The Fed's first hike in three years drew a rally, but the debate moved to what lifts long yields. Refined-product tightness, not crude, is where the energy shock is priced. Compute rental rates keep rising even as neocloud financing risk grows, and mortgage rates above 7% brake housing.
Markets priced roughly three more Fed hikes through June 2027 after the first increase in three years, with sell-side forecasts split between October and December. Bassman calls the long-end move a credibility repricing, not inflation, while El-Erian says Fed credibility has not primarily driven higher long yields. Bianco counters that the Fed targets PCE, where core inflation is 3.3% and rising.
Saudi Arabia restored about half its East-West pipeline capacity after last week's drone attack, leaving Brent near $105 and WTI around $102, while two IRGC tanker attacks in Hormuz and an Iranian statement that the strait stays closed keep the supply-risk premium live. Goldman closed its diesel timespread call and recommended a long European gasoline position for mid-2027.
Oracle says GPUs up for renewal in Q1 were renewed or resold at a 20% premium, most four years or older; Nebius is raising listed rates for H100, H200, B200 and B300; and CoreWeave reported pricing up 25% in July. The offset is financing: CoreWeave proposed $3bn of convertible notes and its shares turned negative.
The average 30-year mortgage reached 7.24%, and pending home sales fell 4.7% year over year. Housing starts and permits fell while homebuilder confidence sat near COVID-era lows, and Lennar cut its full-year delivery target. Einhorn argues land-light financing costs roughly 3-4% of sales price in margin and gets worse when demand slows.
Goldman closed its diesel timespread call and recommended a long European gasoline position for mid-2027 because refiners shifting yield toward diesel are tightening gasoline; TheValueist relays that and argues refiners are the ultimate winners, with at least two more earnings cycles before the products market normalises.
Amazon agreed to buy up to $8bn of backup generators from Generac with a warrant for an equity stake, and Generac shares rose about 18%. Eaton's data-center revenue is running near +65% with orders +85%, and management raised 2026 organic growth guidance to 10% with 12% possible.
michaelsikand says his prior focus was mispriced AI bottlenecks, international names and private-market proxies, and that he now thinks the alpha is in cheap Mag 7 stocks about to reap returns on their capex, naming Meta, Amazon and Nvidia as his top three.
Ceresna recommends a January 2027 TLT 85/77 long strangle near $1.79 with January implied volatility still at the bottom of its one-year range after rebounding from about 10% to roughly 12%; Yardeni says he would not buy long-term bonds ahead of the BOJ decision but sees a 5% 10-year yield as a good return over 6-12 months.
Gerstner puts next-year data-center capacity at about 25GW, half of it for OpenAI and Anthropic, against SemiAnalysis's 43GW; Northwise models KEEL needing $1.74bn of additional common equity, rising to $2.36bn on a six-month tenant-acceptance delay, and jiahanjimliu ranks CoreWeave first, IREN second and Nebius last on incremental AI ARR per dollar of 2026 dilution.
Citi models DRAM price increases continuing through 2031 starting at +23.1% in 2027 and separately puts HBM growing 25x to $100bn by 2027; against that, Micron's latest quarter showed an 84.56% gross margin versus a prior best of 58.87% and a negative margin in fiscal 2023, and bubbleboi argues less DRAM per accelerator is inevitable.
CoreWeave proposed $3bn of convertible senior notes and its shares turned negative; an $800m NBIS convertible holder was cited selling roughly 15m shares; stockmarketnerd says every neocloud will need to dilute or lever up to fund projected buildouts; and KEEL's model needs $1.74bn of equity, rising to $2.36bn on a delay.
Subramanian warns that more than three short-end hikes raises financing costs for a Russell 2000 she says no longer behaves like a domestic cyclicals index, and the QQQ/IWM ratio is read as showing the small-cap rotation has already failed.
The AI Beneficiaries Index relative to the AI At Risk Index has fallen 46% from its late-2025 peak to near its lowest since early 2025, and Einhorn argues AI providers lack network-effect moats so users capture most of the value.