U.S. Treasury yields hold while Aussie yields retreat after 2011-era peak run up

The U.S. 10-year Treasury yield held at 5.230%, its highest since July 2007, while the 30-year yield remained near 5.556% as crude oil rose above $106 per barrel following President Trump's rejection of an Iranian ceasefire proposal involving the Strait of Hormuz. The RBA raised its cash rate 25bps to 4.60%, a 15-year high, while markets price a nearly 70% probability of another 25bp Fed hike in October amid persistent inflation concerns. Wednesday's PCE inflation data and Friday's nonfarm payrolls report are key near-term catalysts that could reinforce the higher-for-longer rates outlook and extend pressure on global bonds and equities.

Analysis

The relevant transmission is not simply higher discount rates: persistent energy-led inflation raises the probability that nominal growth remains firm while real consumption weakens. That combination is most damaging to long-duration, cash-burning software and discretionary equities, where valuation support depends on a declining terminal-rate assumption; it is relatively supportive of upstream energy, refinanced-later banks, and insurers reinvesting float at higher yields. AI infrastructure spending also becomes a two-sided risk: power, cooling, and financing costs can absorb a larger share of hyperscaler capex budgets, slowing the conversion of GPU demand into downstream software revenue.

The next 3-5 trading days are binary around inflation and labor data, but a sustained energy shock would take 1-3 months to appear in core-services inflation expectations and earnings revisions. The curve is already deeply restrictive in absolute terms, so outright duration shorts have unfavorable convexity if either data release disappoints or geopolitical risk resolves; the cleaner expression is to own inflation/energy beneficiaries against rate-sensitive consumers. A move lower in crude, easing inflation expectations, or a material deterioration in payrolls would quickly reverse the cyclical leg of this thesis.

The AI-training-halt angle is not actionable without confirmation of duration, compute commitments, and whether it reflects a safety pause versus a capacity or financing constraint. A broad AI selloff would likely create dispersion rather than invalidate semiconductor demand: power-management, grid, and data-center cooling suppliers retain better near-term backlog visibility than application-layer names. DJT has no evident operating linkage to rates, energy, or AI training spend; treat any correlation-driven move as noise rather than a macro position.

Key Decisions for Investors

• For the next 1-3 months, implement long XLE / short XLY in equal dollar risk rather than an outright Treasury short. The pair captures energy cash-flow upside and consumer margin/discretionary pressure; exit if crude falls below $90 for two consecutive weeks or inflation expectations roll over materially.
• Buy 2-3 month put spreads on ARKK or IGV after any post-data relief rally, targeting a 10-15% downside with defined premium at risk. The thesis is multiple compression in unprofitable and long-duration growth if policy expectations reprice higher; invalidate on a clear soft inflation print followed by a material decline in real yields.
• Maintain a watchlist rather than initiate an AI-specific short: monitor VRT, ETN, PWR and CEG against software proxies such as IGV. Go long the infrastructure basket versus IGV only if hyperscaler capex guidance remains intact while evidence confirms that the training interruption is isolated; missing data are the duration of the halt and associated compute-contract changes.
• Use October/November Treasury futures options for downside-rate hedging rather than adding naked duration shorts: buy modest TY put spreads or pay fixed selectively after strong data. Take profits if the 10-year yield rises another 25-35bp, since policy/geopolitical de-escalation creates significant reversal risk from elevated yield levels.