street, the factory or the garage, this generation of American university students made their choice.
Go to SpaceX, or Google, and get a steady, decent six-digit trainee offer; stick one head in a crowded "Hacker House" in San Francisco, and squeeze up and down with a group of just as crazy young people.
What would you do if it were you?
Charles Muehlberger, a leading student at Princeton University, did not hesitate to tear the first ticket and jump into a canoe.
Only four weeks later, he not only started his own AI pioneer company, but also went all the way to Barcelona to promote his ambitious idea of deploying the open-source AI model offline to local equipment.
This sense of urgency, driven by the past, has even led him to take a decision that is contrary to the dogma of the traditional elite: a year off.
"He who builds now can speak in the future. Charles says.
It's not just an old story about "college students rebelling against entrepreneurship."
Murberg is not an example.
In San Francisco this summer, there was an anomaly: students of the Ivy League, Stanford, MIT, who systematically rejected interns from large technology companies and Wall Street, pouring into a place called "Hackhouse."
Many of these students are not even sure if they want to go back to school after summer.
For decades, the path of elite college students has been clear as a highway.
Got an internship at the top company, graduated and went up the stairs. Finance, consultancy, big factory, three tracks, one to Rome.
This path has become an identity tag.
"I Intern in Goldman Sachs," "I Intern in Google," and these sentences are sometimes higher than the GPA in famous school social circles. At Harvard and Walton's alumni dinner, the first sentence of my introduction was always in the institution.
This combination of punches has been in place for decades and never missed.
But AI is turning this once-stable highway into a road without destination.
The most direct impact comes from software engineering.
The CS profession was once a safe haven for students in science and technology. Learning to get into a big factory, starting with a six-digit salary per year, could not be more stable.
Silicon Valley ' s com-farming culture has even given rise to a whole range of lifestyles: apartments in the Bay Area, Tesla, and weekends of drinking at Napa Valley. The entrance to this lifestyle is a big factory offer.
It's completely different now. Cursor, Copilot, Claude Code, one person and AI can do the work of the three former junior engineers.
So the Murbergs made a seemingly crazy choice: instead of making a screw in a big factory that could be replaced at any time, they built their own machines.
These hack houses and incubators in San Francisco are spread throughout Mission District and Soma.
The threshold is oddly low and it is uniquely configured: free accommodation, mentoring, social networks of entrepreneurs, full set.
Some incubators even provide small start-up finance and legal support and hate to leave all the logistical obstacles to starting a business intact.
The only cost: a full commitment to an AI entrepreneurship project.
No salary. No big factory. No good brand endorsement on LinkedIn.
There are only one bed, one computer, and a group of people of your age who are as anxious and excited as you are.
The biggest difference with college dormitories is that no one is assigned. Everyone has voluntarily given up more comfortable options to squeeze up and down.
At 2:00 a.m., the people in the next bed were still adjusting the model parameters. At 7:00 a.m., someone in the public area has drawn the drawings before the whiteboard. Half the delivery boxes were stacked in the kitchen, and no one had time to cook.
The rhythm here is very different from the big factory. The direction of the product is discussed at 9 a.m., the actual users are tested at noon, the code-to-code model is written in the afternoon, the night is repeated and the night is changed.
Seven days a week, no weekends. Demo was handed over one month later and the investors met two months later. The rhythm is fast enough for you to doubt yourself.
"There's more to be learned here for a month than to listen to a whole semester in a stairwell. This has become a consensus in the hacker's home.
One entrepreneur said that universities taught knowledge systems and hacker houses taught survival instincts. The difference is between simulators and live ammunition.
Some projects have even received official support from top universities.
Princeton AI Lab organized a road show for entrepreneurs to build bridges between students and investors. Stanford's entrepreneurship programme allows students to start a full-time business without dropping out of school.
The university is also aware of the embarrassing fact that if it does not embrace the current, students will vote with their feet. Direct drop-out.
There was a precedent last year. Two 21-year-old Ivy dropped out of school and founded an AI company in the Bay region, which received millions of dollars in financing in a few months.
Such stories are spread more quickly on campus than any admissions campaign. Every success story weakens the core narrative of the university: "You need to finish your studies before you can go into the world."
By the way, the risk is also real.
The vast majority of AI entrepreneurs die within 12 months.
Without a curriculum vitae of the experience of a major firm, it would be more difficult to get back on track once the business fails. The opportunity cost of taking a year off of school is not offset by the phrase "learning a lot".
Parents are harder to explain -- you tell them you're going to drop Google's internship and stay in the top and bottom bunks, and you're going to be quiet for a while, and then a video call.
The harsher part is that hackers' homes are packed with the same smart, same desperate people.
You think you're the one with the most ideas? Stanford PhD in the next bed may have made a better prototype than you.
The competition is no less intense than, or even higher than, any major plant interview. Because there's no HR to screen your opponent, no structured interview to protect you.
The market is an interviewer and the product is a questionnaire.
One of the incubator's mentors told the truth in private: "There may be only three out of 100 people who can actually make something." But the achievements of the three will make the other 97 feel that they should come.”
3% of success rate. It's like the early wind hits.
Everyone thinks they'll be 3%.
Back to Murberg.
He's betting on one thing: in the AI window, speed is worth more than education.
The most interesting part of this collective flight of elite students is its direction.
Not fleeing competition — San Francisco is no less intense than Wall Street. Not to escape pressure - the work of the hacker house is no less intense than that of the investment bank.
They fled with certainty.
The Murbergs chose to use one kind of uncertainty against another. Does it count? Nobody knows.
But one thing is clear: there is a structural shift in the flow of elites. From large factories to start-ups, from stable salaries to equity bets, from the career ladder to the entrepreneurship wilderness.
People who are familiar with the history of technology will immediately remember 2007. After iPhonepublish, the smartest group of young people came to move to Internet business. That time, the winner defined the next decade. The lost people have added a "co-founder" to their resume.
This time, AI's window might be shorter and bet more.
References:
Editor: David
I don't like it.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.