Top Stories
Anthropic Set to Close $30B Round, Leapfrogging OpenAI as Most Valuable AI Startup
Anthropic is expected to close a funding round exceeding $30 billion this week at a valuation above $900 billion, according to Bloomberg. The deal, co-led by Sequoia Capital, Dragoneer Investment Group, Altimeter Capital, and Greenoaks Capital Partners — each investing roughly $2 billion — would make Anthropic the world’s most valuable AI startup, surpassing OpenAI’s $852 billion last-round valuation.
The round’s speed is striking. Anthropic’s February Series G valued the company at $380 billion post-money, meaning the implied valuation has more than doubled in three months. That trajectory is being driven by the financial performance reported last week: $10.9 billion projected Q2 revenue, a 130% jump from Q1, and a first-ever operating profit of $559 million.
The timing is deliberate. With OpenAI’s confidential S-1 already filed and both companies eyeing fall IPOs, Anthropic is locking in its valuation narrative before public market scrutiny begins. Peter Thiel’s Founders Fund and General Catalyst are also participating. If the round closes as expected, Anthropic will have raised over $60 billion in total funding in 2026 alone.
Pentagon Tests Rival AI Models in Race to Replace Claude
The Department of Defense is actively testing AI models from OpenAI and Google to find replacements for Anthropic’s Claude, which has been embedded in classified military workflows including the Maven Smart System used in operations against Iran. Testing began in early March, three days after Defense Secretary Pete Hegseth designated Anthropic a supply-chain risk over the company’s insistence on guardrails for military applications.
The evaluation involves 25 of the Pentagon’s “power users” comparing how competing models handle the same classified queries. Early results show significant behavioral differences between models given identical prompts, and evaluators have found that tailoring prompts to specific models yields better results — an implicit acknowledgment that Claude’s performance advantage came partly from the military having optimized its workflows for that particular model.
The situation captures the central contradiction of the current moment: the US government is simultaneously trying to replace Anthropic over safety guardrails while Anthropic’s valuation doubles, the Vatican platforms its co-founder as a moral voice on AI, and every major investor races to give it more money. The market has clearly decided that safety-focused AI development isn’t a liability — even if the Pentagon disagrees.
Source: Bloomberg, CryptoBriefing
Intuit Cuts 17% of Its Workforce as AI Reshaping Accelerates
Intuit announced it’s eliminating 3,000 jobs — 17% of its workforce — joining the growing list of major tech companies restructuring around AI capabilities. The cuts come alongside a fiscal Q3 showing $8.56 billion in revenue and $12.80 adjusted EPS, but Intuit shares are down more than 40% this year.
CEO Sasan Goodarzi insisted the cuts have “nothing to do with AI” and are about simplifying operations. That claim strains credibility when every structural detail points the other direction: the company is actively investing in baking AI into TurboTax, QuickBooks, and Credit Karma, and the job categories being eliminated overlap heavily with tasks these AI features are designed to automate.
Intuit joins Meta (8,000 jobs), Salesforce, Microsoft, and Google in the 2026 layoff wave that has now eliminated over 113,000 tech positions while AI infrastructure spending exceeds $725 billion. Affected US employees get 16 weeks base pay plus two weeks per year of tenure, with a final day of July 31.
Source: TechCrunch, CNBC
Agents and Autonomy
OpenAI Codex Can Now Control Your Mac While You Sleep
OpenAI shipped a “Locked Use” feature for Codex on May 22, allowing the agent to continue operating macOS applications even when the screen is locked and the display goes dark. The feature uses an Apple authorization plug-in with temporal and behavioral safeguards, letting developers trigger and monitor long-running tasks remotely from their phone.
Codex now has over 4 million weekly users, and the combination of mobile access, locked-screen operation, and goal-oriented task execution represents a meaningful step toward truly autonomous AI agents. The agent can click through windows, type, navigate menus, and interact with the clipboard in whitelisted applications.
There are guardrails: it can’t automate Terminal, can’t interact with system-level admin prompts, and relocks on local input. The feature is also unavailable in the EEA, UK, and Switzerland — presumably due to regulatory concerns about autonomous software agents operating without direct human supervision.
Quick Hits
-
Magnifica Humanitas full text published: The Vatican released the full 42,300-word text of Pope Leo XIV’s AI encyclical. Key demands include a ban on autonomous weapons, independent AI oversight bodies, and mandatory human control over consequential decisions. Christopher Olah’s presence at the presentation drew pointed attention, with the Anthropic co-founder calling on the Church to help “ensure the gains of AI are shared globally.” CNN, PBS
-
Chinese AI models hit 61% of global API traffic: Chinese providers — led by Xiaomi (21.1% market share), Alibaba, MiniMax, and DeepSeek — now account for the majority of global API traffic on OpenRouter, up from under 2% a year ago. The driver is pricing: Chinese frontier models are 15-30x cheaper than Western peers for comparable workloads. AI in China
-
Anthropic launches Claude for Small Business: A new package bundles Claude with connectors for QuickBooks, PayPal, HubSpot, Canva, Docusign, Google Workspace, and Microsoft 365 — ready-to-run workflows targeting businesses that lack dedicated AI teams. Anthropic
-
Anthropic reaffirms ad-free Claude: While OpenAI expands its self-serve Ads Manager for ChatGPT — removing the $50,000 minimum spend and adding CPC bidding — Anthropic announced Claude will remain ad-free, saying “advertising incentives are incompatible with a genuinely helpful AI assistant.” CNBC
Worth Watching
Anthropic is threading a needle that shouldn’t be possible. The company is simultaneously fighting the US government in court over military AI guardrails, partnering with the Vatican on AI ethics, posting its first profit, and about to close a funding round that would make it the world’s most valuable startup. The market is sending an unambiguous signal: safety-focused AI development is a competitive advantage, not a handicap. Whether that signal holds through IPO scrutiny is the next test — but for now, Anthropic’s bet that you can be principled and profitable appears to be paying off in spectacular fashion.