Tech News — Week of 2026-05-22 to 2026-05-29#

Story of the Week#

Anthropic has officially unseated OpenAI as the world’s most valuable artificial intelligence startup, closing a jaw-dropping $65 billion Series H funding round that catapults its valuation to $965 billion. This historic changing of the guard arrives alongside the launch of Claude Opus 4.8 and a novel “Dynamic Workflows” tool for orchestrating swarms of AI subagents. The massive capital influx proves that explosive enterprise demand is rapidly reshaping the generative AI hierarchy, placing Anthropic squarely in the driver’s seat for the next era of frontier model development.

Top Stories#

[SpaceX’s Wild Ride Toward a $1.8 Trillion IPO] · Various Elon Musk’s SpaceX executed a successful orbital test of its Starship V3 prototype, marking a massive leap forward for the economics of orbital logistics, even though the Super Heavy booster was lost during descent. The milestone coincides with the company’s formal IPO filing, initially detailing a $28 trillion total addressable market before adjusting its valuation target down to a more grounded $1.8 trillion. With Blue Origin’s New Glenn rocket suffering a catastrophic launchpad explosion this week, SpaceX’s dominance in space—and its deepening, multi-billion dollar defense ties—remains practically uncontested.

[The AI Boom Hits the Reality Distortion Field] · Various The unit economics of generative AI are facing severe stress tests. Uber’s president openly questioned the ROI of its AI budget after exhausting it in just four months, struggling to connect rising token consumption to actual consumer value. Meanwhile, as DeepSeek permanently slashes its flagship V4-Pro AI model pricing by 75% to violently undercut Western competitors, Meta is launching paid subscriptions across Facebook, Instagram, and a $19.99/mo “Meta One” chatbot tier just to offset its monstrous infrastructure burn.

[Agentic AI Infrastructure is a Security Nightmare] · Various As the tech industry races to deploy autonomous agents, a critical vulnerability in the Starlette Python framework has exposed the credentials of millions of AI agents relying on Model Context Protocol (MCP) servers to access external systems. Recognizing the systemic risks of agentic workflows, AWS and Google have rushed to implement strict operational guardrails and sandboxed execution environments. The real-world friction is already apparent, with the Bank of England completely locking out Anthropic’s powerful Mythos tool due to unresolved cybersecurity fears.

[The Vatican and Silicon Valley Align on AI Oversight] · Various In an unprecedented move, Pope Leo XIV released Magnifica Humanitas, a 42,300-word encyclical warning of AI’s threat to human dignity and employment. Co-presented by Anthropic’s Christopher Olah, the document schooled tech billionaires on moral limits while demanding strict government regulation. It represents a fascinating strategic alignment between the Vatican and the AI safety wing, particularly as US federal regulation stalls and President Trump abruptly scrapped a frontier AI oversight order following intense lobbying from Elon Musk and Mark Zuckerberg.

[The Energy Transition Hits a Tipping Point] · Various Global electricity generated by wind and solar power has officially surpassed gas for the first time in history, hitting 22% and generating a record 531 terawatt-hours in April. The global transition could soon accelerate even further as MIT researchers backed by startup Rock Zero have commercialized a low-heat method to extract battery-grade lithium using weak acids in plastic tanks. This process eliminates the need for extreme-temperature roasting, severely undercutting traditional hard-rock mining costs to under $6,000 per metric ton.

Also Worth Knowing#

Valve’s Pricing Power: Valve hiked the price of its 1TB OLED Steam Deck to a staggering $949 due to rising component costs, yet the popular gaming handheld still completely sold out in North America in under 24 hours.
Robinhood’s Risky Bot Trading: Robinhood is officially allowing users to deploy autonomous AI agents to trade their life savings, though the brokerage explicitly warns that agentic trading carries a high risk of total investment loss.
Linux’s Double Whammy: A decade-old critical vulnerability (CVE-2026-46333) was discovered in mainline Linux kernels granting instant root access, arriving exactly as solo maintainers face severe burnout from a 20% spike in AI-generated “drive-by” bug reports.
Google Insider Trading: Federal prosecutors charged a Google security engineer with fraud after he allegedly weaponized unreleased 2025 “Year in Search” data to net $1.2 million betting on Polymarket.
Ferrari’s Design Miss: Ferrari’s first all-electric vehicle, the $640,000 “Luce” co-designed by Jony Ive, is facing a brutal online backlash for resembling a Nissan Leaf or an Apple product rather than an aggressive Italian supercar.