Executive Summary
A global sell-off in semiconductor stocks erased $1.3 trillion in market value Friday, triggered by a disappointing earnings report from Broadcom that punctured the euphoria around artificial intelligence spending. The rout coincided with a stark warning from Elon Musk that a “catastrophic” shortage of U.S.-made memory chips threatens national security, even as China announced breakthroughs in training advanced AI models on its own domestic chips. In the Middle East, U.S. forces shot down Iranian drones and struck radar sites near the Strait of Hormuz, prompting Iranian ballistic missile launches toward bases in Kuwait and Bahrain and threatening a fragile regional ceasefire. Meanwhile, President Vladimir Putin publicly rejected a direct peace proposal from Ukrainian President Volodymyr Zelenskyy, dismissing the offer as insincere as China’s Xi Jinping prepared for a state visit to North Korea.
AI & Technology
Broadcom Earnings Trigger Global Chip Stock Rout, Erasing $1.3 Trillion
A sharp sell-off in semiconductor stocks swept through global markets on Friday, erasing approximately $1.3 trillion in value from U.S.-listed chipmakers and threatening to end Wall Street’s nine-week winning streak. The rout was triggered by Broadcom’s quarterly earnings report, which failed to meet the most bullish revenue forecasts for its custom AI chip business. Its shares fell nearly 20% over two sessions, dragging the Philadelphia Semiconductor Index down 10.3% in its worst day since March 2020. Nvidia lost 6%, shedding over $300 billion in market capitalization, while Micron Technology dropped 13% and Marvell Technology sank 17%. The weakness spread to Asia, where South Korea’s SK Hynix fell 10% and Samsung Electronics dropped 6%.
Analysts said the sell-off reflected an increasingly high bar for companies to impress investors after a stunning, narrow rally driven by AI optimism. “You’ve had a lot of people here that were just blindly buying the dip,” said Dennis Dick, a proprietary trader at Triple D Trading. “Blindly buying the dip had been winning you money, but that ended today.” The market’s reaction to Broadcom’s report, which was solid but not spectacular, suggests that valuations had become detached from near-term business fundamentals. The sell-off was compounded by stronger-than-expected U.S. jobs data, which showed employers added 172,000 jobs in May, raising concerns that the Federal Reserve may have less room to cut interest rates this year.
The Philadelphia Semiconductor Index, despite the two-day rout, remains up 73% for the year, indicating that the underlying thematic investment case for AI hardware remains largely intact. The immediate test will be whether other major chip companies can meet the market’s lofty expectations in upcoming earnings reports. “After such massive gains a ‘correction’ for recent winners was sorely needed for a reset,” said Andrew Jackson, an equity strategist at Ortus Advisors.
China Announces AI Self-Reliance Breakthroughs Amid Chip Shortage Warnings
Chinese researchers and companies announced a series of advances this week aimed at reducing the country’s dependence on foreign technology, as Elon Musk warned of a “catastrophic” shortage of U.S.-made memory chips for AI. A team including Huawei Technologies successfully used a cluster of at least 1,000 of its Ascend 910C chips to conduct full-parameter training for the massive DeepSeek-V4-Pro model, a complex process previously dominated by U.S. chips like Nvidia’s. Separately, a state-backed consortium established the Beijing Space Intelligent Computing Research Institute, aiming to launch a pilot satellite for orbital AI computing by the end of 2028.
The developments signal a strategic pivot by China’s tech sector toward foundational, long-term capabilities. The Huawei-led project directly addresses a critical weakness: China’s domestic semiconductors have largely been relegated to running finished AI models, not building or refining them. “The exploration will help enhance the self-reliance of China’s AI industry chain,” a Shenzhen government social media post stated. This push is being fueled in part by a reverse brain drain, exemplified by the appointment of former OpenAI researcher Yao Shunyu as Tencent’s chief AI scientist. “My personal goal is that in China we should establish a long-term AGI organization,” Yao said at a Tencent event in Beijing.
The announcements present a coordinated picture of a nation investing across the entire AI stack, even as Western companies grapple with supply constraints. In a recent conversation with JPMorgan CEO Jamie Dimon, Elon Musk stated that the United States currently has “zero” high-volume computer memory chip fabrication plants, calling the situation “catastrophic” for national security and technological competitiveness. He noted that Micron’s planned facilities will not reach volume production until 2028-2030. The immediate test for China will be whether these research breakthroughs can scale commercially and withstand the pressure of ongoing U.S. export controls.
Anthropic Engineers Embed at NSA for Cyber Ops as Company Calls for AI Pause
Anthropic has embedded roughly half a dozen engineers inside the National Security Agency to help deploy its advanced Mythos AI model for offensive cyber operations, according to a Financial Times report, even as the company publicly urges a global slowdown in AI development. The engineers are customizing the model for specific applications, with one source suggesting it could be used to infiltrate networks in countries like China and Iran. This arrangement persists even as Anthropic is locked in a legal battle with the Pentagon, which designated the company a “supply-chain risk” after it refused to allow its commercial Claude model to be used for mass domestic surveillance or fully autonomous weapons.
The company’s simultaneous advocacy for a global pause, detailed in a recent research paper, has drawn criticism from skeptics who view it as a marketing ploy. Anthropic has limited Mythos to a select group of vetted partners through its Project Glasswing, citing the risk of misuse. The NSA’s use of the model was reportedly exempt from the company’s ban on military applications for autonomous weapons and surveillance. The legal conflict stems from a collapsed $200 million contract with the Department of Defense. A California judge temporarily blocked the Pentagon’s “supply-chain risk” designation, but a D.C. appeals court denied Anthropic’s request to halt the ban while litigation proceeds.
This situation highlights the deepening entanglement between leading AI labs and national security agencies, even amid public calls for caution. It raises unresolved questions about the governance of dual-use technology and the credibility of corporate safety advocacy when the same company is actively enabling state-level cyber capabilities. Anthropic is also preparing for an initial public offering that could value it above $1 trillion, according to reports.
OpenAI Agrees to Voluntary U.S. Government AI Review
OpenAI has agreed to comply with a voluntary executive order from President Donald Trump that requests AI companies allow the federal government to review their models before release. The order, signed on Tuesday, asks for access to models 30 days before public debut and establishes a benchmarking process to assess advanced cyber capabilities. OpenAI’s head of countries, George Osborne, confirmed the company’s participation, stating it was “quite right that democratic governments have a big role to play.”
Separately, the Trump administration is exploring a more radical proposal: taking equity stakes in leading AI companies, including OpenAI and Anthropic. President Trump told reporters on Friday that he had discussed a deal in which the federal government “essentially becomes a partner with the companies,” and said further meetings with AI labs would occur soon. This concept gained momentum after Senator Bernie Sanders proposed a one-off, 50 percent tax on AI labs’ stock to fund a U.S. sovereign wealth fund, an idea met with immediate criticism from Silicon Valley figures.
The executive order itself represents a significant but deliberately limited step. An earlier, more stringent draft required a 90-day review window and was reportedly delayed and watered down following pressure from the tech industry. The final version explicitly disclaims creating any mandatory “preclearance” regime, a clause that critics say undermines its effectiveness. Representative Don Beyer (D-VA) called it “underwhelming policy” that creates a “wild west environment for AI development.” The parallel discussions about government equity stakes suggest the administration is considering multiple, potentially conflicting, avenues for influence over a critical industry.
Geopolitics & Security
U.S. and Iran Exchange Strikes, Threatening Strait of Hormuz Ceasefire
U.S. forces shot down four Iranian drones over the Strait of Hormuz on Friday and subsequently struck Iranian coastal surveillance radar sites at Goruk and on Qeshm Island, according to U.S. Central Command. The military said the drones posed an immediate threat to maritime traffic. Iran responded by firing ballistic missiles toward U.S. bases in Kuwait and Bahrain, with CENTCOM reporting six were intercepted and a seventh failed to reach its target. Kuwait and Bahrain activated air raid sirens and urged residents to take shelter, though there were no reports of U.S. casualties.
The exchange is the latest in a series of escalating tit-for-tat strikes over the past two weeks that threaten a fragile ceasefire in place since early April. Iran’s Revolutionary Guard claimed it targeted the Ali Al Salem airbase in Kuwait and the headquarters of the U.S. Navy’s 5th Fleet in Bahrain, a claim CENTCOM disputed as false. The U.S. is enforcing a blockade on Iranian ports in response to Tehran’s effective control over the Strait of Hormuz, a crucial chokepoint for global oil shipments. The conflict has sent energy prices spiking and created political challenges for President Trump’s Republican Party ahead of midterm elections.
With both sides postured for further action, the immediate risk is a miscalculation that ruptures the ceasefire entirely. The repeated targeting of surveillance infrastructure suggests the U.S. is attempting to degrade Iran’s ability to monitor and potentially interdict shipping in the strait. Indirect negotiations for a longer-term agreement have stalled, with Tehran demanding access to billions in frozen oil revenue, sanctions relief, and leverage over the strait. The unresolved question is whether these calibrated strikes will pressure Iran back to negotiations or provoke a wider regional conflagration.
Putin Rejects Zelenskyy’s Public Proposal for Direct Peace Talks
Russian President Vladimir Putin on Friday dismissed a public offer from Ukrainian President Volodymyr Zelenskyy for a face-to-face meeting to negotiate an end to the war, calling the proposal insincere and saying he saw “no point” in such talks. Speaking at the St. Petersburg International Economic Forum, Putin described Zelenskyy’s open letter, published Thursday, as “rude” and “boorish,” arguing it was designed to make a meeting impossible rather than to facilitate one. Zelenskyy responded hours later in a nightly video address, stating that Putin’s rejection showed the Kremlin had “no wish to end the war” and calling it a “weak response.”
The exchange marks the first direct public communication between the two leaders since Russia’s full-scale invasion began in February 2022 and underscores the complete diplomatic stalemate as the conflict enters its fifth year. Zelenskyy had framed the offer as a necessity, arguing that the United States is “fully focused on the issue of Iran” and that Ukraine cannot wait for the war to return to the center of Washington’s attention. He stated the starting point for any talks would be “the front line today,” explicitly rejecting earlier proposals that would have forced Ukraine to cede occupied territory.
The rejection, coupled with Putin’s reaffirmation of his territorial goals, suggests no near-term off-ramp for the war, with both sides continuing to pursue military pressure as their primary leverage. Analysts note the letter underscores Kyiv’s growing confidence on the battlefield even as Zelenskyy conceded a significant disadvantage in ballistic missiles. The key question is whether mounting Russian battlefield losses and internal elite discontent might eventually compel Moscow to consider a diplomatic off-ramp it has so far refused.
Xi Jinping to Visit North Korea Amid Heightened Nuclear Tensions
Chinese President Xi Jinping will make a state visit to North Korea on Monday and Tuesday, meeting with leader Kim Jong Un for the first time in Pyongyang in nearly seven years. The trip, announced by both countries’ state media on Friday, comes amid a flurry of diplomatic activity by Beijing, which recently hosted U.S. President Donald Trump and Russian President Vladimir Putin. The Chinese Foreign Ministry said the visit would “advance ties and strengthen regional peace and stability.”
Analysts say the timing is significant, as it follows North Korea’s recent public display of its nuclear capabilities and its deepening military cooperation with Russia. Just a day before the visit was announced, North Korean state media reported that Kim had toured a new nuclear-material production facility, highlighting advances in producing fissile material. This display, along with Pyongyang’s recent codification of an “automatic nuclear launch” policy, appears designed to project strength ahead of Xi’s arrival and signal that denuclearization is off the table.
The visit underscores China’s effort to reassert its influence over Pyongyang as Kim has cultivated a closer partnership with Moscow, including sending troops and conventional weapons to support Russia’s war in Ukraine. China remains North Korea’s largest trading partner and primary provider of aid. The trip also follows recent trilateral talks among the U.S., Japan, and South Korea, which analysts say have galvanized Pyongyang. It is unclear whether Xi will press Kim to moderate his nuclear posture or if the visit will instead solidify a united front against U.S.-led pressure in Northeast Asia.
Israel Kills Lebanese Soldiers as Hezbollah Rejects Ceasefire
Israeli airstrikes killed several Lebanese soldiers, including an officer, on a road in southern Lebanon early Saturday, marking a significant escalation as the Lebanese army has typically avoided direct conflict with Israel. The strike near Nabatieh occurred amid intense cross-border fighting that continued despite a U.S.-brokered ceasefire deal between Israel and the Lebanese government, which was rejected by Hezbollah on Thursday. Separately, Israeli gunfire killed a seven-month-old Palestinian baby and wounded his parents at a checkpoint near Hebron in the occupied West Bank on Friday evening, an incident the Israeli military said it was investigating after soldiers perceived a threat from an accelerating vehicle.
The Lebanese military’s casualties represent a dangerous broadening of the conflict, drawing in a state institution that has sought to remain on the sidelines. Prime Minister Nawaf Salam, addressing the crisis, stated the war “is not being fought for our sake, but on our land and at the expense of our people,” and criticized Iran for using Lebanon as a “bargaining chip” in its negotiations with the U.S. Hezbollah, meanwhile, claimed it carried out around 20 attacks on Israeli troops in southern Lebanon on Friday, responding to what it called Israeli ceasefire violations.
The U.S.-mediated ceasefire, announced but immediately rendered moot, would have allowed Israel to continue airstrikes while requiring Hezbollah to halt fire, a condition the group dismissed as “surrender.” The failure of the deal and the subsequent lethal strike on Lebanese soldiers underscore the limited control the Lebanese government exerts over Hezbollah and the precarious position of state institutions caught in the crossfire. The direct attack on the Lebanese army risks triggering a wider state-on-state conflict and further destabilizing Lebanon, where nearly one million people are already displaced.
Trump Directs New Acting Spy Chief to Shrink Intelligence Staff
President Donald Trump has instructed his newly appointed acting Director of National Intelligence, Bill Pulte, to initiate a significant reduction in staff within the Office of the Director of National Intelligence (ODNI). In an interview with The Wall Street Journal published Friday, Trump said he wanted Pulte to “start the process” of firing employees, calling the office “unnecessary” and “too big.” He suggested the ODNI, which oversees 18 intelligence agencies including the CIA and NSA, should perhaps be terminated altogether.
The directive comes amid existing controversy over Pulte’s appointment. Trump named Pulte, the current director of the Federal Housing Finance Agency and a political loyalist with no prior intelligence experience, to the acting role on Tuesday. He is replacing outgoing DNI Tulsi Gabbard, who resigned in May. Trump told the Journal that an acting director is “less shackled” and has “more power” to enact rapid change. An ODNI official noted that Gabbard had already reduced the office’s staff by nearly 50% under her “ODNI 2.0” initiative, saving an estimated $1 billion.
Critics, including senior Republican lawmakers, have condemned the appointment as a move to further politicize the intelligence community. Senate Majority Leader John Thune (R-S.D.) said, “We don’t need a weaponized DNI. We need professionals there.” The push for cuts follows a separate, explosive whistleblower allegation from a former Social Security official, who claims the Trump administration planned to mark 2.7 million living people as dead in federal records as part of an immigration enforcement effort. The immediate focus will be on the scale and speed of Pulte’s actions at an office central to U.S. national security coordination.
Economy & Markets
AirTrunk Commits $30 Billion to AI Data Centers in India
Blackstone-backed data center operator AirTrunk announced a $30 billion investment to develop 5 gigawatts of AI data center capacity in India by 2030, marking one of the largest single commitments to the country’s digital infrastructure. The plan, which follows AirTrunk’s recent entry into the Indian market via acquisition, was personally welcomed by Prime Minister Narendra Modi, who framed it as a validation of India’s growing role in the global digital economy. The announcement coincides with intense capital pressure on the world’s largest technology firms, as Alphabet seeks to raise $85 billion in fresh equity to fund its own AI infrastructure build-out.
The parallel developments underscore a global scramble for both physical AI infrastructure and the capital required to build it. AirTrunk’s massive bet reflects India’s strategic push to become a major hub for cloud and AI compute, aided by government incentives like long-term tax exemptions for foreign cloud providers. This investment wave is projected to increase India’s total data center capacity from about 1.5GW today to as much as 8GW by 2030. Meanwhile, Alphabet’s pursuit of new capital, even after securing over $55 billion in debt since November, highlights the unprecedented financial demands of the AI arms race. Analysts note that Google’s free cash flow is expected to turn negative for several years as it ramps spending.
The competition extends beyond capital and concrete to talent and technological vision. In a separate but related signal from the AI sector, former OpenAI CTO Mira Murati re-emerged publicly to promote her new venture, Thinking Machines Lab, after 18 months of operating quietly. Her careful re-entry into the spotlight, where she previewed a novel “interaction model” for AI, suggests a calculated effort to maintain relevance in a market dominated by giants. The question now is whether the flood of capital into infrastructure will create a sustainable foundation for global AI development or lead to overcapacity and financial strain.
Meta and Alphabet Fundraising Sparks Investor Jitters Over AI Costs
Meta’s stock fell more than 5% on Friday following a Financial Times report that the company could raise tens of billions of dollars in a stock offering to fund its artificial intelligence investments. The report, which a Meta spokesperson called “pure speculation,” came as rival Alphabet announced plans to raise $85 billion through equity sales, a move that contributed to a 3.8% single-day drop in its stock and erased roughly $10 billion each from the personal fortunes of co-founders Larry Page and Sergey Brin. The divergent market reactions highlight investor unease over the massive capital expenditures required for AI infrastructure, with Meta raising its 2026 capex forecast to as high as $145 billion and Alphabet hiking its guidance to $190 billion.
This spending spree is colliding with a severe semiconductor supply crisis, a point underscored by Elon Musk in a recent conversation with JPMorgan CEO Jamie Dimon. Musk stated that the United States currently has “zero” high-volume computer memory chip fabrication plants, calling the situation “catastrophic” for national security and technological competitiveness. He noted that Micron’s planned facilities in Idaho and New York will not reach volume production until 2028-2030, and even then will only meet a fraction of the demand driven by AI and robotics.
Amid this capital and supply race, Meta is seeking to differentiate its AI models by focusing on health capabilities. Alexandr Wang, head of Meta Superintelligence Labs, stated that health is a “critical” area for the company’s future models, citing the performance of its recently released Muse Spark model. He acknowledged Muse Spark is not yet at the tier of leading frontier models from OpenAI or Anthropic but said it performed better than expected internally. The immediate financial market volatility reflects deep uncertainty over the return on these historic investments, with the broader chip stock sell-off adding to the pressure on tech giants to justify their spending.
From the Timeline
The Political Weaponization of AI and Calls for Regulation
A significant debate is unfolding around the potential nationalization and regulation of AI, sparked by Bernie Sanders’ proposal for a 50% public stake in major AI companies. @DavidSacks dissected the political dynamics, arguing that while the proposal resonates due to AI leaders’ own “massive job loss” warnings and fears of ideologically biased philanthropy, government ownership could lead to a “Central Government AI” with Orwellian control over information. He cautioned conservatives against supporting regulations that could be weaponized by future administrations. Meanwhile, @paulg criticized the current administration’s tone, sharing a White House webpage he found embarrassingly reminiscent of propaganda for a “third world dictator,” reflecting a broader sentiment of governmental overreach in tech circles.
The AI Infrastructure Boom and Compute Economics
The staggering financial scale of AI infrastructure is coming into focus, with SpaceX emerging as a dominant force. @chamath highlighted that SpaceX’s compute revenue from deals with Anthropic and Google alone runs at a $26 billion annual rate, a sum @EMostaque noted is comparable to major cloud providers. This compute scarcity is driving innovation in efficiency, as seen with @hardmaru announcing the Sakana AI RSI Lab in Tokyo, focused on “compute-efficiency” and “sample-efficient” self-improving AI, a direct challenge to the “brute-force status quo.” @ID_AA_Carmack also pondered if chip fabrication could be optimized for wafer throughput over pure performance to alleviate constraints.
Building the “Agent-First” Software Stack
A clear theme is the architectural shift towards AI agents as primary users of software. @naval succinctly predicted that “Software platforms are going to be rebuilt for agent-first.” This is being operationalized by builders like @garrytan, who shared a detailed framework for an “agent company” using a central “gBrain” orchestrator and narrowly scoped specialist agents to improve output quality and prevent context bleed between clients. The economic imperative for this shift is underscored by @ClementDelangue, who provided data showing that agents using efficient developer tools like a dedicated CLI succeed more often (94% vs 84%) and use up to 6x fewer tokens than those rebuilding workflows from scratch, proving that “good tools are cached intelligence for agents.”
Crypto’s Evolution Beyond Bitcoin
Thought leaders are pushing back against the reductive narrative that crypto’s health is solely tied to Bitcoin’s price. @brian_armstrong argued that “Crypto touches every area of finance,” noting growth in derivatives, stablecoins, and prediction markets. He also showcased Coinbase’s deepening infrastructure play, announcing an investment in ProShares’ money market ETF designed for stablecoin reserves post-GENIUS Act. This aligns with a view of crypto maturing into a broader financial stack, a sentiment echoed by @tobi, who celebrated the integration of Shopify with Vercel’s v0 to easily build Next.js storefronts, pointing to the mainstreaming of web3-enabled commerce tools.
Venture Capital Anecdotes and Industry Culture
A mix of criticism and insider humor colored discussions about VC culture. @pmarca and @wolfejosh engaged with a thread about “worst VC stories,” which included a Sequoia partner allegedly passing on Cloudflare due to gender bias and a miscommunicated, awkward pitch meeting with Andreessen Horowitz. @pmarca also quipped with the insider phrase, “He’s an autist, but he’s our autist,” highlighting the industry’s tolerance for atypical brilliance. Meanwhile, @chamath reiterated his belief that today’s tech giants will look “tiny” in 10-15 years due to the order-of-magnitude capabilities concentrated in a smaller subset of companies, a view he shared on the All-In podcast.
Platform Governance and Narrative Control
Experts raised alarms about the manipulation of public information and platform governance. @wolfejosh amplified an investigation alleging a “weaponization of Wikipedia” by a small group of editors to spread a negative narrative about Hindu American organizations, framing it as an attack on Indian Americans. This concern over narrative control intersects with platform policy, as @elonmusk announced that X will directly message users who interact with a post later corrected by Community Notes, a novel approach to combating misinformation. Separately, @elonmusk engaged with discourse framing the term “far right” as a propaganda tool against ordinary people.
Economic Data and Market Reactions
Market-moving news and economic data prompted immediate commentary. @DavidSacks highlighted a strong May jobs report, declaring it a beating for the “jobs apocalypse narrative” often associated with AI. @zerohedge noted Trump’s positive reaction to the Iran situation, linking it to market performance, while also reporting on the Chicago Bears’ planned move to Indiana and the DOJ observing vote counting in LA, stitching together a tapestry of political and economic signals.