The battle lines of Iran and Ukraine are intersecting. In the same week that Russia equipped Iranian drones with Starlink, Ukraine destroyed 40% of Russia's oil export capacity. Capital markets have finally grasped it: these are not two isolated conflicts, but a merging global risk network.
1| Two Wars Converge: Ukraine Bombs Russian Oil Ports, Russia Gives Iran Starlink
Ukraine launched consecutive drone strikes on three major Russian oil export ports this week. Primorsk on March 23rd, Ust-Luga on the 25th, and the Kirishi refinery on the 26th. Reuters calculates that about 2 million barrels per day of export capacity were taken offline, accounting for 40% of Russia's total export capacity, "the most severe oil supply disruption in modern Russian history."
In the same week, Fortune reported that Russia is providing Iran with upgraded Shahed drones, featuring new AI computing platforms, jet engines, Starlink communication capabilities, and anti-jamming equipment. Western intelligence also shows Russia is providing Iran with real-time satellite positioning of US military assets. On the other side, Ukraine signed a security agreement with Saudi Arabia to share counter-drone technology, and Zelenskyy made secret visits to the UAE and Qatar.
University of Pittsburgh political science professor Spaniel says "we are not at a real world war yet," but the warring parties, weapons supply chains, and intelligence networks of the two conflicts are already intersecting. Washington is still considering diverting part of its military aid to the Middle East. (Continuation of yesterday's report)
(Sources: Fortune / Reuters / Moscow Times / FPRI / Time)
2| Mag 7 Evaporates Over $2 Trillion, Market Prices In Rate Hike for the First Time This Year
The Magnificent 7 have collectively lost over $2 trillion in market value from their historical highs. Microsoft is down 32% from its October high, its worst start to a year in company history. Meta is down 25%, Alphabet down 15%, and Nvidia and Amazon have turned negative for the year. The S&P 500 fell for five consecutive weeks, its longest losing streak since 2022.
A key expectation flip occurred in the futures market. Traders pushed the probability of a rate hike within the year to 52%, breaking the 50% threshold for the first time. The Atlanta Fed tracker shows a 19.8% probability of a 25 basis point hike. The Fed held rates steady in its March meeting, but Powell "expressed concern" about the lack of progress on inflation. Global forecasting agencies have raised their CPI expectations to 4.2%, far exceeding the Fed's forecast of 2.7%. EY-Parthenon raised the probability of a US recession to 40%.
Institutional funds are shifting from tech growth stocks to energy, defense, and domestic manufacturing. The Mag 7 gained 107%, 67%, and 25% over the past three years; now all seven stocks are in the red for the year. The problem isn't just oil prices; it's that $650 billion in AI capital expenditure has become a burden in a rising rate cycle.
(Sources: Fortune / CNBC / Yahoo Finance / EY-Parthenon / Atlanta Fed)
3| Anthropic Warns Government: Mythos Model Makes Large-Scale Cyber Attacks "More Likely"
Anthropic is privately warning senior US officials that its unreleased "Mythos" model makes a large-scale cyber attack in 2026 "more likely to occur." Previously, information about the model was accidentally leaked due to a CMS system failure, after which the company confirmed Mythos is its "most powerful model to date," representing a "step-change in capabilities."
Axios reported that the model "far surpasses any other AI model in cyber capabilities," allowing agents to "autonomously penetrate corporate, government, and municipal systems with stunning precision." A Dark Reading survey showed 48% of cybersecurity professionals list agent AI as the top attack vector for 2026, surpassing deepfakes and all other threats. Cybersecurity stocks fell on the news, and Evercore issued commentary.
QuantumBit concurrently reported that Claude discovered a 20-year-old system vulnerability in 90 minutes. These two events point to the same inflection point: the growth rate of AI's offensive capabilities is now faster than its defensive capabilities. Frontier labs are no longer just selling products; they are redefining who can breach whom.
(Sources: Axios / Fortune / CNBC / Evercore / QuantumBit)
4| Lilly Bets $2.75 Billion on AI Drug Discovery, Rewriting Traditional R&D Timelines
Eli Lilly signed a $2.75 billion AI drug discovery agreement with Hong Kong's Insilico Medicine, paying $115 million upfront for global exclusive development and commercialization rights. Insilico has used generative AI to develop 28 drugs, nearly half of which are in clinical stages. The two companies have collaborated since 2023, aiming to compress the cycle from target discovery to new drug application from the traditional several years to a few months.
This is the largest single agreement for AI drug discovery by a major pharmaceutical company. Previous AI pharma deals were mostly research collaborations and milestone payments; Lilly directly buying commercialization rights effectively prices AI drug pipelines using traditional pharma valuation logic. STAT News points out that Insilico's core competency lies in using AI for both target discovery and molecular design simultaneously, parallelizing the two most time-consuming steps. When AI is not just an辅助工具 but the pipeline itself, the valuation logic of the pharmaceutical industry may be rewritten.
(Sources: CNBC / Bloomberg / STAT News)
Also Worth Knowing ↓
CLARITY Act stablecoin yield compromise reached, DeFi tokens face headwinds. Senators Tillis and Alsobrooks reached a principled agreement: passive holding yields are banned, active rewards are retained. Protocols like Uniswap, Aave, dYdX may be restricted. Bank committee review scheduled for late April. Stablecoins are being legislatively redefined from speculative tools to payment instruments, forcing a structural adjustment to DeFi's yield models. (Source: CoinDesk / Congress.gov)
Musk warns US chip capacity will soon exceed power supply, a problem China doesn't have. Speaking at Davos with BlackRock CEO Fink, he said the fundamental limiting factor for AI deployment is electricity, as chip capacity grows exponentially but power cannot keep up. The bottleneck in the AI arms race is shifting from supply chains to the power grid. (Source: Fortune)
Southeast Asian stablecoin payments go "invisible," crypto card business surges. Users no longer directly interact with crypto assets as stablecoins are embedded into daily consumption payments. This aligns with the CLARITY Act's direction of positioning stablecoins as payment tools, a signal that crypto is moving from speculation to infrastructure, evident in both legislation and product development. (Source: CoinDesk)
Cambridge University's new memristor chip reduces AI switching current by a million times. New materials solve the energy consumption and precision bottlenecks of traditional memristors, potentially fundamentally changing the energy efficiency of AI inference. Still in the lab stage, but the direction echoes Musk's point that "electricity is the limiting factor." (Source: Tom's Hardware)
Trump's AI agenda becomes an independent campaign issue for the first time in midterm elections. A new political group formed specifically to push AI policy into the 2026 midterm elections. AI governance is moving from the executive order stage into electoral politics. (Source: New York Times)
