SignalNotNoise

Beijinger
@TKAWPB & @georgiatech & @mitSloan alum. Pursuing the perfect portfolio. Everything finance & tech.

[Art of Speculation @ArtofSpecuycky]

如果Warsh像1999年的格林斯潘，AI牛市可能才刚刚开始
最近在读格林斯潘的《动荡的世界》和《繁荣与衰退》，突然想到一个问题：Warsh和1999年的格林斯潘，到底有多像？

先说他们像的地方
两个人有几个共同点是确实存在的。
都偏自由市场，相信竞争和创新的力量远大于政府干预。都认为央行很难精准识别泡沫，主动刺破泡沫的代价往往高于事后救助，因此对资产价格上涨有更高的容忍度。都反对过度前瞻指引，更喜欢保留灵活性。
最核心的相似点是对生产率革命的信仰。
格林斯潘在《繁荣与衰退》里反复强调一个观点：生产率才是一切繁荣的来源。美国之所以能一次次走出危机，靠的是技术创新不断提高生产率。从铁路到电力到汽车到互联网，旧产业死去，新产业崛起，创造性破坏才是资本主义真正的生命力。
1999年他面对的正是互联网革命。当时按传统菲利普斯曲线模型，低失业率本应带来更高的通胀压力。但格林斯潘认为互联网革命改变了生产率，因此不必机械地依赖旧模型，选择维持相对宽松的利率。
现在回头看Warsh最近的公开讲话，逻辑有相似之处。他认为AI基础设施有潜力成为强大的通胀抑制力量，如果AI能让企业生产效率持续提升，美联储就不需要急着转鹰。
这里有一个重要的区别要说清楚。现在AI还处于CapEx先爆发、收益后兑现的阶段，真正像互联网90年代那样让全社会生产率大幅提升，至少还要几年时间验证。Warsh可能相信AI的长期潜力，但AI已经压低通胀这个结论目前证据还不够充分。格林斯潘当时已经能看到生产率数据的改善，而Warsh现在更多是在押注一个尚未完全兑现的未来。

第二个相似点：对通胀指标的处理方式
格林斯潘时代，美联储开始越来越重视PCE。PCE本身有其合理性：权重动态变化，能捕捉消费者的替代效应，覆盖范围比CPI更广。从历史数据来看，PCE往往比CPI低0.3到0.5个百分点。这意味着在同样的经济环境下，PCE框架通常会给货币政策更大的灵活空间。这一趋势后来在伯南克时代正式确立，将PCE作为2%通胀目标的锚定指标。
Warsh对传统通胀指标一直持保留态度。他曾公开表示，核心PCE并不能完美反映底层通胀趋势。相比单一指标，他似乎更倾向于同时观察核心PCE、中位数CPI和修剪均值通胀。修剪均值的计算方式是把所有商品价格涨跌从低到高排列，剔除掉两头最极端的项目，在高通胀周期里通常会给出比核心PCE更温和的数字。
如果这种思路进入Fed决策框架，货币政策可能获得更大的解释空间。从格林斯潘时代强化PCE，到Warsh可能引入更多元的观察指标，方向是一致的：为货币政策保留更多灵活性。

第三个相似点：重回建设性模糊
很多人误以为Warsh反对前瞻指引，这个理解有些片面。
他真正反对的，是鲍威尔时代那种近乎明牌的沟通方式。
格林斯潘时代，美联储开始逐渐走向透明，但他始终坚持一个原则：给市场方向，不给市场剧本。他会告诉市场自己关注什么、担心什么，以及经济大致会往哪个方向走，但绝不会把未来几年每一次利率会议的路径提前画出来。
他最著名的一句话是：如果你觉得听懂了我的话，那你一定误解了我的意思。
这当然不是故弄玄虚。格林斯潘相信，央行最重要的资产之一就是相机抉择的灵活性。经济是动态变化的，央行必须保留随时调整方向的权利，一旦提前承诺太多，就会被自己的话绑住手脚。
Warsh批评的，恰恰是鲍威尔时代越来越强的路径依赖。每季度公布点阵图，每次会议后召开长时间发布会，市场甚至开始逐字逐句拆解每一个委员未来几年的利率预测。这种极度透明的好处是预期稳定，代价则是美联储越来越像一个需要不断兑现承诺的机构。
2021到2022年就是最典型的例子。在通胀暂时论的框架下，美联储长时间维持宽松预期。当通胀超预期失控时，政策转向变得异常痛苦，因为市场早已把之前的承诺交易进了价格。
所以Warsh真正想回到的，可能是格林斯潘式的建设性模糊：多谈长期目标与风险，少谈具体点位和时间表，给市场方向感，同时保留足够的调整空间。
这也是为什么华尔街现在最担心的，反而是确定性消失了。过去几年市场习惯了Fed提前告诉你答案，而Warsh可能把这个习惯打破，留下的是更大的不确定性。
市场担心的不是Warsh会取消点阵图，而是他会降低市场对点阵图的依赖，弱化长期降息承诺。甚至在首场会议上，上调通胀预期，重新定义中性利率，释放出比市场想象更鹰派的长期利率路径。
如果真是这样，最大的变化可能不是利率本身，而是市场突然发现，过去几年习惯的那种Fed永远提前告诉你答案的时代，结束了。剩下的，是要重新学会在更大的不确定性里做决策。

如果Warsh真的走这条路，股市会经历什么？
我觉得会有三个阶段。
第一阶段是估值扩张。市场会开始交易一个叙事：Fed不会轻易打断AI投资周期。前提是AI生产率开始真正兑现，一旦这个叙事成立，PS和PE都会扩张，最先受益的是AI基础设施、半导体、软件平台和高弹性成长股。逻辑和1998到1999年互联网那波完全一样，市场愿意给更高的倍数，因为他们相信美联储站在这边。
第二阶段是赢家通吃。1999年不是所有科技股都涨，最后真正活下来并且涨疯的，是那些收入模式最扎实的公司——网络基础设施、龙头软件和核心平台。纯概念的小票逐渐掉队。类比今天，最后可能集中在AI算力、光通信、数据中心基础设施、AI云平台这几条线。有真实收入和盈利能力的公司会越来越贵，纯叙事的标的会慢慢被抛弃。
第三阶段是估值脱离基本面。市场不再关心利润多少、PE多少、收入能不能兑现，只关心下一个买家会不会出更高的价。这个阶段格林斯潘当年选择了等待，最终等来了2000年的泡沫破裂。Warsh会做同样的选择吗？这个问题放在后面再说。

但很多人搞错了一件事：加息会压住牛市吗？
1999年格林斯潘从4.75%一路加息到6.5%，按传统逻辑科技股应该崩。
结果纳指那一年涨了超过80%。
原因在这里。利率是估值模型里的分母，但盈利增长是分子。当分子的增长速度超过分母的压缩速度，股市照样疯涨。1999年经济增长4%到5%，生产率大幅提升，市场认为即使利率高一点，公司利润还是会高速增长，所以PE继续扩张。
这给现在一个很重要的参考：加息本身不能终结牛市。利率高到盈利跟不上估值，或者市场发现增长故事是假的，才是牛市真正的终点。

这次和1999年有几个本质差异不能忽略
1999年美国联邦债务占GDP只有50%左右，财政赤字压力远小于今天。现在政府融资需求巨大，市场对通胀重新抬头极度敏感。
格林斯潘在《动荡的世界》里自己也承认，他当年最大的判断错误之一，是低估了金融体系在泡沫中积累的系统性风险。今天的杠杆比1999年更高，传导速度也更快。
所以即使Warsh风格偏向市场派，他也很难复制格林斯潘那样连续几年极度宽松的金融环境。更像的剧本是允许AI牛市继续演绎，但在通胀和金融条件之间不断拉扯，在经济增长、通胀和资产价格之间寻找新的平衡。

最后想讨论个问题
如果央行开始相信AI带来的生产率革命，市场会不会再次经历一场1999式的超级牛市？
如果AI生产率革命真的成立，AI CapEx继续增长、龙头科技不断创新高、PS和PE继续扩张的组合。市场会重新讨论传统DCF模型是不是低估了AI。
反过来，如果企业发现AI的投资回报率没有想象中那么高，CapEx开始放缓，今天所有高PS的公司都会经历1999年互联网泡沫后期的估值杀。
格林斯潘在《动荡的世界》里写道，他最后不得不承认，人类的非理性、动物精神和风险厌恶，是任何经济模型都无法完全预测的变量。传统经济学在危机面前失灵，只能抓住行为经济学这最后一根稻草。
事实证明他错了一半，又对了一半。互联网泡沫破灭了，但Amazon和Google活了下来，互联网真的改变了世界。他不是看错了方向，只是低估了市场会有多疯狂。
这句话放在今天的AI周期里，依然值得反复思考。
如果Warsh真的像1999年的格林斯潘，他会不会也犯下同样的错误，明知道泡沫正在形成，却因为相信技术革命，选择让市场自己走到终点？
这个问题的答案，比任何一次FOMC会议都更重要。

[Ricky Ho @rickyho_1989]

The charts highlight something that many investors still underestimate: AI is rapidly becoming a scale business rather than a pure technology business.

At the frontier, Anthropic remains the intelligence leader. Claude Fable 5 scores 60 on the Artificial Analysis Intelligence Index, ahead of Opus 4.8 at 56 and GPT-5.5 at 55. However, the second chart shows that this final increment of intelligence comes at a very steep cost. Fable 5 costs roughly $3.25 per benchmark task versus $1.78 for Opus 4.8 and $0.99 for GPT-5.5.

In other words, Anthropic appears to be operating on the far right side of the efficiency curve where each additional point of intelligence requires disproportionately more compute, more reasoning tokens, and more inference cost. The company currently owns the performance crown, but it is paying heavily for it.

The more interesting story is happening below the frontier. DeepSeek V4 Pro Max delivers 44 intelligence points for around $0.06 per task. MiMo V2.5 Pro delivers 42 points for roughly the same cost. Gemini Flash reaches 50 points while remaining dramatically cheaper than the frontier models. The intelligence gap between the best and the “good enough” models remains meaningful, but the cost gap is becoming enormous.

This is exactly what happens when a technology industry starts to mature. The market initially competes on absolute performance. Eventually it begins competing on performance per dollar.

The scatter plot makes this particularly clear. The most attractive area is not necessarily the top-right corner where intelligence is highest. It is the upper-left area where intelligence remains high while costs collapse. Many Chinese models now sit squarely in that region.

This is why the narrative that “the cheapest model wins” is just as wrong as the narrative that “the smartest model wins.” Neither is sufficient. The winners are likely to be the models that sit on the Pareto frontier, delivering the highest intelligence for a given cost level. Today that frontier includes Anthropic at the very high end, Google in the middle, and a growing group of Chinese labs including DeepSeek, Qwen, MiniMax, and Xiaomi at the efficiency end.

For investors, this has an important implication. Falling inference costs are not necessarily bearish for AI infrastructure. Historically, lower computing costs have almost always led to more consumption rather than less. We saw it with storage, bandwidth, cloud computing, and semiconductors. As intelligence becomes cheaper, companies simply run more agents, automate more workflows, and expand AI usage into new domains.

The charts therefore reinforce a view that we have held for some time: AI is not becoming less valuable because models are getting cheaper. AI is becoming more ubiquitous because models are getting cheaper.

Anthropic currently leads the intelligence race. Chinese labs are increasingly leading the efficiency race. Google appears to be competing effectively across both dimensions. The result is an industry that is simultaneously moving up the intelligence curve and down the cost curve. That combination is exactly what drives mass adoption.

The most bullish takeaway is that the industry is no longer trading off intelligence against cost. It is increasingly improving both at the same time. Historically, that is the hallmark of a technology entering its exponential adoption phase rather than approaching saturation.

[Chase Passive Income @chasedownleads]

Kansas just renamed itself Iran to unlock $300 billion in federal funding

This is a genius

[Epik Spor Anları #WC2026 @epicsp0ranIari]

2006 Dünya Kupası'nda sahaya çıkan 764 futbolcudan sadece 4'ü kaldı. 🐐

[Formula 1 @F1]

FIRST WIN IN RED ❤️

#F1 #BarcelonaGP

[ArtButMakeItSports @ArtButSports]

Statue of Liberty, Frédéric Auguste Bartholdi & Gustave Eiffel, dedicated 1886, 📸’s by @natlyphoto, PannaPhoto

[Timeless Sports @timelesssports_]

2016 NCAA Champions.

2026 NBA Champions.

[Hunter Biden @HunterBiden]

To everyone so eager to cancel someone for a tattoo they got at age 22, a drunk text, a selfie they took in the middle of a mental health crisis:

Show us your laptop.
Show us your iCloud.
Open your entire digital life to your worst enemy. No context. No filter. No explanation.

You won’t.

You won’t because you know what I know. Any one of us, frozen at our worst moment, photographed in our lowest hour, looks like a monster. Looks like a stranger. Looks like someone who deserves to be cast out.

That is not who we are.

My mom and baby sister were killed in a car accident when I was just a kid. Cancer took my brother Beau, my best friend and my rock. I battled alcoholism. I battled addiction. I chose the coward’s way out more times than I can count.

For years I believed the defining chapters of my life were written by tragedy, loss, and shame.

I no longer believe that.

Pain can shape us. Loss can humble us. Failures can leave scars that never fully fade. But none of them have the authority to define us.

And it sure as hell ain’t the critic that counts.

That authority belongs to us alone-the person in the arena.

Every setback presents a choice. Play the victim, or cut the bullshit and take ownership for who we become next.

Life does not determine our character. It reveals it.

Again and again we are asked the same question. When shit happens, what next?

We are not defined by what happened to us. We are not defined by the worst photo, the worst text, the worst tattoo, the worst night. We are defined by the person we choose to become. And by the courage to choose that person, every single day.

So before you reach for the gavel - show us your laptop.

You won’t.

The whole world saw mine. And I am still here. Still becoming. Still choosing. Still standing.

That is the only definition that matters.

[ArtButMakeItSports @ArtButSports]

Cast of Laocoön and his Sons (Roman version of a lost Greek original), 100BC-50AD

[Bojan Tunguz @tunguz]

Our Anthropic overlords deciding which prompts the peasants are allowed to use.

[ol’ stocky ⛳️ @oldstocky]

I can understand why Italians saw this and decided to invent unemployment

[terminally onλine εngineer @tekbog]

me to mythos: hi

mythos:

[NBA Memes @NBAMemes]

Jalen Brunson’s iconic three-point celebration has been with him since day one. 👏🏻

[sports live tweeter yadira @jonmoxIeys]

SHORT MEN BELIEVE IN UR DREAMS

[dylan @dylangonzalez21]

He’s so awesome

[ArtButMakeItSports @ArtButSports]

Hercules and Nessus, by Giambologna, 1599

[Boring_Business @BoringBiz_]

State of AI right now

[F1 TROLL @f1trollofficial]

Verstappen, Norris and Leclerc at the Monaco GP