5月份坚持了几天务必读完的一本书，最近的AI美股涨的有点无法理解了，4月份美元基金净值涨了40%+，5月份涨幅进一步逼近了60%，越来越不可思议了。当下的行情最值得的对标的就是2000那次互联网危机，所以找了类似的书来读一读，还是蛮有收获的。最意外的收获其实是一次大的泡沫不只是行业泡沫，不只是行业内企业的兴衰，而是更多其他企业的裹挟而入，和当时的社会经济环境相结合的产物。这么看来，我们今天的AI行情依然还处于早期阶段。
Disclosure was the least intrusive form of supervision—like a mother’s telling her child to keep the cookie jar in plain sight. Or as Louis D. Brandeis had explained, “Sunlight is said to be the best of disinfectants; electric light the most efficient policeman.”In other words, the requirement to disclose motivated a CEO not to do ill and generally not to violate the law, but it did not ensure that he would build value for the owners.
披露是干预最少的监督，包括透明化、信息平权的相应努力，这套机制在AI时代会更好用。
In the 80s, companies such as Oracle, Sun Microsystems, and Microsoft followed Intel’s lead. Soon,Silicon Valley was known as a place where nerdy, postadolescent programmers worked for modest wages but got rich on stock options—just as Ben Graham had imagined. Of course, as every economist since Adam Smith could tell you, nothing of value—not even lunch—is free.Stock options were not costless; it was just that their cost went unaccounted for.
80年代硅谷造富的工具居然是员工期权。所以那个时代的股权还没有今天的分配方式，今天的分配更甚一筹了。当然期权不是免费的午餐，期权发多了的稀释对股东也十分讨厌。
Nortel Networks was a disastrous example; under the pressure of analysts’ expectations, it acquired nineteen companies between 1997 and 2001 in a quest to transform itself from traditional telephony to data networking. The stock steadily climbed—until investors realized that Nortel had issued $32 billion in stock for largely worthless assets—whereupon Nortel’s stock collapsed from $80 to $2.”
北电的失败源于泡沫时期的疯狂收购。从这点上看，近期的市场大规模收购似乎还没有发生，近期唯一的系列收购硬要算的话也就是spacex和xAI之间的事情了。相信未来一定会有这么一波，一旦数据中心的收益可以被市场溢价高估，未来的收购就有可能在数据中心上再现？静待吧。唯一的问题是AIDC门槛偏高了，不像当年买光缆埋光缆那么容易。那会是什么呢？这是个好问题。
By summer 1995, Congress was awash in proposals to reduce the scope of regulators, freeze the budget of the SEC for five years, eliminate some of its commissioners, and reduce disclosure requirements.The very day Blumenthal raised the alarm, Netscape Communications,an Internet company with a sensational product—a browser for surfing the Web—but with only two quarters of sales history and no profits, went public.
政治和经济的关系也是很紧密的。95年的监管预算缩减对放松监管有直接影响，才有了后来的泡沫。这轮的危机如果来的话，也应该有个类似的时刻？比如放松对数字货币的监管？我只是再猜。
It was in September 1998, just as the currency crisis in Russia was roiling markets, that eBay, the Internet auction king, went public. The stock was offered at $18. After a full day of trading, it closed at $477. By year-end the stock was at $241. It strains credulity to report that eBay was trading at 1,800 times its previous year’s earnings, but at least eBay had earnings.
eBay这样的公司火箭式上涨，1800x P/E估值和今天的SpaceX按112x P/S的IPO估值相比也没差太多了。巨型公司上市肯定是个不好的信号，但值得注意的是泡沫破裂到了2000年之后，也就是2年以后了。所以未来2年还是要坐紧、抓好，保持一定的激进，尤其是对新东西、新公司。
The Nasdaq index and the Dow Jones average had tripled over this period to, respectively, 1,720 and 8,000.’ This steep ascent might have impressed upon investors the need for caution, but its effect was otherwise. If General Electric could trade at forty times earnings, who was to say what the limit was for Amazon? True, Amazon had no earnings, but in the wonderland of the late 90s, the lack of profits was a boon.
纳指和道指涨3倍，带来了新的其实更可怕的问题：就是蓝筹水涨船高以后，新经济股票就更没有边际了，这种基础估值抬升带来对未来新经济股票的更大的想象空间是加速泡沫的核心一环。目前这件事似乎正在路上，如果做硬件的企业能有50x P/E，模型企业可能就能有100x P/S，这个可能才是未来的主升浪所在。
To grasp the transforming power of fiber optics marked one as a visionary, whereas skeptics—the faithless—were “enemies of the future.”° To doubt technology’s promise, to insist on evidence one could see, was to insult the new religion; it was to be trapped in old-world precepts. Had Columbus seen America before he sailed?
先相信还是先看见，哥伦布肯定是因为相信所以看见，所以技术乐观主义是必要的。
“Faith is central to every process of innovation,” Gilder adjured. “The act of creation is a religious act,” and “the investor who never acts until the financials affirm his choice [is] doomed to mediocrity by [his] trust in spurious rationality.”’ Spurious rationality. The Internet economy was something greater; it existed on a higher plane than the merely rational.
对创新的宗教般的狂热，其实是虚假理性的说法很新颖。
The total of investment surged from $3 billion in 1990 to $60 billion in 1999, implying that venture firms were becoming more confident but also less discriminating.
He repeated again and again that the Valley was spawning the “largest legal wealth creation in history.” The phrase struck a popular chord. If Doerr, who had a history of success with earlier technology companies, such as Compag and Lotus, was plugging the Web, it seemed safe for everyone else. In fact, he told The New Yorker, two years after the Netscape IPO, the Web might well be “underhyped.”
风险投资从30亿美元快速扩张至600亿美元，就很难继续精挑细选模式了。特别是当大人物都对此深信不疑的时候，市场只会更加确认。这个点上，大人物往往更容易出错而不是做对。
He used CMGI stock for every deal he made,and though he acquired companies pretty much on a whim, each (well publicized) deal further raised his stock. This was the game that Kozlowski at Tyco and many others were playing, but Wetherell’s stock was far more virile. Over the five years from its IPO to early 1999, CMGI’s stock rose 140 times, giving it a market value of more than $6 billion.None of CMGI’s various businesses were close to making money.”But the parent generated cash by periodically selling a piece of one of its startups in an IPO. What Wetherell had really built was an IPO farm——the quintessential Internet business. According to Business Week, Wetherell was becoming “a major force on the Web.” Microsoft and Intel each acquired a minority stake.
换股收购、做大做强，股价涨了140x。虽然CMGI的业务都不赚钱，但股价不断上涨。这种类似孵化企业的业务模式让我想起来了当年的ITVC的说法，无疑是市场高点的幻想了——关键是当时的市场条件让这种幻想确实可能实现。
Nonetheless, Wetherell, who spent significant time in market chat rooms,deduced that investors hated the deal. The cash flow standard of Diller simply could not coexist with the airier standards of the new economy. Wetherell backed away from the merger. Wetherell’s choice—stock price over real cash flow—epitomized the bubble as nothing else. The top-performing IPO of 1999 was Internet Capital, an incubator of business Web sites similar in structure to Wetherell’s CMGI. It had gone public in August at $6. It closed on December 31 at $170.
做局者本身还是的清醒，当收购盈利企业做大利润的计划和做大市值相悖时，收购就可以放弃了。收购的标准时股价优先原则。1999年表现最好的IPO居然是一家孵化器，4个月涨了接近30x。
By then, the price-earnings ratio of the average technology stock was well over two hundred. Nasdaq, which had been under 2,000 only sixteen months earlier, broke 5,000. By the peak of the bubble, the first quarter of 2000, a technology company was going public and doubling every other day—a speculative orgy without precedent in public markets.*
2000年的一季度时泡沫高峰，当天上市的一个企业后续股价居然能每天翻倍。这时科技股的平均P/E达到了200x。大盘指数是从1998年下半年开始加速，16个月涨了2.5x。
Such diverse ventures reflected his view that expertise was less critical than talent, which is to say, Skilling believed that Enron could fearlessly enter terrain in which it had little or no experience. Enron thus became something of an open laboratory.
Fastow’s new solution bordered on the diabolical: to create a vehicle that was separate and yet not separate, that was independent of Enron and yet controlled by Fastow personally. In short: a personal fief within the public company. This demand was so counter to the proper spirit of the public corporation as to testify to the unhappy debasement generally of managerial culture—to the long years of executive profiteering. Indeed, Fastow seems to have felt that profiteering was nothing less than his due.
安然是监管放松的产物，也是开放试验的产物。但他本身并不是互联网泡沫下的怪胎，却刺破了这个泡沫。
Nineteen ninety-eight was the year of eBay and Amazon, and Enron wanted in on the party. It wanted to exploit the Web and, just as importantly, to exploit the stock market’s love of the Web. EnronOnline would trade energy products—that much was predictable—but Enron would also trade broadband, that is, capacity on the fiber-optic networks that were crisscrossing the country to carry Internet traffic. More improbably, Enron invested in a fiber network itself, acquiring and building a patchwork system that stretched for 18,000 miles.
1998年随着ebay和亚马逊的上市，安然也想进军互联网了——计划将互联网带宽作为大宗商品的交易，具体交易带宽容量，这个还是很有创意的。这种传统企业进军泡沫新业务的荒诞想法，才是泡沫的高峰，相比之下银行、地产、钢铁等传统产业都要开始AI业务了，才可能是这个泡沫的巅峰吧。
Broadband, the high-speed lines that conveyed millions of e-mails and other data without which modern man would be unable to function, was the shovel to the Internet’s gold. (The businesses were a natural fit, because fiber-optic lines were laid alongside gas pipelines.”) Skilling characteristically decided that fiber-optic capacity could be traded. As it happened, bandwidth capacity was in huge oversupply; in terms of its scarcity value, one might as well have traded air. No matter—Enron, as usual, invested in a capital asset so as to gain entrée as a trader, which in this case meant that it strung together a national fiber network.
In the half-decade after deregulation, telecom companies borrowed $1.6 trillion from banks and enlisted Wall Street to sell $600 billion in bonds.® They raised billions more in stock sales. Most of this money was spent on woefully redundant networks.
宽带网络/带宽是互联网经济中的第一座金矿。电信公司居然借来了2.2万亿美元并投入进去了。相比今天的capex投入还都是大厂的现金流，没有动用太多信贷，足以说明这个泡沫才刚刚起步。也确实如此。
Of course, the trouble with using mergers to increase earnings (as we saw with Tyco) is that one can never stop. In retrospect, this unhappy marriage spelled the end of the dot-com bubble. But few people realized it at the time. There was no outward signal—no obvious trigger. The public appetite for dot-coms was, merely,exhausted by the unceasing parade of IPOs. In the middle of March, two months after the merger had been announced, Internet stocks cracked,and when they did, investors, not unlike the familiar cartoon character.In a mere month, Yahoo was cut in half, Wetherell’s CMGI surrendered two-thirds of its once-stupendous valuation, and Ventro, the brainchild of David Perry, who had driven West in such a haste, was reduced, with similar dispatch, from $239 to $26, or almost 90 percent. In mid-April 2000,the Nasdaq plunged 10 percent in a single day.
泡沫破裂的跌幅是十分凶险的，yahoo可以一个月跌掉一半。CMGI 跌掉2/3，大盘可以一天掉10%。这样的事情要小心。即便是美光、英伟达将来也可能会有跌掉一半甚至更多的可能。
An investor who bought on the dip, as the soothsayers of the 90s had so unceasingly urged, over the ensuing twelve months saw technology issues cut in half yet again. By early 2001, the Nasdaq had plunged from 5,000 to under 2,000, and the value of Amazon had shrunk from a peak of $37 billion to under $5 billion. Internet stocks had been merely the most exotic flower in the hothouse—the last to bloom, the first to expire. It was simply realistic to think that others would expire, too. First the cool of evenings, then the killer frost.
跌幅大的时候左侧可能亏更多，亚马逊的市值从370亿跌到了50亿，15%不到了。最后盛开、最先凋零，这才是可怕的事实。
Telecoms were another matter. The investment in telecoms was at least twenty times greater than in dot-coms.” People thought of their WorldCom shares—even their Global Crossing shares—as safer, less adolescent, than the plunging dot-com stocks. They were more integral to the general economy.
The vehicle for the masquerade was the capacity swap. Carriers had long exchanged leases on pieces of each other’s networks without the exchange of cash or any accounting effect. In the latter half of 2000—that is, at the very time their performance was souring—the game significantly changed. Carriers began to swap broadband leases for large, identical sums of cash. The money was simply round-tripped; checks were sent in both directions to no net effect. Nonetheless, each carrier booked its side of the swap as revenue.
这种宽带交易的swap比今天的循环投资还要可恶，这其实是明目张胆的造假了。只是通过估值技术来实现利润，而不是现金流。要特别小心这种估值繁荣。
Such opportunism was relatively unknown to dot-com executives, many of whom were either idealistic, deluded, or both. Executives at eToys, Webvan, and many others never sold a share; they believed their own stories. Telecom executives were not so naive. Criminal insider trading is difficult to prove, and the executives, later, would protest they had been as unaware as anyone. But the pattern of massive selling, unwarranted optimism, and often deceptive disclosure suggests otherwise.
比起电信大佬们，互联网的创始人还是比较天真，相信自己的故事。电信大佬们就很容易看清现实，只是忙于套现了。
Ronald Barone, an analyst with UBS Warburg, afterward confessed to feeling “brain-drained.” This did not keep him from raising his forecast. Another, Curt Launer, of Credit Suisse, diligently reported, “Trends in all of our modeling inputs point to earnings of $2.20 in 2002.”
Life is rarely that clean. What it shows, rather, is that the executives were increasingly eager to unload the stock just as Enron was reporting increasingly bogus numbers.
分析师吹票永远是泡沫中不可缺少的一环。吹票吹到脑子被榨干。
Wall Street can be fooled, as the memo writer for Global observed, but not forever. When investors catch a scent of weakness, they will turn with surprising haste. All former affection is forgotten; now they want only to sell. And so it was with Enron. Month after month the stock declined—$80 at the end of January, $68 at the end of February, then $58, $62, and $52 at the end of May.
Just when it seemed the public had heard the worst imaginable, there came to light some fresh and more offensive tale of executive wrongdoing. And with each new episode the market fell lower. As John Kenneth Galbraith had written, in The Great Crash, of 1929: “The worst continued to worsen.”
安然事件对泡沫破裂还是至关重要的影响。安然甚至不是一家互联网公司，但其做法刺破了互联网泡沫，这也是需要十分小心和警惕的事情。刺破AI泡沫的，未必是AI公司，可能是其他巨无霸。
it had fallen 11 percent in September alone, then 5 percent more at the start of October. One week after the Fastow in dictment, the S&P 500 closed at 776—exactly half of the peak it had reached in March 2000. The Dow, barely above 7,000, had fallen almost 40 percent. This was its steepest slide in thirty years. Some $7 trillion of the public’s savings had evaporated. The Nasdaq, now close to 1,100,had fallen 78 percent, the largest slide by a major index since the Great Depression. Of some 825 companies that had gone public in 1999 and 2000, 715 were below the offer price.
2000年3月见顶，9月份大盘下跌11%，10月中旬已经Dow跌幅到40%了，纳斯达克跌了78%，泡沫破裂不需要太长时间，6个月就够了，甚至3个月。这个是需要特别去避免的，做好期权仓位全部归零的准备。
The collapse of communism at the end of the 1980s and the daunting advances made in Silicon Valley throughout the ’90s gave the public a heady appetite to invest in shares and ultimately to speculate. However, these factors alone do not explain the headlong abandonment of moral standards.Nor does the refrain, popularized by Alan Greenspan, that America had suddenly become infected by greed.” People have always been greedy,and they have always been subject to speculative frenzy.
2000年泡沫也主要源于当时冷战获胜带来的信心，让大家勇于投资股票。但监管跟不上，人性中的贪婪被逐步放大，从泡沫走向疯狂，最后奔溃。当前的AI乐观主义其实有类似的心里效果。
A better measure of the speculation is that during the Great Crash per se—meaning the swift correction that occurred, despite little change in the economy, in October and November of ’29—stocks fell by half. Coincidentally, in 2000-02, the S&P 500 fell by exactly the same amount; the Dow, a little less; the Nasdaq, considerably more. Billions upon billions were squandered in an array of redundant industries: dot-coms, energy, telecoms, and so forth.
泡沫崩盘的跌幅大致都是50%，个股就完全不同了。这个可以做个极端假设。