{{date}}
Palantir might be the most overvalued firm of all time
What would make it worth buying?
For a few days in March 2000, as the dotcom bubble neared bursting point, Cisco was the world’s most valuable company. Now the seller of networking gear is a cautionary tale, even if it is also an enduring success, with real earnings per share of four and a half times what they were back then. Investors became so exuberant about the firm’s prospects 25 years ago that they valued it at more than 200 times its annual profit, around $1trn in today’s money. Starting from a valuation that stratospheric, Cisco’s solid- but-unspectacular growth was a bitter disappointment. Its market value is now $280bn.
No one can accuse Palantir, a data-analysis outfit and the most searingly hot stock of 2025, of unspectacular growth. It reported revenue of $1bn for the second quarter of this year, 48% higher than for the second quarter of 2024 and quadruple the figure for the same period in 2020. Silicon Valley types seek out companies satisfying the “rule of 40”, meaning that the sum of their operating margin and year-on-year sales growth, both expressed in percentage points, is higher than 40. Palantir’s score on that measure is 94: higher than any other enterprise-software firm with equivalent or greater sales. Among the world’s 25 biggest companies by market value—of which Palantir is one—only Nvidia, with its near-monopoly on artificial- intelligence chips, scores higher.
Any investor would want a piece of that. The trouble is that Palantir’s market value has already soared to $430bn (see chart 1), more than 600 times its past year’s earnings and nearly triple the equivalent multiple for Cisco (or, indeed, Nvidia) at its peak. Software firms often prefer to express their valuation in terms of underlying sales, which puts Palantir’s multiple at around 120. For comparison, in 2005, the year before the Oxford English Dictionary added the verb “Google”, Google’s price-to-sales ratio peaked at 22.
You need not look far to explain why Adam Parker of Trivariate Research, an investment firm, has published a note entitled “Could Palantir be the best short idea?” Writing in late May, he examined the ratio of enterprise value (which adjusts market value to account for debt and cash on the balance- sheet) to forecast sales for the coming year. On this measure Palantir then scored 73 and now scores 104. Mr Parker looked for other listed companies that had hit a multiple of 70 since 2000. Excluding financial firms and those with annual revenue of less than $50m, he found 14, the largest of which has a market value around a quarter of Palantir’s. That was Strategy (formerly MicroStrategy), a firm that sells some software but pitches itself to investors as a “bitcoin treasury company”, with a value derived from its cryptocurrency holdings rather than its sales.
Mr Parker also looked at the shareholder returns such companies have generated. He first lowered the bar to an enterprise-value-to-sales ratio of 30, since so few firms have ever hit Palantir’s heights. He then measured their subsequent returns after first hitting this level, relative to the S&P 500 index (see chart 2). A year after its multiple first hit 30, the median firm had underperformed the index by 22% and seen its multiple contract to 18. What, then, would it take to make Palantir’s shares worth buying? The firm helps everyone from spooks to fast-food chains analyse their data better and thereby improve their operations. Its blistering recent growth comes, in large part, from enthusiasm over adopting AI for such purposes. Palantir’s competitive advantage derives not just from its software and clever engineers, but from a high-level security clearance allowing it to process classified information from America’s defence and intelligence agencies. This gives it a “moat” with which to fend off competitors. It will certainly need one. To reduce its price-to-sales valuation to “only” Google’s at its peak in 2005, while maintaining its current share price, Palantir needs to multiply its revenue by 5.6—substantially more than the barnstorming progress it has made over the past five years. Doing this over the next five would require an annual growth rate above 40%. Sustained revenue growth of that magnitude or more is possible, even at Palantir’s scale: Google (now called Alphabet), Meta and Nvidia have all managed it. Yet none was priced to do so in advance—and they are, after all, among the most successful firms in history. On a call with analysts and investors after announcing Palantir’s most recent results, Alex Karp, the chief executive (pictured), admitted that “this is a once in a generation, truly anomalous quarter.” The remarkable thing is that shareholders need the company’s revenue to continue to grow at a similar rate just to have a good chance of breaking even. There is no allowance for an upstart competitor, a scandal that makes clients wary or even a mere slowdown. If any of those strike, Cisco will need to make way for a new cautionary tale.
Palantir：史上最大泡沫，还是一场惊天豪赌？ 估值已然登天，怎样的未来才配得上它？ 2000年3月，互联网泡沫破裂前夕，思科（Cisco）曾短暂登顶全球市值最高的公司。如今，这家网络设备销售商已成为资本市场的一则“警世恒言”。尽管它至今仍是一家成功的企业，真实每股收益是当年的四倍半，但投资者25年前对其前景的狂热，曾一度将它的市盈盈率推高至200多倍，估值换算成今天的货币将超过1万亿美元。从如此高耸入云的估值出发，思科后来那“稳健却不算惊艳”的增长，带给投资者的只能是深深的失望。如今，它的市值是2800亿美元。 而Palantir，这家数据分析公司，作为2025年资本市场上最灼手可热的明星，没人能指责它的增长“不够惊艳”。公司公布的今年第二季度财报显示，其营收高达10亿美元，比2024年同期增长48%，更是2020年同期的四倍。硅谷信奉一条“40法则”，即一家公司的营业利润率与年销售额增长率之和（均以百分点表示）应高于40。Palantir的得分是94分，在同等或更高销售额的企业软件公司中，无出其右。放眼全球市值最高的25家公司——Palantir已是其中之一——也只有那个几乎垄断了人工智能（AI）芯片的英伟达（Nvidia）能更胜一筹。 任何投资者都想从这样的增长中分一杯羹。然而，问题就在于，Palantir的市值已经飙升至4300亿美元，市盈率超过600倍，几乎是巅峰时期思科（或英伟达）的三倍。软件公司通常更喜欢用市销率（股价与销售额之比）来衡量估值，按此计算，Palantir的市销率高达120倍左右。做个对比：2005年，也就是牛津英语词典收录动词“Google”的前一年，谷歌的市销率最高点也不过22倍。 投资公司Trivariate Research的亚当·帕克（Adam Parker）最近发布了一篇颇具火药味的报告，标题是《Palantir会是最佳做空标的吗？》。他在五月底的文章中，考察了“企业价值与未来一年预期销售额之比”这一指标。当时Palantir的数值是73，如今已是104。帕克先生梳理了自2000年以来，所有触及过70倍大关的上市公司。在排除了金融公司和年收入低于5000万美元的公司后，他只找到了14家，其中市值最大的公司也仅有Palantir的四分之一。这家公司名叫Strategy（前身为MicroStrategy），它虽也销售软件，但向投资者兜售的概念却是“比特币资产储备公司”，其价值更多来源于持有的加密货币，而非软件销售。 帕克还研究了这类公司的后续股东回报。由于很少有公司能达到Palantir的高度，他首先将标准放宽至“企业价值与销售额之比达到30倍”。然后，他衡量了这些公司在首次达到这一水平后的相对回报（与标普500指数对比）。结果显示，在首次触及30倍估值的一年后，这些公司的中位数表现比指数落后了22%，其估值倍数也收缩至18倍。 那么，究竟需要发生什么，才让Palantir的股票值得买入？ 这家公司的业务，是帮助从情报机构到快餐连锁的各类客户更好地分析数据，从而改善运营。近期令人咋舌的增长，很大程度上源于市场对AI应用于此的热情。Palantir的竞争优势，不仅在于其软件和聪明的工程师，更在于它拥有高级别的安全许可，能够处理来自美国国防和情报机构的机密信息。这为它构建了一条深不见底的“护城河”，足以抵御竞争对手。 但这条护城河必须足够坚固。我们来算一笔账：为了让Palantir的市销率降至“仅仅”是谷歌2005年巅峰时的水平，同时维持其当前股价，它的营收需要增长到现在的5.6倍——这比它过去五年间风驰电掣般的增长还要多。要在未来五年内实现这一目标，意味着年均增长率必须持续高于40%。 当然，即便对于Palantir这样的体量，持续如此高速的营收增长也并非天方夜谭：谷歌（现在的Alphabet）、Meta和英伟达都曾做到过。但天壤之别在于，这些巨头当年的股价，并未提前预支这份成功——毕竟，它们是商业史上最成功的公司。在最新财报发布后的分析师电话会议上，Palantir的首席执行官亚历克斯·卡普（Alex Karp）也承认：“这是一个时代一遇的、真正反常的季度。” 最值得玩味的是，股东们需要公司以类似的速度持续增长，才有可能获得盈亏平衡的机会。当前的股价里，没有为任何意外留下空间：无论是突然杀出的竞争对手，让客户望而却步的丑闻，亦或仅仅是一次增长放缓。 一旦以上任何一种情况发生，思科恐怕就要让位，把“警世恒言”的头衔交给Palantir了。
Palantir might be the most overvalued firm of all time
What would make it worth buying?
For a few days in March 2000, as the dotcom bubble neared bursting point, Cisco was the world’s most valuable company. Now the seller of networking gear is a cautionary tale, even if it is also an enduring success, with real earnings per share of four and a half times what they were back then. Investors became so exuberant about the firm’s prospects 25 years ago that they valued it at more than 200 times its annual profit, around $1trn in today’s money. Starting from a valuation that stratospheric, Cisco’s solid- but-unspectacular growth was a bitter disappointment. Its market value is now $280bn.
No one can accuse Palantir, a data-analysis outfit and the most searingly hot stock of 2025, of unspectacular growth. It reported revenue of $1bn for the second quarter of this year, 48% higher than for the second quarter of 2024 and quadruple the figure for the same period in 2020. Silicon Valley types seek out companies satisfying the “rule of 40”, meaning that the sum of their operating margin and year-on-year sales growth, both expressed in percentage points, is higher than 40. Palantir’s score on that measure is 94: higher than any other enterprise-software firm with equivalent or greater sales. Among the world’s 25 biggest companies by market value—of which Palantir is one—only Nvidia, with its near-monopoly on artificial- intelligence chips, scores higher.
Any investor would want a piece of that. The trouble is that Palantir’s market value has already soared to $430bn (see chart 1), more than 600 times its past year’s earnings and nearly triple the equivalent multiple for Cisco (or, indeed, Nvidia) at its peak. Software firms often prefer to express their valuation in terms of underlying sales, which puts Palantir’s multiple at around 120. For comparison, in 2005, the year before the Oxford English Dictionary added the verb “Google”, Google’s price-to-sales ratio peaked at 22.
You need not look far to explain why Adam Parker of Trivariate Research, an investment firm, has published a note entitled “Could Palantir be the best short idea?” Writing in late May, he examined the ratio of enterprise value (which adjusts market value to account for debt and cash on the balance- sheet) to forecast sales for the coming year. On this measure Palantir then scored 73 and now scores 104. Mr Parker looked for other listed companies that had hit a multiple of 70 since 2000. Excluding financial firms and those with annual revenue of less than $50m, he found 14, the largest of which has a market value around a quarter of Palantir’s. That was Strategy (formerly MicroStrategy), a firm that sells some software but pitches itself to investors as a “bitcoin treasury company”, with a value derived from its cryptocurrency holdings rather than its sales.
Mr Parker also looked at the shareholder returns such companies have generated. He first lowered the bar to an enterprise-value-to-sales ratio of 30, since so few firms have ever hit Palantir’s heights. He then measured their subsequent returns after first hitting this level, relative to the S&P 500 index (see chart 2). A year after its multiple first hit 30, the median firm had underperformed the index by 22% and seen its multiple contract to 18. What, then, would it take to make Palantir’s shares worth buying? The firm helps everyone from spooks to fast-food chains analyse their data better and thereby improve their operations. Its blistering recent growth comes, in large part, from enthusiasm over adopting AI for such purposes. Palantir’s competitive advantage derives not just from its software and clever engineers, but from a high-level security clearance allowing it to process classified information from America’s defence and intelligence agencies. This gives it a “moat” with which to fend off competitors. It will certainly need one. To reduce its price-to-sales valuation to “only” Google’s at its peak in 2005, while maintaining its current share price, Palantir needs to multiply its revenue by 5.6—substantially more than the barnstorming progress it has made over the past five years. Doing this over the next five would require an annual growth rate above 40%. Sustained revenue growth of that magnitude or more is possible, even at Palantir’s scale: Google (now called Alphabet), Meta and Nvidia have all managed it. Yet none was priced to do so in advance—and they are, after all, among the most successful firms in history. On a call with analysts and investors after announcing Palantir’s most recent results, Alex Karp, the chief executive (pictured), admitted that “this is a once in a generation, truly anomalous quarter.” The remarkable thing is that shareholders need the company’s revenue to continue to grow at a similar rate just to have a good chance of breaking even. There is no allowance for an upstart competitor, a scandal that makes clients wary or even a mere slowdown. If any of those strike, Cisco will need to make way for a new cautionary tale.
Palantir：史上最大泡沫，还是一场惊天豪赌？ 估值已然登天，怎样的未来才配得上它？ 2000年3月，互联网泡沫破裂前夕，思科（Cisco）曾短暂登顶全球市值最高的公司。如今，这家网络设备销售商已成为资本市场的一则“警世恒言”。尽管它至今仍是一家成功的企业，真实每股收益是当年的四倍半，但投资者25年前对其前景的狂热，曾一度将它的市盈盈率推高至200多倍，估值换算成今天的货币将超过1万亿美元。从如此高耸入云的估值出发，思科后来那“稳健却不算惊艳”的增长，带给投资者的只能是深深的失望。如今，它的市值是2800亿美元。 而Palantir，这家数据分析公司，作为2025年资本市场上最灼手可热的明星，没人能指责它的增长“不够惊艳”。公司公布的今年第二季度财报显示，其营收高达10亿美元，比2024年同期增长48%，更是2020年同期的四倍。硅谷信奉一条“40法则”，即一家公司的营业利润率与年销售额增长率之和（均以百分点表示）应高于40。Palantir的得分是94分，在同等或更高销售额的企业软件公司中，无出其右。放眼全球市值最高的25家公司——Palantir已是其中之一——也只有那个几乎垄断了人工智能（AI）芯片的英伟达（Nvidia）能更胜一筹。 任何投资者都想从这样的增长中分一杯羹。然而，问题就在于，Palantir的市值已经飙升至4300亿美元，市盈率超过600倍，几乎是巅峰时期思科（或英伟达）的三倍。软件公司通常更喜欢用市销率（股价与销售额之比）来衡量估值，按此计算，Palantir的市销率高达120倍左右。做个对比：2005年，也就是牛津英语词典收录动词“Google”的前一年，谷歌的市销率最高点也不过22倍。 投资公司Trivariate Research的亚当·帕克（Adam Parker）最近发布了一篇颇具火药味的报告，标题是《Palantir会是最佳做空标的吗？》。他在五月底的文章中，考察了“企业价值与未来一年预期销售额之比”这一指标。当时Palantir的数值是73，如今已是104。帕克先生梳理了自2000年以来，所有触及过70倍大关的上市公司。在排除了金融公司和年收入低于5000万美元的公司后，他只找到了14家，其中市值最大的公司也仅有Palantir的四分之一。这家公司名叫Strategy（前身为MicroStrategy），它虽也销售软件，但向投资者兜售的概念却是“比特币资产储备公司”，其价值更多来源于持有的加密货币，而非软件销售。 帕克还研究了这类公司的后续股东回报。由于很少有公司能达到Palantir的高度，他首先将标准放宽至“企业价值与销售额之比达到30倍”。然后，他衡量了这些公司在首次达到这一水平后的相对回报（与标普500指数对比）。结果显示，在首次触及30倍估值的一年后，这些公司的中位数表现比指数落后了22%，其估值倍数也收缩至18倍。 那么，究竟需要发生什么，才让Palantir的股票值得买入？ 这家公司的业务，是帮助从情报机构到快餐连锁的各类客户更好地分析数据，从而改善运营。近期令人咋舌的增长，很大程度上源于市场对AI应用于此的热情。Palantir的竞争优势，不仅在于其软件和聪明的工程师，更在于它拥有高级别的安全许可，能够处理来自美国国防和情报机构的机密信息。这为它构建了一条深不见底的“护城河”，足以抵御竞争对手。 但这条护城河必须足够坚固。我们来算一笔账：为了让Palantir的市销率降至“仅仅”是谷歌2005年巅峰时的水平，同时维持其当前股价，它的营收需要增长到现在的5.6倍——这比它过去五年间风驰电掣般的增长还要多。要在未来五年内实现这一目标，意味着年均增长率必须持续高于40%。 当然，即便对于Palantir这样的体量，持续如此高速的营收增长也并非天方夜谭：谷歌（现在的Alphabet）、Meta和英伟达都曾做到过。但天壤之别在于，这些巨头当年的股价，并未提前预支这份成功——毕竟，它们是商业史上最成功的公司。在最新财报发布后的分析师电话会议上，Palantir的首席执行官亚历克斯·卡普（Alex Karp）也承认：“这是一个时代一遇的、真正反常的季度。” 最值得玩味的是，股东们需要公司以类似的速度持续增长，才有可能获得盈亏平衡的机会。当前的股价里，没有为任何意外留下空间：无论是突然杀出的竞争对手，让客户望而却步的丑闻，亦或仅仅是一次增长放缓。 一旦以上任何一种情况发生，思科恐怕就要让位，把“警世恒言”的头衔交给Palantir了。