在华尔街取得成功的关键在于：a) 赚钱，以及 b) 以不同于他人的方式赚钱。业绩达标只是入场券，超越同行才能赢得荣耀。
The keys to succeeding on Wall Street are to a) make money and b) make money differently than others. Performance is table stakes; outperformance gets you the glory.
人工智能正在颠覆这一逻辑。英伟达价值5000亿美元的联盟囊括了全球四大另类资产管理公司中的全部四家，以及前八名中的六家。黑石集团和阿波罗全球管理公司正联合向Anthropic提供数十亿美元资金，用于购买算力。Anthropic旗下专注于现实世界应用的AI咨询业务部门，从黑石集团、海曼弗里德曼、高盛、泛大西洋投资、伦纳德·格林和阿波罗全球管理公司手中筹集了15亿美元。
AI is messing with that logic. Nvidia’s $500 billion consortium includes the four largest alternative-asset managers in the world and six of the top eight. Blackstone and Apollo are jointly providing billions of dollars for Anthropic to buy compute. Anthropic’s AI-for-the-real-world consultancy offshoot raised $1.5 billion from Blackstone, Hellman & Friedman, Goldman Sachs, General Atlantic, Leonard Green, and Apollo.
几乎每一家未列入上述名单的大型私募股权公司，都在支持OpenAI的竞争性方案。（高盛一如既往地两边下注。）这里有一张便捷的图表。
Nearly every big private-equity firm not on that list is backing OpenAI’s competing version. (Goldman, being Goldman, is in both.) Here is a handy chart.
这并非华尔街的常规操作模式，那里的目标是从竞争对手手中抢夺投资机会，而非共享。上一次私募股权行业联手收购那些单家机构无法独立承担的大型交易时，结果却是历史上最糟糕的收购案例之一。
This is not how things usually work on Wall Street, where the goal is to snake investments away from rivals, not share them. The last time private equity clubbed up on deals too big for any one of them to handle solo, we got some of the worst buyouts in history.
部分拥挤现象可以用所需资金的巨大规模来解释。“为什么你能让五到六位顶级玩家坐在同一张桌子上？因为蛋糕足够大，人人有份。”布鲁克菲尔德资产管理公司首席执行官康纳·特斯凯上周对我说道。他认为，随着投资者在人工智能领域各自占据一隅，该领域仍将出现赢家与输家；布鲁克菲尔德押注于欧洲的主权人工智能，蓝猫头鹰资本是人工智能领域的房东，阿波罗则是愿意承担芯片风险的贷款方。
Some of the crowding can be explained by the sheer amount of money needed. “Why can you get five or six of the leading players at one table? Because there’s enough to go around,” Brookfield Asset Management CEO Connor Teskey told me last week. He thinks there will still be winners and losers in the AI trade as investors stake out corners within it; Brookfield has made bets on sovereign AI in Europe, Blue Owl is an AI landlord, Apollo is a lender willing to take on chip risk.
但在我看来，这些究竟是不同的赌注，还是对同一巨大赌注的营销包装，尚不明确。人工智能开始看起来像是一笔“IBGYBG”交易，这是华尔街的一个术语，曾出现在美国政府关于2008年次贷危机的事后调查报告中：“我会走，你也会走。”如果它成功了，即使没有人显得像脱颖而出的天才，每个人都会赢。如果它失败了，每个人都会输，但也没有人显得像脱颖而出的傻瓜。
But whether those are different bets or marketing gloss for the same big bet is less clear to me. AI is starting to look like an “IBGYBG” trade, a Wall Street phrase that made it into the US government’s postmortem report on the 2008 subprime crash: “I’ll be gone, you’ll be gone.”Everyone wins if it works, even if nobody looks like a standout genius. Everyone loses if it doesn’t, but nobody looks like a standout moron.
讽刺的是，就在投资界刚刚回归到一个似乎又能实现超额收益的市场时，人工智能却横空出世。二十年的“轻松赚钱”时代，让每个人都看起来像是毫无差别的天才。一个更加分裂、充满风险且动态变化的世界——战争、通胀和关税并存的世界——或许本会鼓励投资者在他人随大流时另辟蹊径。“我认为，我们开始会看到群体思维的吸引力有所减弱，而逆向思维也将逐渐显现成效，”我在2024年初写道。哎呀，打脸了。
The irony is that the investing world wasjust returning to a place where outperformance seemed possible again when AI arrived. Two decades of easy money had made everyone look like undifferentiated geniuses. A more fractured, risky, and dynamic world of wars, inflation, and tariffs might have encouraged investors to zig where others zagged. “I think we’ll start to see the pull of groupthink become a little weaker, and contrarianism start to pay off,” I wrote in early 2024. Oops.
值得注意的是，普华永道最近的一项研究估计，仅全球数据中心领域的支出到2050年就可能突破30万亿美元。该公司表示，这一建设规模将超越铁路、互联网和电气化等以往的基础设施扩张浪潮。
Notable A recent PwC study estimates global spending on data centers alone could top $30 trillion through 2050, a buildout the firm says would surpass infrastructure booms such as railroads, the internet and electrification.