Welcome back to Daily Zaps, your regularly-scheduled dose of AI news ⚡

Here’s what we got for ya today:

Let’s get right into it!

Sign Up | Advertise | Tools

Anthropic has closed a $65 billion Series H round at a $965 billion post-money valuation — surpassing OpenAI's $852 billion valuation for the first time, and marking what may be the company's final private fundraise before an IPO. The round was co-led by Altimeter Capital, Dragoneer, Greenoaks, Sequoia, Coatue, and D1 Capital Partners, with chip giants Samsung, SK Hynix, and Micron joining as strategic infrastructure partners. A $15 billion slice of the raise reflects previously committed investments from hyperscalers, including Amazon's $5 billion tranche announced last month.

Anthropic said its annualized revenue run rate crossed $47 billion earlier this month, and The Wall Street Journal reported this week that the company expects a 130% revenue surge to bring it to its first operating profit. That growth has been driven almost entirely by enterprise demand for Claude Code, which has turned the company into the dominant AI vendor inside large organizations. The fresh capital will go toward safety research, compute expansion, and scaling the Claude platform — with Anthropic hinting that its most powerful model yet, Mythos, is a few weeks away from a broader release.

Official announcement → Anthropic

On the same day as its historic fundraise, Anthropic released Claude Opus 4.8 — just 41 days after Opus 4.7, the fastest major model upgrade the company has ever shipped. The new model arrives with Dynamic Workflows, now in research preview: a system that lets Claude Code orchestrate hundreds of parallel subagents simultaneously, enabling codebase-scale migrations from kickoff to merge in a single session. Anthropic says early testers found Opus 4.8 "more likely to flag uncertainties and less likely to make unsupported claims" — Bridgewater Associates specifically noted it proactively catches errors that other models routinely miss.

The fast release cadence is no accident. OpenAI's Codex and Google's Gemini 3.5 Flash both launched in the past month, and Anthropic is clearly matching tempo. Pricing stays flat at Opus 4.7 rates. The company also signaled that the long-awaited Mythos-class models — held back since April over cybersecurity concerns — are moving through final safety reviews and will reach all customers "in the coming weeks." What Anthropic is building toward is not just a better model; it's the full agentic stack, with Claude Code as the execution layer and Opus as the brain running it.

Release notes → Anthropic

Bloomberg published leaked renders this week of Apple's planned AI overhaul for iOS 27, revealing a standalone Siri app designed to compete directly with ChatGPT, Claude, and Gemini. The new Siri will surface from the Dynamic Island rather than from the bottom of the screen, support full chat history, and allow users to upload documents and photos — all the hallmarks of a modern AI assistant. A revamped swipe-down search will route queries through a rebuilt Siri powered by Google's Gemini technology under the hood, presenting results in a card-style interface without opening a browser.

Apple's advantage here is not the model — it's the install base. ChatGPT now has 900 million weekly active users. Apple has 2.5 billion devices. The company is leaning on the same playbook it used with Google Search: partner with the best external AI now, build its own local models in parallel, and let distribution do the heavy lifting. WWDC is June 8 — that's when the full iOS 27 roadmap goes public, and when the world will find out exactly how far Apple is willing to go to make Siri relevant again.

Read more →

Robinhood launched agentic trading this week, letting users connect AI agents — Claude, ChatGPT, or any MCP-compatible system — to a dedicated sub-account loaded with a pre-set budget. The agent can read your full portfolio, analyze concentration risk, scan analyst notes, and execute equity trades autonomously inside that wallet. Users get push notifications for every trade. Some orders require manual approval; a Robinhood fraud team reviews anything suspicious. The company is also launching a virtual credit card for AI agents — 3% cashback, currently for Gold Card holders — so the same agent that manages your stocks can pay your bills.

This isn't a demo or a pilot — it's live for Robinhood's full 27 million customer base, with options, crypto, futures, and prediction markets coming next. Stripe, Amazon, and Google have all launched agentic payment products in the past year, but Robinhood is the first mainstream consumer brokerage to hand an AI agent a real portfolio and let it run. The MCP integration means any developer can plug a custom agent into Robinhood's trading infrastructure today. The age of autonomous AI finance just got a consumer-facing on-ramp.

Full breakdown →