Visit https://ai.22vresearch.com/
Contact Mwhaling@22vresearch.com
In this week's video, I explain why I think the adjustment to the AI midcycle slowdown is ending. The second-derivative deceleration is now consensus, positioning has been cleansed, and the bears are coming out of the woodwork, historically a constructive sign while token demand continues. Tech momentum realized volatility hit 87, retail's 2x levered products gave back 62%, quant managers surrendered a third of their year, and 87% of S&P 500 semiconductors registered oversold. That's a speed crash inside a structural bull market built on token demand while Samsung's 10-day rate of change selloff on good news was the largest since COVID and Lehman.
The valuation setup is striking: Samsung trades near a 4 PE on 2026 estimates with operating profit potentially exceeding $217 billion, more than its cumulative profit over the past 40 years. Memory shortages are now projected beyond 2030 while capacity grows only 20–30% annually against doubling demand. This is exponential demand versus linear supplyhe , the core of the short-abundance, long-scarcity framework.
The next theme is consumer agents, a bigger token catalyst than coding agents, with Goldman estimating agents could drive token consumption up roughly 24x. And the crypto shift is now unavoidable: Scott Bessent's speech and Mohamed El-Erian's op-ed reframe digital assets, stablecoins, and tokenization as US financial statecraft. Traditional investors can no longer have no view.
Timestamps
• (00:00–02:55) Setup: bubbles, parabolas, and speed crashes; the midcycle slowdown is ending; a reminder of why this is the AI Macro Nexus, AI is now the dominant macro force of the cycle.
• (03:15–06:32) Sentiment and technical cleanse: tech momentum 60-day realized vol at 87; a structural bull market built entirely on token demand; still early despite two years of bubble calls.
• (06:32–11:08) The memory speed crash: Samsung's 19-fold profit surge, a 21% 10-day rate of change (largest since COVID and Lehman), 4 PE valuations, and why "short abundance, long scarcity" remains the framework.
• (11:34–15:30) Positioning flush: quant long/short managers' worst stretch since December 2023, retail 2x levered products down 62%, 87% of S&P semis oversold historically strong entry signals.
• (15:30–20:58) Memory reality check: shortages now projected beyond 2030, Musk's terafab, capacity growing 20–30% vs. demand doubling, why the fracking comparison fails against exponential digital demand.
• (20:58–22:38) Concentrated basket update: 50-day retest, RSI reset, Nvidia holding the 200-day with a MACD buy signal at its lowest valuation in a decade.
• (22:38–28:27) The token expenditure index, demystified: the coding-agent capability shock, enterprise FOMO, bill shock and Dylan Patel's spend going from $100K to $11 million annualized in six months for a 90-person firm.
• (28:27–34:40) Consumer agents, the next major theme: Meta's CTO on why consumer AI has been slow (product layer, not model weakness), and Apple closing at all-time highs despite raising prices 20% on memory costs.
• (34:40–40:05) The capacity myth: Microsoft, Google, Amazon, and Oracle backlogs totaling nearly $2 trillion; SemiAnalysis capex forecast at $11.1 trillion versus Goldman's $7.6 trillion.
• (40:27–51:21) Crypto's Bessent moment: digital assets, stablecoins, and tokenization reframed as US financial statecraft; the Andreessen 2014 parallel; the 40-name crypto index; Bitcoin absorbing Saylor's sale and closing higher.
• (51:36–53:54) The Fed: why July is the only month a hike makes sense, why one-and-done at 25bps is the risk case, and why no move likely means tailwinds for Bitcoin.