Tech • AI • Robotics • Game
SELL EVERYTHING (Including Your Dollars), PPI God Candle, Lee Jae Myung Says "We Must Live
9.3/10TL;DR
A sharp rotation out of AI-linked stocks intensified after Anthropic disclosed slower-than-hoped revenue growth, underscoring how even enormous expansion can disappoint a market priced for near-exponential gains.
KEY POINTS
AI trade stumbles despite huge growth
A selloff spread across AI-related equities as investors reassessed revenue expectations for frontier model companies. CoreWeave was described as down roughly 50% over the past month, and the broader AI trade has weakened even as underlying demand remains high. The move reflected a gap between still-rapid business growth and valuations built on even faster assumptions.
Anthropic’s investor update resets expectations
The immediate catalyst was Anthropic’s latest revenue disclosure. Quarterly revenue was cited at $16.8 billion, implying an annualized run rate of about $67.2 billion. That figure is extraordinary by normal standards, but investors had been looking for a bigger upside surprise, with some expecting a path closer to $90 billion or even $100 billion-plus.
The problem is deceleration, not decline
The market reaction highlighted the difference between growth and acceleration. AI labs had been widely framed as businesses capable of roughly 10x annual revenue expansion, and any slippage from that pattern now carries outsized consequences for sentiment. Even a move from 10x to 9x growth can materially alter forward valuation models when stocks are priced for sustained exponential curves.
Distillation revenue loss proved smaller than feared
A key new data point was the share of revenue affected by model distillation, which was estimated at 5% to 7%. That was lower than some traders expected, countering more extreme claims that revenue would collapse once questionable usage was stripped out. Even so, at Anthropic’s scale, that percentage still represents billions of dollars of activity tied to unauthorized or policy-violating extraction of model outputs.
Distillation remains technically hard to police
Preventing distillation appears difficult because suspicious traffic can resemble legitimate use. Frontier labs face abuse routed through wrapper applications, enterprise tools and large account networks, making it hard to distinguish ordinary prompts from attempts to recreate model behavior. The challenge is magnified when demand is already exploding and usage is fragmented across many intermediaries.
Competition may be slowing runaway dominance
Another drag on expectations is customer behavior across big tech. Large companies have incentives to avoid dependence on any single model provider, creating what amounts to an anti-monopoly balancing force in AI procurement. As one lab gets too far ahead, rivals often receive new commercial support, limiting the odds of a clean winner-take-all outcome.
Institutional investors are rotating, not fleeing risk entirely
The tone around markets turned apocalyptic in places, but the actual move was more selective. The Nasdaq was described as down only about 0.3% to 0.4%, while the Dow was up close to 1%, suggesting rotation rather than wholesale liquidation. Capital has been moving out of the highest-beta AI names and into more traditional software, industrial and defensive exposures.
Prominent managers cut AI-heavy positions
Altimeter Capital founder Brad Gerstner said the firm had sold all of its CoreWeave position after reducing exposure over the prior three months. He said AI had overshot and was entering a consolidation period, while the fund shifted roughly 10% of its book into Meta and Tesla. The trade reflected preference for large platforms with AI upside but broader earnings support.
Inflation data added to the pressure
Markets also had to digest hotter producer-price data. One widely cited view labeled it the hottest PPI print in four years, reinforcing concerns that inflation and interest rates may stay elevated. Higher rates are especially painful for speculative growth trades and for heavily financed real-asset sectors such as development and infrastructure.
Outside tech, infrastructure and luxury offered a different lens
Alongside the market discussion, a construction technology company described growing demand for trenchless underground utility work as power, fiber and grid projects expand. At the same time, alternative inflation gauges for affluent consumers showed steep price increases in luxury goods and services, from Chanel bags to Hamptons homes and Virgin Galactic seats. Together, those signals pointed to an economy where capital spending remains active but pricing power is highly uneven.
CONCLUSION
The AI boom has not ended, but the market is becoming less willing to pay for perfection. With inflation still hot and growth expectations stretched, the next phase may reward companies that can show durability rather than just explosive headline expansion.
Ask a question