Skip to content
Daily Brief News Daily Brief

5 AI stories from June 30, 2026: GitHub Copilot billing shock, Google DeepMind talent exodus, Five Eyes cyber warning, Alphabet's $84.75B raise, and RAISE US workforce fund

· by Pondero Newsdesk · 5 stories

AI news daily brief: 2026-06-30

Copilot's first metered month closed with bills up to 50x higher than expected, the same day Alphabet printed the largest equity raise in US history to fund the compute behind those costs. Alongside: six Google DeepMind researchers gone to rival labs, a Five Eyes cyber advisory with a months-not-years urgency clock, and a $500 million bipartisan workforce fund from the four biggest AI vendors.

GitHub Copilot's first metered billing cycle closed with developers reporting 10x to 50x cost jumps on agentic workflows

June 30 marks the end of Copilot's first full month under token-based "AI Credits" billing, which replaced flat subscription tiers on June 1. The new model prices one credit at $0.01, consumed at published API rates per model. Per GitHub's announcement, base plans held flat: Copilot Pro at $10/month, Pro+ at $39/month, and a new Max tier at $100/month. Code completions and Next Edit suggestions remain free on all paid plans. The meter runs on agentic sessions, frontier model access, and multi-step autonomous tasks.

The cost shock hit developers running extended agentic workflows. One developer projected a monthly bill jumping from roughly $29 to $750. Another shared a screenshot showing a projection near $3,000, up from around $50. The structural change that caught users off guard: exhausting your credits stops premium features entirely, with no fallback model under the new system. GitHub CPO Mario Rodriguez framed the shift as a sustainability move, noting that short chats and multi-hour autonomous sessions were priced the same under the old model. Cursor and Claude Code both offer flatter agentic pricing, though rates shift periodically.

Full story: GitHub Copilot billing shock

Six Google DeepMind researchers left for Anthropic and OpenAI in five months, costing Alphabet more than $245 billion in market value

Four departures in ten days during June capped a five-month run that began when Denny Zhou (reasoning research) left for Meta's Superintelligence Lab in February. On June 18, Noam Shazeer, a co-author of the 2017 "Attention Is All You Need" Transformer paper and a key Gemini contributor, announced he was joining OpenAI. Two days later, John Jumper, the 2024 Nobel laureate in Chemistry for AlphaFold, confirmed a move to Anthropic. That single departure erased more than $245 billion from Alphabet's market capitalization in one trading session per The Next Web.

On June 24, Jonas Adler (AI coding products) and Alexander Pritzel (pretraining) also confirmed moves to Anthropic per TechCrunch. Sergey Brin, who oversees Google's AI Coding Strike Team, put the internal stakes plainly in a memo: "To win the final sprint, we must urgently bridge the gap in agentic execution and turn our models into primary developers of final code." For operators tracking Claude Code's roadmap, Adler's and Pritzel's expertise in coding products and pretraining landed directly on the Anthropic team building it.

Full story: Google DeepMind researcher exodus

Five Eyes agencies warned that frontier AI will reshape cyber defenses in months, not years

On June 22, cybersecurity directors from the US (NSA and CISA), UK (NCSC), Canada (CSE), Australia (ASD), and New Zealand (GCSB) signed a joint advisory on frontier AI and offensive cyber risk. The central claim: "The rapid pace of frontier AI development means cyber risk assumptions can become outdated in months, not years." The agencies called out AI-enabled autonomous vulnerability scanning, AI-assisted exploit generation, and AI-powered social engineering as near-term risks already in use.

Four immediate actions topped the advisory for executives: reduce attack surface by limiting unnecessary system access, accelerate patching cadence because AI compresses the discovery-to-exploitation gap, treat legacy systems as "strategic liabilities" rather than deferred maintenance, and tighten identity and access controls. The agencies noted defenders have the same tools: organizations integrating AI into security operations can detect vulnerabilities earlier and respond faster. The statement is the highest-credibility pre-legislative signal available from any Western government. Operators in regulated industries or with government contracts should treat the four action items as near-term compliance expectations.

Full story: Five Eyes AI cyber warning

Alphabet closed an $84.75 billion equity raise, the largest in US corporate history, to fund AI compute infrastructure

Alphabet completed the raise across three structures: a $30 billion underwritten public offering of Class A and Class C shares alongside mandatory convertible preferred depositary shares, a $10 billion concurrent private placement to Berkshire Hathaway, and a $40 billion at-the-market share program that launches in Q3 2026. The final total of $84.75 billion exceeded the initial $80 billion announcement after institutional demand triggered an upsize within 24 hours per Alphabet's Form FWP filed with the SEC. The previous US equity raise record stood at $70 billion, set by Brazil's Petrobras in 2010.

Berkshire's $10 billion placement matters because the firm built its reputation on avoiding open-ended capital-intensive businesses. Its participation signals a view that AI infrastructure returns are durable, not speculative. Alphabet's 2026 capex guidance sits at $180-190 billion, covering data centers, custom silicon, and network capacity for training and serving AI models. The $84.75 billion raise covers roughly half the annual plan in a single transaction. CEO Sundar Pichai said demand for AI solutions currently exceeds available compute supply, and that core AI response costs dropped more than 30% since Gemini 3, per the investor presentation.

Full story: Alphabet $84.75B equity raise

RAISE US launched with $500 million from Amazon, Anthropic, Microsoft, and OpenAI to retrain US workers displaced by AI

Former US Commerce Secretary Gina Raimondo and former Indiana Governor Eric Holcomb announced RAISE US on June 25, structured as a nonpartisan nonprofit with a $1 billion fundraising target. Amazon, Anthropic, Microsoft, and the OpenAI Foundation anchor the initial $500 million, joined by Bank of America, Cisco, General Motors, IBM, Eli Lilly, and the Stephen A. Schwarzman Foundation, among others. AFL-CIO President Liz Shuler holds a board seat. Initial state programs are live in Arkansas, Connecticut, Maryland, and Utah per the RAISE US launch release.

Raimondo framed the stakes directly: "America has a technology strategy for leading the global AI competition. It does not yet have a people strategy." An Anthropic Economic Index survey of 9,700 workers conducted in June 2026 found that more than 35% expect AI to perform most of their work tasks within 12 months. The bipartisan co-leadership is structural: Raimondo is a Democrat, Holcomb a Republican, and early state partnerships span red and blue states. Published outcome data from the state pilots arrives 12 to 24 months out. Track the Arkansas and Connecticut cohorts first, as those states have the most established workforce-retraining infrastructure.

Full story: RAISE US AI workforce initiative

Sources