Iran Struck Back. Chips Rallied Anyway.

GOOD MORNING, Thursday looked like a geopolitical risk-off day — until the chip trade took over. Iran retaliated against U.S. targets across the Gulf, tanker traffic through Hormuz nearly stopped and investors were forced to rethink the inflation risk from another energy shock. Yet the Nasdaq still jumped 1.3% as semiconductors rallied 3.1%. Micron announced more than $250 billion of U.S. investment and Meta disclosed plans to put its own AI chip into production in September. The market’s message was unusually clear: geopolitical risk still matters, but AI capital spending remains powerful enough to overpower it — at least for a day.

MARKETS | TLDR

The Nasdaq rallied despite renewed U.S.-Iran fighting: The Nasdaq gained 1.30%, the S&P 500 rose 0.81% and the Dow added 0.27% as chip stocks led a broad rebound.
Semiconductors were the engine: The Philadelphia Semiconductor Index jumped 3.06%, with Micron up 4.5%, Applied Materials up 3.2% and SanDisk up 7.6%.
Oil fell even as Hormuz traffic nearly stopped: Brent dropped to $76.30 and WTI to $72.08 as concerns about demand and hopes for eventual de-escalation outweighed immediate supply fears.

MICRON

AI Spending Just Turned Into a $250 Billion Factory Bet

Micron said Thursday it would increase planned U.S. manufacturing and technology investment to more than $250 billion through 2035, driven by surging demand for memory in AI systems. The company also poured first concrete at its Clay, New York fabrication site more than a quarter ahead of schedule and said the expanded buildout supports a long-term goal of producing roughly 40% of its DRAM in the United States.

The scale matters because memory is becoming one of the clearest bottlenecks in the AI stack. Accelerators such as Nvidia GPUs cannot deliver their full performance without large amounts of high-bandwidth memory and advanced storage. Micron’s decision to accelerate capacity is therefore not a speculative bet on consumer electronics; it is a direct response to the compute intensity of AI workloads.

The market rewarded the signal. Micron rose 4.5% and the broader chip index gained 3.1%. But the long-term tension is worth watching: scarcity creates pricing power, while $250 billion of new investment is designed to reduce scarcity. The strongest AI suppliers are now spending aggressively to defend the very economics that made them winners.

Meta Wants to Own More of the AI Cost Curve

Meta plans to put a custom AI chip codenamed “Iris” into production in September, according to an internal memo reviewed by Reuters. The chip is part of Meta’s MTIA program and will be manufactured by TSMC after a six-week testing cycle. The company expects to deploy about 7 gigawatts of compute infrastructure in 2026 and aims to roughly double that capacity to 14 GW next year.

The strategic logic is cost control. Meta is still spending heavily with external suppliers — including Nvidia and AMD — but every workload it can shift onto a chip designed specifically for its own software reduces dependence on merchant accelerators. Meta expects AI infrastructure spending of as much as $145 billion this year, so even modest efficiency gains can become economically meaningful at that scale.

The broader read-through is that hyperscalers are moving from simply buying AI infrastructure to vertically integrating it. Meta wants a new custom chip every six months through 2027. Google already has TPUs, Amazon has Trainium and Microsoft has Maia. The AI capex boom is still good for semiconductor demand, but over time it may shift more economics from general-purpose suppliers toward the companies controlling workloads, software and custom silicon.

HEADLINES

Only two tankers crossed the Strait of Hormuz early Thursday: Traffic fell close to a standstill after renewed attacks, while a Qatari LNG tanker was hit by a projectile — showing that physical energy risk remained far worse than Thursday’s oil-price decline implied.
Iran struck U.S. military targets in Kuwait, Qatar and Bahrain: The retaliation followed fresh U.S. strikes on Iran and strained a fragile truce, keeping geopolitical risk firmly inside the market’s inflation outlook.
Micron added another $3 billion for U.S. semiconductor supply chains: The company will support domestic wafer and materials capacity, including $500 million of financing for GlobalWafers’ Texas facility, showing how AI investment is spreading into upstream suppliers.
The U.S. labor market remained resilient: Lower weekly jobless claims helped reinforce the view that the economy is still strong enough to absorb geopolitical volatility — but also strong enough to keep the Fed cautious.
Technology is expected to drive another strong earnings quarter: Analysts expect roughly 24% year-over-year S&P 500 earnings growth in the second quarter, leaving AI-heavy tech responsible for much of the market’s fundamental support.

UPCOMING

U.S. CPI — July 14: June consumer inflation is the next major test of whether the spring energy shock is fading fast enough to keep the Fed patient.
Major U.S. bank earnings — July 14–15: JPMorgan, Goldman Sachs, Bank of America, Citigroup and other lenders will provide a read on trading, dealmaking, lending and credit quality.
U.S. PPI — July 15: Producer prices will show whether cost pressure is cooling upstream or whether energy and tariff effects are still working through business margins.
TSMC earnings — July 16: The world’s dominant advanced-chip foundry will provide the clearest near-term test of whether AI demand remains strong enough to justify the semiconductor sector’s extraordinary capital cycle.

DEEP INSIGTHS

Oil Tanker Traffic Through Hormuz at Near Standstill

Read this for the physical-market reality behind Thursday’s crude move. Before the war, the strait handled roughly 20% of global oil supply and 125–140 vessels a day. Early Thursday, only two tankers crossed. That divergence — oil prices falling while actual shipping conditions worsen — is exactly the kind of disconnect that can reverse abruptly if the disruption persists.

Meta to Put AI Chip Into Production in September

This is useful for understanding how hyperscaler economics are changing. Meta’s plan is not simply to add more compute, but to control more of the silicon layer itself. If custom accelerators reduce cost per workload, the companies spending the most on AI infrastructure can eventually capture more of the margin now earned by external chip suppliers.