Anthropic’s funding round underscores AI agents as near‑critical infrastructure

What changed: Anthropic is close to closing a $30 billion funding round that would value the company at over $900 billion, surpassing OpenAI’s last reported valuation and marking a dramatic jump from roughly $380 billion just a few months ago. The company is also projecting its first quarterly operating profit, with Q2 revenue expected to hit $10.9 billion (up from $4.8 billion in Q1), and has committed to paying SpaceX about $1.25 billion per month through May 2029 for GPU compute under a $45 billion contract.

Why it matters: Capital at this scale signals that frontier models—and the agentic systems built on top of them—are being treated as long-term, strategic infrastructure rather than experimental tools. For anyone building or buying AI agents, this suggests that the leading model providers are unlikely to be short‑lived vendors and that competition between Anthropic and OpenAI will keep pushing capabilities and pricing.

Try/watch: Treat your primary model provider like a cloud or payments dependency: negotiate longer‑term commitments where possible, develop a backup provider for critical agent workflows, and keep an eye on how Anthropic’s new capital shows up in better tools, latency, and reliability for agents.

US postpones AI oversight order, leaving agentic AI largely self‑regulated (for now)

What changed: A planned US executive order that would have asked companies to voluntarily submit new AI models for government review up to 90 days before release—aimed at assessing security risks—has been abruptly postponed. Reporting indicates the signing was canceled after direct calls from several tech CEOs, with the president warning that the order could undermine US competitiveness in AI.

Why it matters: With no new federal review process, companies deploying powerful AI agents retain wide latitude over how they test, secure, and release systems that can act autonomously in sensitive domains. This keeps regulatory friction low for builders, but also increases the burden on internal governance, red‑teaming, and safety practices.

Try/watch: If you’re rolling out agentic systems (for example, tools that can access production data, financial systems, or customer accounts), assume regulators may later scrutinize today’s decisions; adopt voluntary pre‑release reviews, incident logging, and alignment checks now so you’re not retrofitting governance under pressure.

‘AI interns’ frame the next wave of digital employees in offices

What changed: A new column from Korea JoongAng Daily describes how companies are preparing for “AI interns” in the workplace—agentic systems that function as digital employees handling routine office work. The piece notes that Google is reportedly developing an AI agent ahead of its annual conference and argues that managing these systems so they don’t threaten corporate security mirrors the challenge of onboarding human staff with the right access controls.

Why it matters: Framing agents as interns or junior staff helps organizations think concretely about scope, supervision, permissions, and performance expectations. It also reinforces that security and governance for agents must be designed, not assumed—especially when they are given access to email, documents, or internal tools.

Try/watch: Pilot “AI intern” roles in clearly bounded functions—such as meeting-note drafting, simple report assembly, or ticket triage—while enforcing role‑based access, approval workflows for sensitive actions, and clear human escalation paths.