字号·· | 护眼
CNBC

英伟达丑闻之际芯片股反弹,“坏消息出尽即见底”这句老话应验'Markets bottom on bad news' maxim rings true as chip stocks rally amid Nvidia scandal

点「原文对照」整页切到原文,或双击某段只看那段的原文。

“坏消息出尽即见底”是那种听起来很聪明的经验法则之一,它应验的频率恰好高到足以让这句华尔街老话至今仍稳稳立足。

"Markets bottom on bad news," is one of those smart-sounding rules of thumb that proves true just often enough to remain a Wall Street cliché in good standing.

对AI股票的残酷惩罚持续了数周,上周是否已到高潮,并在周一因一连串不利头条新闻而反转走高?过去一周,美国选民在全国大部分地区以压倒性多数反对数据中心建设,而Anthropic在下调IPO目标估值后不久,一场性丑闻与间谍丑闻就击中了整个AI经济的化身——英伟达的黄仁勋。

Did the brutal weeks-long punishment of AI stocks culminate last week and reverse higher Monday after a pileup of adverse headlines? In the past week, American voters resoundingly opposed data-center construction across much of the country and Anthropic lowered its target IPO valuation just before a sex-and-espionage scandal struck the one-man embodiment of the entire AI economy, Nvidia's Jensen Huang.

在纳斯达克100指数周一上涨3.4%、费城半导体指数(SOX)飙升7%之后,尽管英伟达下跌4.3%,“市场在坏消息达到高潮、经历一次严重挫折后上演强劲反弹”的说法目前还算站得住脚。

The story of a market taking a cue from a crescendo of bad news after a nasty setback to stage a resounding bounce holds together well enough for now, after Monday's 3.4% jump in the Nasdaq-100 and 7% rip in the Philadelphia Semiconductor Index (SOX), even as Nvidia shed 4.3%.

仅从战术证据来看,科技硬件板块进入本周时已经相当超卖,具备了反弹条件。多家交易台报告称,在截至上周四的风险出清中,机构对该板块的净卖出强度在历史上罕见。纳斯达克100指数在多项指标上的超卖程度,与3月30日伊朗抛售低点时相当。芝加哥期权交易所标普500波动率指数出现了一波通常与风暴过境相关的漂亮飙升,冲高至28,随后回落至20以下。

Based purely on the tactical evidence, the tech-hardware complex came into the week pretty well washed out and primed to bounce. Numerous trading desks reported historically rare intensity of net selling in the group by institutions into last Thursday's risk purge. The Nasdaq-100 was as oversold on several metrics as it had been at the March 30 Iran-selloff low. The CBOE S&P 500 Volatility Index has put in a nice spike of the sort usually associated with a passing storm, shooting up to 28 and then receding below 20.

贝尔德策略公司的期权策略师洛朗·拉斯科夫斯基表示,近期纳斯达克指数的暴跌和避险情绪的升温,“具有许多投降式抛售的特征”:期权交易量创下纪录、波动率、偏斜度和相关性急剧飙升、VIX曲线倒挂,以及被迫平仓……因此,市场当前的格局是,下行左尾正在收窄,而上行右尾正在变厚。几位技术分析大师(包括下文将引用其观点的宏观风险顾问公司的约翰·科洛沃斯)认为,市场还没有给出一个足够好的“正中好球”,让人敢于大举逢低买入。

The recent dump in the Nasdaq and uptick in risk aversion, "had many elements of capitulation," says Laurent Laskowski, options strategist at Baird Strategas: "record options activity, a sharp spike in volatility, skew and correlations, an inverted VIX curve, and forced liquidations…The setup, therefore, is one where the market is seeing the [downside] left tail narrowing while the [upside] right tail is fattening." Several technical-analysis gurus (including Macro Risk Advisors' John Kolovos, whose work is cited below) argue the market hasn't quite served up a "fat pitch" to buy the pullback aggressively.

也许如此。但FanGraphs计算得出,在美国职业棒球大联盟中,只有约7%的投球落在好球区正中央。尽管这些“正中好球”贡献了全部本垒打的约23%,但其余四分之三飞出墙外的球,都是在好球区边缘、条件并不完美的情况下打出的。

Maybe so. But FanGraphs calculates that in Major League Baseball only about 7% of all pitches are in the very center of the strike zone. While those fat pitches result in some 23% of all home runs hit, the other three-quarters of balls that fly over the wall come with less-perfect setups around the edges of the zone.

我也不太会极力反驳这轮以科技股为中心的涨势还会持续一段时间,原因除了上述几点,还有估值已变得远没有那么苛刻。上周我曾指出,大型科技股此前尚未对中度承压的市场环境作出反应,但已越来越接近作出反应,并且“建立在AI叙事之上的牛市,几乎肯定需要AI叙事重回正轨才能继续推进。否则,等权重标普500指数跑赢市值加权版本持续很长一段时间,通常就被称为熊市。”周一的小幅上涨可能是一个开始,但仍有待验证的地方很多。

I wouldn't argue too strenuously against this tech-centered rally carrying on for a bit, for these reasons as well as the fact that valuations have become far less demanding. Last week, I noted Big Tech had not to that point responded to moderately stressed conditions but was getting closer, and "the bull market built on the AI story will almost surely need the AI story to get back on track to keep advancing. A prolonged period when the equal-weight S&P 500 outperforms the market-cap-weighted version is otherwise known – usually – as a bear market." Monday's blip higher could be a start, with plenty still to prove.

但市场由AI驱动的这一部分近来表现出不那么值得投资者信任,给看涨理由留下了足够多的结构性损伤迹象,以至于信心无法立即恢复。

But this AI-driven part of the market has lately shown itself less worthy of investor trust, leaving enough hints of structural damage to the bull case that faith can't be restored right away.

在6月底的一篇专栏中,我曾问道:“5月14日对这个市场来说是否是一个具有重大意义的时刻,至少从战术层面看?”那一天距今已近六个月,标普500指数当时首次冲上7500点;此后该指数再也未能达到逃逸速度,事实上直到昨天才再次突破7500点。

In a late-June column, I asked, "Was May 14 a consequential moment for this market, at least tactically?" That date, nearly six months ago, was when the S&P 500 spurted higher to reach the 7500 level for the first time; the index has never reached escape velocity since and in fact crossed above 7500 again just yesterday.

5月14日正是备受吹捧的AI芯片新秀Cerebras上市之日,其股价翻了一倍多,最高达到386美元,当时据报道特朗普总统与中国领导人在结束伊朗战争方面取得了进展。自那以后,Cerebras已暴跌至97美元,半导体板块下跌17%,而伊朗冲突仍在持续。

May 14 was when highly touted AI-chip upstart Cerebras came public and its shares more than doubled to a high of $386, as President Trump and China's leader were reported to have made progress on ending the Iran war. Since then, Cerebras has collapsed to $97, the semiconductor group is off 17% and the Iran conflict continues.

一个月后,SpaceX的IPO似乎是新时代科技热情的又一个明显顶峰;它或许并非AI热潮的重要顶部,但如果确实是,那么此后的市场走势可能与我们已经看到的情况相似。

A month later the SpaceX IPO was another apparent culmination of New Era tech enthusiasm; it might not have been an important peak in AI excitement, but if it were the market action since would possibly resemble what we've seen since.

或许,市场连续两个季度拒绝为30%的年化盈利增长买单,这半年的清算期已足以在更为温和的预期之上构筑一个新的底部。我一直关注的是晚周期震荡对市场的扰动,表现为盈利增长见顶、资本开支见顶、指数集中度见顶、美联储收紧以及信贷事件突发。这一切解释了市场数月来的停滞,而这些因素依然存在。

Perhaps a half-year reckoning period during which the market has refused to pay up for 30% annual earnings growth for two straight quarters is enough to build a new base constructed on more modest expectations. I've been focused on late-cycle tremors unsettling markets in the form of peak earnings growth, peak capex, peak index concentration, a tighter Federal Reserve and credit flareups. This all explains the market's multi-month stall, and they remain.

但重要的是,要对另一种可能保持开放心态:这些最终可能不过是周期中段的动荡,正如摩根士丹利首席投资官迈克·威尔逊一直所主张的那样。

But it's important to remain open to the chance that these could end up amounting to mere mid-cycle turbulence, as Morgan Stanley chief investment officer Mike Wilson has been arguing.

不过,首先,行情需要展现出比一次漂亮的、广泛的单日反弹更多的进展。另一条经验法则是,反弹行情至少需要站上20日移动均线,才能证明它不仅仅是条件反射式的死猫跳。这意味着标普500指数需要较周一收盘上涨1.3%,纳斯达克100指数需要上涨3.7%,费城半导体指数需要上涨9%。

First, though, the tape will need to show more progress than a nice, broad single-day comeback. Another rule of thumb holds that a rebound rally needs at least to carry above the 20-day moving average to prove that it's not merely a reflexive dead-cat bounce. This would require a 1.3% gain from Monday's close for the S&P 500, 3.7% for the Nasdaq-100 and 9% for the SOX.

虽然我们无法预判各项针对英伟达的调查将对其中国销售、领导层或长期盈利能力产生何种影响,但可以肯定的是,这场风波的爆发时机对市场而言颇为巧合。

Without pretending any clear sense of what the various Nvidia investigations will mean for its China sales, leadership or long-term earnings power, we can say for sure that the scandal arrived at a fairly opportune time for the market.

在人们持续担忧当前利润不可持续、且该集团仍易受数据中心建设放缓冲击的背景下,英伟达的估值已被打压至多年低位。这使得在一天的交易中,其他所有半导体股票纷纷上扬,而英伟达市值蒸发约2000亿美元,成交额高达约800亿美元。

Valuations have been pounded to multi-year lows on persistent worry that current profits are unsustainable and the group remains vulnerable to any slowdown in data-center builds. This allowed, for a day, every other semiconductor stock to whistle higher as Nvidia lost some $200 billion in market cap on a blistering $80 billion in dollar-volume traded in the stock.

当然,标普500指数整体在光学上也变得更便宜了,半导体板块支撑了预期盈利,但投资者不愿给予该板块盈利可持续性的认可。

The S&P 500 as a whole, of course, has grown optically cheaper, too, with semis bolstering the projected earnings and investors declining to give the group credit for their sustainability.

问题在于,随着超大规模云服务商疯狂支出,这些盈利中最终能转化为自由现金流的部分将减少。正如这张对比估值指标的图表所示,以自由现金流衡量,市场看起来比以盈利衡量要贵得多。

The issue remains that less of those earnings are set to manifest as free cash flow as the hyperscalers spend manically. As this chart comparing the valuation gauges shows, the market looks a good deal less cheap on free cash flow than earnings.

从当前水平看,市场的一个合理看涨理由可能是:Alphabet、亚马逊、Meta和微软等巨头将恢复收割自由现金流,投资者或许愿意为其给予比那些炙手可热却难以令人信赖的芯片制造商更高的估值倍数。不过,这目前尚无法被证实。

A proper bull case for the market from here might be that the likes of Alphabet, Amazon, Meta and Microsoft will resume harvesting free cash flow, on which investors might place a higher multiple than they're willing to bestow on those hot but hard-to-trust chip makers. Not yet a provable argument.

《纽约客》这篇评论詹姆斯·格雷克《电话:一部新史》的文章读来有趣又富有启发性,它不可避免地让人联想到电信业的诞生与当下AI建设热潮之间的相似之处。庞大的私人融资基础设施建设、 destabilizing 人与技术互动的新习惯,以及不断有人呼吁政府监管以换取近乎垄断的权力。书中引用了AT&T 1909年的一则广告,称电话是“‘文明……扩大人类智慧最强大的影响力之一’的工具”。

This enjoyable and illuminating review in the New Yorker of James Gleick's "The Telephone: A New History" inevitably summons parallels between the advent of telecommunications and the current AI buildout. Huge privately financed infrastructure buildout, destabilizing new habits of interaction with technology and constant calls for government regulation in exchange for near-monopoly power. The book cites an AT&T ad from 1909 calling telephones "tools of 'civilization . . . one of the most powerful influences for broadening human intelligence.'"

穆迪分析经济学家马克·赞迪有一个颇有帮助的推文串,指出迄今为止AI为美国经济增加的就业岗位多于它取代的岗位。不过主要原因是令人清醒的,大部分收益来自数据中心建设以及为持有依赖AI股票的人创造的财富。对信息经济的求职者来说,这算不上什么安慰。

Moody's Analytics economist Mark Zandi has a helpful thread suggesting AI is so far adding more jobs than it's costing the U.S. economy. The main reasons are sobering though, with most gains coming from data-center construction and wealth creation for those who own AI-dependent stocks. Cold comfort for information-economy job seekers.

上周的选举引发民主党接管国会,其规模和传递的信息让人将其与1994年共和党以“与美国契约”取得的大胜相提并论。那场选举发生在比尔·克林顿第一个任期内,凸显了深刻的社会裂痕和经济不稳定。

Last week's election triggering a Democratic takeover of Congress brought comparisons to the 1994 "Contract with America" Republican sweep for its magnitude and message. That election, in Bill Clinton's first term, highlighted deep societal fissures and economic instability.

事后看来,许多人将90年代视为一个共识与繁荣的平静时代,但从94年那场反弹式选举、持续的企业裁员、对联邦债务水平的危言耸听、1995年俄克拉何马城爆炸案,到对技术变革的不安,那个十年击中了许多当今叙事的节拍。芝加哥联储这篇论文《90年代工作保障的衰落:失业、焦虑及其对工资增长的影响》可以为那些未曾经历那个时代的人填补一些空白。

With hindsight, many view the '90s as a placid era of consensus and prosperity, but between the '94 backlash election, constant corporate downsizing, scare-mongering over Federal debt levels, the 1995 Oklahoma City bombing and discomfort with technological change, that decade hit many of today's narrative beats. This Chicago Fed paper, "The decline of job security in the 1990s: Displacement, anxiety, and their effect on wage growth" can fill in some blanks for those not around for it.

祝疫情时期投机性科技股牛市见顶五周年快乐。距离标普500指数在2022年1月见顶、美联储准备遏制飙升的通胀(或许是通过引发衰退)还有不到两个月时,几个由散户交易者推动的关键子板块已经见顶。

Happy fifth anniversary to the peak of a pandemic speculative-tech bull market. Not quite two months before the S&P 500 hit a top in January 2022 as the Fed prepared to choke off surging inflation, perhaps by causing a recession, several key retail-trader-driven subsectors peaked.

2021年11月的第三周是特斯拉相对表现的峰值,也是Coinbase和Wisdom Tree云计算ETF(WCLD)的绝对顶部。

The third week of November 2021 was the relative-performance peak in Tesla, as well as the absolute top in Coinbase and the Wisdom Tree Cloud Computing ETF (WCLD).

五年回溯期是纳斯达克综合指数未能跑赢标普500指数的少数几个不足25年的时间跨度之一。大盘能够承受如此内部的投机性爆发,这是一件好事,但也很好地提醒人们,市场中最有趣的角落不会永远保持那种状态。

The five-year lookback period is one of the few time spans shy of 25 years over which the Nasdaq Composite has not outperformed the S&P 500. It's a plus that the broad market can withstand such internal speculative blowoffs, though also a good reminder that the most fun pockets of the market don't stay that way forever.

阅读媒体原文 ↗