The Saudi East–West pipeline is particularly important because it was one of the major routes allowing crude to avoid the disrupted Strait of Hormuz. Its temporary shutdown adds another potential bottleneck at precisely the wrong time. The Houthi advance creates a second problem: Bab el-Mandeb controls access between the Red Sea and Arabian Sea. Reuters describes Iran-aligned Houthi gains as creating a new challenge for global shipping while Hormuz is already destabilised. Energy/refining/nuclear/defence: likely relative winners. Airlines/logistics/chemicals/discretionary/Asian energy importers: losers After sticky CPI and the energy shock, the question is no longer simply whether the Fed changes rates; it is the balance between energy-driven inflation + resilient labour demand + weak
The biggest change overnight is a sharper split between AI fundamentals and the broader market. The S&P 500 fell 0.58%, Dow 1.18% and Nasdaq 0.32% as Brent approached US$100 and Treasury yields stayed near 4.8%. Yet semiconductor names tied to custom AI infrastructure held up better, helped by a major new Amazon–Qualcomm agreement. Micron is already shipping HBM4 in volume for Nvidia Vera Rubin, delivering >2.8 TB/s bandwidth and >20% greater power efficiency than HBM3E. SK hynix meanwhile approved about ₩54T of new fab investment specifically to address growing AI-memory demand. HBM remains the strongest fundamental layer, but after recent sharp memory rallies coupled with imminent interest rate increases, better entry points key rather than Supercycle over.
$Microsoft(MSFT)$ monetisation shown via Q4 Azure growth reached 43%, and management expects about 45% constant-currency growth next quarter. Importantly, Microsoft says customer demand still exceeds available capacity and that new capacity brought online in Q4 was rapidly monetised. Commercial RPO reached US$678 billion, +84% YoY, while roughly 30% is expected to convert into revenue within 12 months. Microsoft also reported more than 30 million paid Microsoft 365 Copilot seats. CapEx was US$41 billion in the quarter. Accounting changes now put CY2026 reported CapEx near US$175 billion, but Microsoft explicitly said the change is mostly lease classification rather than lower underlying infrastructure investment. FY27 CapEx is still expected t
The first half of 2026 had already priced enormous growth into AI infrastructure stocks. Memory, semiconductors, equipment and related names had risen dramatically. Morningstar notes that some AI hardware stocks had appreciated 2x, 3x or even 4x before earnings. Consequently, merely reporting “excellent” numbers wasn’t enough anymore — companies effectively needed to beat already extraordinary expectations and raise the outlook.
Yesterday’s biggest positive was Treasury intervention in the bond market. The US Treasury announced that it will double long-duration bond buybacks to at least US$4 billion per operation from September through early November. The move followed the 30-year Treasury yield reaching nearly 5.34%, its highest in almost two decades. That is meaningful relief, but I would not interpret it as the end of the bond problem. The underlying issues—US fiscal deficits, inflation and enormous AI infrastructure financing requirements—remain unresolved. They showed that “many” policymakers believe higher rates may ultimately be required if inflation does not continue falling, while three policymakers had already voted for a 25 bp hike at the July meeting. The current policy rate remains 3.50%–3.75%.
The US Treasury announced that it will double long-duration bond buybacks to at least US$4 billion per operation from September through early November. The move followed the 30-year Treasury yield reaching nearly 5.34%, its highest in almost two decades. The result: 10-year yield → ~4.65% 30-year yield → ~5.20% S&P 500 → +0.21% Nasdaq → +0.16% Dow → +0.22%. That is meaningful relief, but I would not interpret it as the end of the bond problem. The underlying issues—US fiscal deficits, inflation and enormous AI infrastructure financing requirements—remain unresolved. 🔴 Fed — more hawkish than the market hoped The July Fed minutes were important. They showed that “many” policymakers believe higher rates may ultimately be required if inflation does not continue falling, while three policy
1. CPI delivered the outcome we wanted July US CPI rose just 0.1% month-on-month and 3.4% year-on-year, down from 3.5% in June. Core CPI rose 0.2% monthly and eased to 2.5% year-on-year. This is not enough to declare inflation defeated, but it materially reduces the urgency for another Fed hike. Markets now place roughly a 60% probability on no September rate change, versus approximately 50-50 before CPI. That is a meaningful positive change from yesterday. My concern shifts away from immediate Fed tightening toward two longer-term issues: energy inflation + US bond supply. #1 opportunity — AI infrastructure just received another validation Cisco’s results are important. Fiscal Q4 revenue rose to US$17.25 billion from US$14.67 billion, while net income jumped to US$3.86 billion. More impor
There is no evidence yet that hyperscaler infrastructure spending is slowing materially. The better investment opportunity continues to move toward the physical AI infrastructure chain: REITs — today’s CPI matters considerably to this sector Soft CPI → lower Treasury yields → positive REIT catalyst. Hot CPI + Brent approaching $90 → higher yields → negative REIT catalyst.