Hormuz Stays Shut|30-Year Yield at the Maginot Line

The 5% Line

The 30-year Treasury yield crossed 5% on Monday and held there Tuesday — the third time in less than three years, but unlike the prior two, this time the cause is not a Fed meeting or an inflation print. It is a chokepoint 7,000 miles away. The Strait of Hormuz has been effectively closed since late February, cutting roughly 14.5 million barrels of daily supply from global markets — the largest oil disruption the International Energy Agency has ever recorded. Brent crude is trading near $108 a barrel, up more than 50% since the conflict began. And that number is now printing itself into the long end of the U.S. yield curve.

The mechanism is straightforward but underappreciated. Higher oil prices feed into headline CPI with a lag of roughly six to eight weeks. The Federal Reserve, which entered 2026 expecting to cut rates by June, watched three FOMC members dissent at the most recent meeting to remove the easing bias entirely. Oxford Economics moved its next cut forecast from June to December this week. Bank of America's Michael Hartnett, who called the 5% threshold the bond market's Maginot Line, noted that every prior time a major asset bubble ended — Japanese bonds in 1989, U.S. Treasuries in 1999, Chinese yields in 2007 — it was a rapid move above that line that opened the door. The U.S. is now spending $1.22 trillion annually on debt service, equivalent to 4.18% of GDP, the highest since the early 1990s. Each day the Strait stays closed extends the higher-for-longer regime further out on the curve.

Treasury Secretary Scott Bessent went on television Monday and told the market the crisis was effectively over. He said the U.S. has absolute control of the Strait and that more than 150 crude carriers can move out immediately. The market did not move. Brent held near $108. Ship operators said it takes both sides to unblock — and Iran has not agreed to anything. Project Freedom, the Navy escort operation, has moved exactly two U.S.-flagged merchant ships through the Strait against a backlog of more than 150. The gap between Bessent's words and what oil is doing is the signal worth tracking. If the 5% level breaks cleanly above the October 2023 high of 5.17%, it would put long yields in territory not seen in close to two decades — and that is the condition Hartnett says begins a disorderly repricing in equities.

The AI Capex Lock-In

The Hormuz shock is landing on an equity market that has been pricing a very different story — an accelerating AI infrastructure cycle that is producing some of the strongest earnings beats in a decade. The two forces are not unrelated. They are pricing the same economy from opposite sides.

Anthropic reportedly committed to spending $200 billion on Google's cloud services and chips over five years. That single contract is large enough to represent more than 40% of the revenue backlog Alphabet disclosed to investors last week. Combined with OpenAI's cloud commitments, AI lab contracts now account for more than half of the estimated $2 trillion in forward cloud backlogs across the major hyperscalers. That is not demand speculation — that is contracted revenue already on the books. The practical consequence arrived Tuesday night when AMD reported first-quarter results. Data center revenue jumped 57% year over year to $5.8 billion, beating expectations by $200 million. Total revenue grew 38% to $10.25 billion. Adjusted EPS came in at $1.37, ahead of the $1.29 consensus. For Q2, AMD guided revenue to $11.2 billion at the midpoint — 7% above what Wall Street had modeled. CEO Lisa Su described it as demand from inferencing and agentic AI driving increasing need for high-performance CPUs and accelerators, and noted that customer forecasts are running ahead of management's initial expectations.

Qualcomm added a different dimension the same day. The company disclosed a win with a major hyperscaler for custom data center AI silicon, a business line that barely existed two years ago. It paired the announcement with a $20 billion share repurchase authorization. Q2 net income rose to $7.37 billion from $2.81 billion a year earlier, driven by the mix shift toward AI infrastructure and automotive. The counter-signal worth tracking: ING's Warren Patterson noted that the oil market is tightening every day the Hormuz blockade holds, and the same energy cost passthrough that is pressing the Fed is also showing up in AMD's client segment, where PC shipments are expected to fall 11.3% in 2026 on a global memory shortage. The strong AI capex story and the deteriorating consumer hardware story are running simultaneously inside the same earnings reports. The condition that determines which dominates: whether the hyperscaler commitments locked in by Anthropic and OpenAI hold even as energy-driven inflation slows broader IT spending.

Power as the Trade

If the AI capex cycle is real and durable, the most direct evidence is not in the chip stocks. It is in the industrial machinery delivering the power and infrastructure those data centers require. That is the chain connecting the two previous stories — and it is running ahead of both.

Caterpillar is up 53% year to date. GE Vernova is up 65%. Both stocks hit fresh highs Tuesday. Caterpillar's Q1 power generation revenue jumped 41% to $2.82 billion on demand for large reciprocating engines and turbines feeding data center sites. Total revenue grew 22% to $17.41 billion. CEO Joe Creed cited a record order backlog as the forward foundation. GE Vernova's Q1 orders surged 71% organically to $18.3 billion, and its Electrification segment booked $2.4 billion in data center equipment orders in Q1 alone — more than the entire 2025 fiscal year combined. CEO Scott Strazik raised 2026 free cash flow guidance to $6.5 billion to $7.5 billion. The Bureau of Economic Analysis confirmed that manufacturing sector profits hit $759.6 billion in Q4 2025, with durable goods up 13% year over year, validating the structural tailwind. These are not speculative AI plays. They are capital equipment companies with contracted backlogs from data center operators who have already committed the money.

The friction is real but manageable so far. Caterpillar absorbed $710 million in tariff-related manufacturing costs in Q1 and still expanded margins. The risk that shuts this down is not tariffs — it is an abrupt cooling in hyperscaler capex guidance. The Anthropic-Google deal is the kind of contract that, if it holds, keeps the backlog intact for years. If AI model economics deteriorate and labs pull back on compute commitments, Caterpillar and GE Vernova's order books are the first visible indicator — orders lead revenue by two to four quarters. Watch GE Vernova's Electrification segment bookings in the Q2 report. The company guided for 110 gigawatts of combined gas turbine backlog and slot reservations by year-end. If that number comes in below 100 in the Q2 update, the infrastructure thesis loses its floor. The weight of evidence today — $18.3 billion in orders, $2.4 billion in data center bookings in a single quarter — points toward the AI power build remaining intact. But the same 30-year yield at 5% that is pressuring equities broadly will eventually price the financing cost of that infrastructure. The Hormuz-driven inflation that pushed yields here is the condition that could tighten the very capital those build-outs depend on. The two stories are on a collision course, and the Q2 earnings season will tell which one is moving faster.