Warsh made hikes live while dropping forward guidance; AI usage is accelerating faster than financing quality; memory and optical scarcity remain real but time-limited; refined-product shortages preserve an inflation tail despite prospective Venezuelan supply.
Warsh reaffirmed the 2% target, said inflation needs clearer progress and rejected mechanical forward guidance. Markets lifted September and December hike odds, yet Mohamed El-Erian argues the near-60% September price is too high because expectations remain anchored. The regime change is less a guaranteed hike than a wider distribution of rate outcomes at every meeting.
AI usage is still accelerating: Warsh cited more than $100 billion of annualized token sales and an enterprise expert reported 13-14x token growth in six months. The counterweight is economics—compute prices are falling 70-80% annually, OpenAI's sequential costs reportedly rose three times faster than revenue, and new capacity is being financed with short-dated debt.
HBM and enterprise SSD supply is expected to stay tight through late 2027, while AAOI commentary says demand is not the problem—capacity is. HBM co-packaging is also back-weighting next-generation GPU deliveries. The disagreement is duration: Paul Kedrosky expects Asian fab investment to create a supply wave in early 2028, making execution and contract length more valuable than blanket semiconductor exposure.
Diesel and crack spreads remain tight as Russia withholds product and Novorossiysk loadings are expected to fall more than 50% month over month. Long-crack-spread positioning is already working in refiners. Prospective Chevron and Halliburton investment in Venezuela could add crude, but it does not immediately solve refinery-product shortages, leaving an inflation tail even if headline oil supply improves.
JulienTechInvest added six ISRG shares at $370 and sold all 40 UBER shares at $77, explicitly rotating from autonomous-ride commoditization risk into surgical robotics.
accounting_ds initiated MCD, citing the franchise cash-flow model and a 20x next-twelve-month valuation; this was a disclosed purchase, not a bullish watchlist mention.
Damnang initiated a small tactical SIVE position after the selloff while retaining a high-risk, high-beta assessment and emphasizing that the company still has much to prove.
grumpierbtday sold LPTH puts after judging the morning decline excessive; the trade is conditional exposure rather than a disclosed outright share purchase.
accounting_ds said more TOYO would be bought only below $4. Until that threshold trades, this remains a conditional idea rather than a fresh purchase.
ParadisLabs identified LITE and AAOI as preferred quality winners and intended allocations, but did not disclose a fill in the cited post.
seedy19tron opened an ABCL short targeting single digits and explicitly maintained an existing MRNA short.
kevinxu established bearish QQQ exposure through SQQQ, citing higher rate odds, low volatility and policy uncertainty rather than claiming the September hike is certain.