AI News Roundup — June 4, 2026
Broadcom lost $315 billion in market cap in a single day after its AI chip guidance failed to meet sky-high expectations, sending shockwaves through the chip sector — while Canada launched its national 'AI for All' strategy targeting 250,000 new jobs, and WWDC 2026 previews confirmed Apple's rebuilt Siri will run on Google Gemini.
Broadcom Falls 14%, Shedding $315 Billion in Market Cap After AI Guidance Disappoints
Broadcom stock dropped more than 14% on June 4, wiping out over $315 billion in market value — one of the largest single-day market cap losses in US history. The paradox: Q2 AI semiconductor revenue of $10.8 billion grew 143% year-over-year and beat forecasts. But CEO Hock Tan's guidance of $16 billion in AI chip revenue for Q3 (which would be 200%+ YoY growth) was seen as falling short of inflated investor expectations. Broadcom's role backstopping the Apollo/Blackstone $36B Anthropic TPU debt deal (announced June 1) makes the stock reaction a broader signal about AI capex sustainability.
Canada Launches 'AI for All' National Strategy — $2 Billion, 250,000 Jobs by 2031
Prime Minister Mark Carney announced Canada's new national AI strategy in Toronto on June 4, targeting an additional $200 billion in economic growth and 250,000 new jobs by 2031. The 'AI for All' strategy includes new legislation, $2B+ in direct investment, a Canadian Sovereign AI Compute Strategy, and a deliberate effort to reduce reliance on US tech giants. Carney positioned Canada as a natural AI sovereignty leader among middle powers — citing its energy supply, engineering talent, and stable democratic institutions. Politico called it Canada's bid to lead the global AI sovereignty race.
Prime Minister of Canada · CBC News · Politico
WWDC 2026 Preview: Siri's Google Brain, Tim Cook's 'Swan Song,' and iOS 27
With WWDC 2026 four days away, CNET's preview called it Tim Cook's 'Swan Song' — a final defining product event before a leadership transition. The headline: Siri is being rebuilt from scratch on a custom Google Gemini model (the $1B/year licensing deal confirmed), with full conversational multi-turn capabilities launching in iOS 27 alongside iPhone 18 in September. Apple is expected to demonstrate on-device AI running on Apple Silicon, developer APIs for Siri extensions, and a completely redesigned AI interface replacing today's Siri experience.
Nvidia Enters the $200 Billion CPU Market With AI Agent PCs
Nvidia announced a push into the CPU market with a new line of AI Agent PCs in partnership with Microsoft, Dell, and HP — powered by Nvidia's Arm-based RTX Spark chips. The move targets the estimated $200 billion CPU market that Intel and AMD have historically dominated. AI Agent PCs are designed to run local AI inference for coding agents, personal assistants, and enterprise workflows without cloud round-trips. The Surface RTX Spark Dev Box (launched at Microsoft Build) is the first device in this category.
AI Safety Community Debates Trump Executive Order — Is 'Voluntary Review' Meaningful?
Following Trump's June 2 AI executive order, AI safety researchers and policy experts are debating whether the voluntary 30-day government preview mechanism has any teeth. Critics argue labs can simply opt out with no consequences. Supporters — including Anthropic — see it as a framework for responsible disclosure that could evolve into binding regulation. OpenAI's resistance to the order while Anthropic embraced it created an unusually public split between the two most prominent AI labs on a policy question.
SpaceX and Anthropic IPO Risks — 'Beware the Ghosts of IPOs Past'
With SpaceX's $75B IPO roadshow underway and Anthropic targeting a fall listing, Reuters published an analysis warning about the structural risks: both companies are valued at multiples that require sustained hypergrowth for years. SpaceX posted a $4.28B net loss in Q1 2026 (up 700%+ YoY) while pursuing orbital AI compute infrastructure. Anthropic's $965B valuation implies 22x forward revenue at a company still less than 4 years old. Reuters cited Snap, Lyft, and Uber as cautionary tales of high-valuation IPOs that disappointed on durability.
Claude Mythos Bypassed Apple M5's Memory Integrity Enforcement in 5 Days
A new detailed report from FelloAI revealed that during Glasswing testing, Claude Mythos Preview bypassed macOS's Memory Integrity Enforcement on Apple M5 chips in five days — exploiting a kernel-level vulnerability that Apple has since confirmed and is patching. The disclosure follows the 27-year OpenBSD bug and 16-year FFmpeg flaw from the Glasswing 30-day update, adding to the case that Mythos represents a qualitatively different level of automated vulnerability discovery.