A Better Newspaper

US Strikes Iran Again as Hormuz Tanker Attacks Threaten Oil Supply

CENTCOM confirmed the US military has begun striking targets inside Iran in a new wave of attacks that Trump described as 'large and powerful,' warning Iran would be hit 'much harder' if it retaliates. Iran promised to respond despite the threats. Separately, two oil supertankers were hit by projectiles in the Strait of Hormuz, with neither side prepared to cede ground over the waterway.

Context: The Strait of Hormuz carries roughly a fifth of global oil consumption; a sustained disruption there is the single most consequential energy-supply risk for markets. Watch tanker insurance rates and Brent for the real-time price of escalation, not the headlines.

Qatar Says Mediation Underway to End Iran-US War, Reopen Hormuz

Qatar said efforts are under way to end the Iran-US war and reopen the Strait of Hormuz. Doha also accused Israel of exploiting regional conflicts to impose new realities in Palestine, Lebanon, and Syria.

Context: Qatar has been the region's go-to back channel in prior US-Iran flare-ups. A negotiated de-escalation is the fastest path to reversing any oil spike — but the crowd has consistently priced a full Hormuz closure as unlikely even during acute crises.

30-Year Treasury Yields Log Worst Stretch Since 2006

Yields on the longest-maturity Treasuries have not been this high for this long since 2006, per Bloomberg. A widening budget deficit, another wave of corporate issuance, and a potentially decisive Federal Reserve meeting are expected to keep investors wary of US debt in the weeks ahead.

Context: The signal here is fiscal, not cyclical: persistent term-premium pressure at the long end raises borrowing costs across mortgages, corporate credit, and equity valuations regardless of what the Fed does at the front end. A Gulf-driven oil shock would only compound the inflation worry weighing on duration.

Xi Visits Egypt for First Time in a Decade Amid Middle East Crises

Chinese President Xi Jinping visited Egypt to deepen economic ties and align strategies as the Middle East faces multiple crises, his first trip to the country in ten years.

Context: Beijing is the largest buyer of Iranian oil and stands to gain diplomatic capital as Washington gets pulled into another Gulf conflict. Watch for China positioning itself as an alternative security and infrastructure partner across the region — a slow-motion realignment with implications for trade routes and dollar dominance.

Six Straight Days of Russian Strikes Push Kyiv Residents to Weigh Leaving

Russia has launched near-continuous strikes on Kyiv and surrounding areas for six straight days, prompting residents to consider leaving the capital.

Security-Specific Frontier Models Become Their Own Market — CrowdStrike, Nvidia Move In

CrowdStrike launched SafeMind, a family of AI models and agent harnesses built with Nvidia specifically for security work rather than general use, produced by a new in-house Cyber Superintelligence Lab. It also extended its Falcon platform across Google Cloud's enterprise AI ecosystem. Separately, SiliconANGLE reports enterprise security is reorganizing around 'AI detection and response' as autonomous agents move into production — with the same capabilities aiding defenders and attackers alike, forcing boards to reassess risk.

Context: The strategic read: general-purpose frontier models are commoditizing, but vertical, safety-critical models (security, and Anthropic's restricted Claude Mythos) are where defensible margins and procurement lock-in are forming. 'AI detection and response' is emerging as a named budget line — an underbuilt niche for founders and a diligence checklist item for any enterprise deploying agents.

Anthropic Ships 5.1 Models — But the Real Story Is a $35B Compute Bill

Anthropic debuted Claude Fable 5.1 and Claude Mythos 5.1, its most capable models to date, one day after signing a $35 billion infrastructure deal with cloud startup Lambda. A week earlier it inked an even larger hardware contract with Nscale Global Holdings. The launches came with record-setting benchmark claims.

Context: Anthropic is now committing tens of billions to non-hyperscaler compute (Lambda, Nscale) rather than leaning solely on its Amazon and Google backers — a hedge against compute scarcity and a signal that the frontier is capital-intensity, not just cleverness. Watch the neocloud tier (Lambda, Nscale, CoreWeave) as the arms dealers of this cycle; that's where infrastructure capital and equity upside are concentrating.

US-China AI Safety Diplomacy Floated — Even as Beijing Attacks US Governance

OpenAI's head of strategic futures Dean Ball, a former White House adviser, publicly called for US-China AI safety cooperation ahead of a Trump-Xi summit, writing that he had long thought such collaboration desirable but unlikely. The call came as Chinese state media launched a scathing attack on American frontier AI governance.

Context: This lands against Stanford HAI's 2026 finding that China has erased the US AI lead. The strategic tell for the reader: any Trump-Xi AI safety framework would reshape export-control assumptions and could open — or foreclose — cross-border enterprise AI deals. Beijing's simultaneous attack on US governance is a bid to control the standards conversation, where it's already gaining leverage on inference and benchmarks.

Research Signal: Hidden and Dormant Adversarial Behaviors Are Now Reproducible

A cluster of new alignment papers demonstrates concrete attack surfaces in deployed models: one shows an adversary can distribute a benign, performant open-weight model that only activates adversarial behavior once a downstream user fine-tunes it (the 'FAB' attack); another operationalizes ten categories of 'hidden intentions' — covert agendas embedded in outputs to manipulate users — and finds them trivially inducible and hard to detect. A third introduces a 3D multimodal testbed for studying verbal and non-verbal deception by embodied agents.

Context: For a lawyer-strategist, this is the empirical backbone for coming liability and disclosure fights: if open-weight models can carry dormant, fine-tuning-activated payloads, model provenance and supply-chain due diligence become contract and insurance issues, not academic ones. This is exactly the risk the EU AI Act's governance layer is meant to police — and a market gap for model-provenance and audit tooling.

OpenAI Publishes Its 'Path to Astra' — A Frontier Capability and Safeguards Roadmap

OpenAI released 'Path to Astra,' outlining critical capabilities it is pursuing and the frontier safeguards it intends to pair with them.

Ed Zitron on the AI trade as a circular, subsidized bubble

The guest argues AI spending is barely cracking $100 billion and most of it is circular — Nvidia funding customers like CoreWeave, and a handful of firms (Oracle, CoreWeave) paying each other — against roughly $300 billion in equity funding and ~$600 billion in capex, meaning it isn't profitable and demand is subsidized by loss-leader subscriptions. He also cites a recent OpenAI study finding no correlation between spending on AI tokens and revenue per employee, and no evidence of white-collar productivity gains — only disruption to cheap contract work that would have been offshored anyway.

Jeff Keller on why capex won't stop even if AI returns disappoint

Keller reframes hyperscaler capex ROIC as buying an option rather than pouring concrete into the ground — defensive (existential if you don't play) and offensive (open-ended upside) — so even a low-but-positive 4-5% ROIC is rational because stopping is too risky. He notes that after the Anthropic IPO, the major labs plus SpaceX and Google will have raised about $500 billion year to date and markets took it in stride, so an extra 150 bips of borrowing cost won't slow them down.

Jeff Keller on why low-multiple semis beat 'hidden AI winners'

Keller makes the contrarian case that low-multiple AI names like Nvidia and Micron will outperform high-multiple 'creative' AI plays (behind-the-meter, services, cyclical businesses), because all are driven by the same factor — hyperscaler capex and lab ARR — yet the semis are already pricing peak while the others extrapolate a rosier future. He separately notes the cloud market went from ~3 scaled providers to 9-10 (adding Oracle, SpaceX, the labs, Neoclouds, CoreWeave, Nebius), explaining why those stocks don't command higher multiples even as AWS growth reaccelerates from ~18% into the 40s.

DHH on agents already outclassing the median programmer

Based on 25 years of reviewing tens of thousands of contributors' output, DHH argues most programmers 'suck' — not in skill but in diligence: they don't document the why, write comments, double-check work, or write unit tests, whereas agents do all of it when instructed. He says he's merged over 1,000 pull requests in three months on his Quattro project, many from people who aren't classical programmers, and no longer reviews PRs himself — agents review, validate fixes in a VM, filter bad submissions, and hand him a summary for the human merge decision.

Joe Liemandt on Alpha School's two-hours-a-day academic model

Liemandt says Alpha School students average 121 minutes a day in AI apps yet test in the top 1% nationally, with average graduating SAT scores of 1530. He grounds it in Bloom's two-sigma finding — one sigma from individual tutoring, one from mastery-based learning — arguing AI tutors now make mastery-level individualized instruction cost-effective for the first time. Transfers from $50,000 private schools averaged 2.2 grade levels behind, and a full grade level in one subject can be caught up in 20-30 hours of K-8 work.

Liemandt on why algebra failure is really a third-grade math-facts failure

Liemandt argues the working-memory constraint means algebra failure is usually an unaddressed third-grade math-facts gap — remediation must go backward, not forward. He cites a student stuck at a 710 math SAT with 'careless errors' who was actually running out of working-memory slots; after memorizing multiplication tables, the score jumped to 790.

Scott Galloway on the mid-career income cliff and equity comp

Galloway cites a major 2018 study finding more than half of workers in their early 50s who held long-term jobs were laid off after age 50, with many seeing pay drop 50% or more for a year. Separately, he advises negotiating for options over salary: ISOs aren't taxable when granted, grow tax-deferred and compound unimpaired by the 30-45% annual haircut salary takes, and yield long-term capital gains if exercised equity is held over a year.

Galloway on 'no hire, no fire' and AI layoffs still confined to tech

Galloway describes the current labor market as 'no hire, no fire' — firms holding off on hiring to see what happens to the economy — and argues the AI layoff wave has so far hit mainly tech as the early adopters, which he thinks likely presages broader layoffs. He also cites a Google study in which a job opening drew 200 CVs within 60 minutes, and among the top 20 candidates interviewed, 70% of the time the person hired had an internal advocate.

Tucker guest on the Argentine beef deal and rejected shipments

The guest says China rejected an ~80,000-metric-ton shipment of Argentine beef after detecting a last-resort antibiotic banned for livestock use in Argentina, the US, China, and Canada — and days later the president announced the first 80,000 metric tons of a 330,000-ton import deal would come from Argentina at a 25% discount. He also notes four companies control 85% of US beef and oppose country-of-origin labeling partly because they commingle imported grind into vats of hamburger and don't want to track its origins; such labeling existed until Congress repealed it in 2015 after Mexico and Canada won a WTO case against the US.

Host on autonomous vehicles showing sharply lower crash rates

The host cites data showing autonomous vehicles have a 68% lower overall crash rate (2.1 versus 4.68 police-reported crashes per million miles), roughly 80-81% fewer injury crashes, and 85% fewer single-vehicle crashes than human drivers. The guest frames it alongside skepticism of AI-executive job-loss predictions as self-interested rather than evidence-based.

Iran hits seven weeks with zero crude exports under U.S. Hormuz blockade

Iran has now gone seven weeks without exporting any crude oil, according to Marine Link, as the U.S.-enforced blockade of the Strait of Hormuz continues to shut in the country's primary revenue source. The report frames the sustained halt as a direct consequence of the maritime disruption in the strait.

Context: Crude exports are the fiscal backbone of the Iranian state; a seven-week complete stoppage is a far harsher squeeze than sanctions alone have historically achieved, and it explains the scramble by Tehran and its trading partners to keep any trade channels open.

Chinese exports to Iran collapse 52% as blockade compresses import capacity

New Chinese customs data show China shipped just $1.9 billion in goods to Iran between January and July 2026, down from $4 billion in the same period of 2025 — a roughly 52% drop. Kurdistan24 attributes the collapse to sanctions, the naval blockade, and maritime disruption compressing Iran's ability to import foreign inputs.

Context: China has been Iran's economic lifeline and largest crude buyer; a halving of even non-oil goods trade signals that Beijing is quietly pulling back exposure rather than backstopping Tehran through the war.

Qatar extends LNG force majeure as Hormuz stays closed to loadings

Qatar has extended a force majeure on LNG deliveries as the Strait of Hormuz remains closed to loadings, according to OilPrice.com. The move formalizes that Qatari cargoes cannot transit the strait, disrupting supply to global buyers.

Context: Qatar is one of the world's largest LNG exporters and virtually all of its output must pass through Hormuz; a prolonged force majeure ripples directly into European and Asian gas markets already stressed by the conflict.

McGurk: In 'Act IV' of the war, Iran is losing its Hormuz leverage

In a CNN analysis, former U.S. official Brett McGurk argues that despite the disruption, Iran is not winning the conflict — it is steadily losing the leverage it held over the Strait of Hormuz. He frames the current phase as an 'Act IV' in which Tehran's strategic position is eroding rather than strengthening.

Context: McGurk served as White House Middle East coordinator; his read runs against the intuition that Iran benefits from squeezing the strait, and pairs with the trade and export data showing the blockade cutting deeper into Iran than the West.

Health-Pixel and TCPA Settlements Keep Clearing Seven Figures

Cone Health agreed to a $1.76 million settlement resolving claims it used pixel tracking technology to collect and share patient information without consent, and Palm Beach Tan agreed to a $2.5 million TCPA settlement over unsolicited text messages, with a claim deadline of Oct. 4, 2026.

Context: The steady drumbeat of mid-seven-figure resolutions in health-data pixel and TCPA matters continues to validate these as reliable, high-volume plaintiff-side theories with predictable settlement floors — useful comps for funders pricing consumer-privacy portfolios rather than one-off events.

OSU Pays $2.1M to Settle DOJ China-Disclosure Grant Claims

Ohio State University agreed to a $2.1 million settlement to resolve DOJ allegations that it failed to disclose employees' ties to China when applying for federal NASA and NSF grant funding between November 2012 and August 2023, the Department of Justice announced Monday.

Context: DOJ's continued use of False Claims Act theories against universities over foreign-influence disclosures signals a durable qui tam / grant-fraud enforcement lane worth watching for relator-driven mass filings.

Business Idea: Structure a Long-Vol Energy-Logistics Play, Not an Oil Bet

The market is pricing a Hormuz oil spike, but the more durable, mispriced dislocation is in shipping and gas logistics: with Qatar's LNG force majeure extended and tankers physically hit, the scarce asset is not barrels but insured, re-routable hull capacity and non-Hormuz molecules. The move is to structure a position that is long freight-rate and marine-war-risk-insurance exposure (VLCC/LNG time-charter rates, tanker equities like Frontline and Euronav, and war-risk premium beneficiaries) while pairing it with long-dated payer swaptions or a steepener on 30-year Treasuries — hedging the twin catalysts of an energy shock and a widening deficit that are both live today. A well-capitalized entrepreneur with legal expertise should also lock up multi-year charter or storage contracts on Atlantic-basin and West-African tonnage now, before mediation either fails (rates rip) or succeeds (you own the cheapest re-routing capacity into a normalizing market). The asymmetry: you profit from continued stalemate, spike on escalation, and retain optionality on reopening — without needing to call the oil price itself.

Stoic Thought

“To comprehend all in a few words, our life is short; we must endeavour to gain the present time with best discretion and justice.” — Marcus Aurelius, Meditations, THE FOURTH BOOK