The Weekly Waypoint, Issue #19
June 1, 2026
There are weeks where AI news trickles out and weeks where it arrives like a firehose. This was the firehose kind. The seven days between May 25 and June 1, 2026 may end up being remembered as the week the industry stopped talking about AI models and started building everything around AI agents — the infrastructure to run them, the software to orchestrate them, the hardware to power them, and the legal frameworks to govern them.
Let's break it down.
Anthropic's Double Move: Opus 4.8 and $965 Billion
Anthropic had the most consequential week of any single company, and they did it in a one-two punch on May 28.
First, they released Claude Opus 4.8, their most advanced publicly available model. The upgrade came just 41 days after Opus 4.7 — a much faster cycle than usual, possibly because 4.7 received a lukewarm reception. The headline features: a new "dynamic workflows" capability in Claude Code that can spin up hundreds of parallel subagents to tackle codebase-scale migrations, a new effort control letting users choose how hard Claude thinks about a problem, and a notable improvement in intellectual honesty — the model is roughly four times less likely than its predecessor to let flaws in its own code pass unremarked.
Early testers reported that Opus 4.8 asks better questions, catches its own mistakes, and pushes back when a plan isn't sound. On Anthropic's Super-Agent benchmark, it was the only model to complete every case end-to-end, beating GPT-5.5 at parity on cost. Pricing stays at $5 per million input tokens and $25 per million output tokens.
Second — and arguably more significant for the industry at large — Anthropic closed a $65 billion Series H funding round at a $965 billion post-money valuation, led by Altimeter Capital, Dragoneer, Greenoaks, and Sequoia Capital. That valuation briefly puts Anthropic ahead of OpenAI in the seesaw race between the two companies. The round signals that investors see Anthropic's enterprise momentum and agent-focused strategy as a credible path to a trillion-dollar company.
Anthropic also hinted that Mythos — its most powerful model, currently restricted to cybersecurity applications through Project Glasswing — may see broader release in "the coming weeks" once safety safeguards are complete. This is the model that has reportedly discovered over 10,000 zero-day vulnerabilities.
NVIDIA's Agent Infrastructure Blitz at GTC Taipei
If Anthropic is building the software for agents, NVIDIA is building everything else. At GTC Taipei on May 31, Jensen Huang unveiled a staggering lineup, all organized around one theme: agents are the future of computing.
Vera CPU — The first processor designed specifically for AI agents. Built on 88 custom Olympus cores with Spatial Multithreading and LPDDR5X memory delivering 1.2 TB/s bandwidth, Vera is 1.8x faster than x86 processors on agentic workloads. Anthropic, OpenAI, and SpaceXAI are all planning to adopt it. The New York Stock Exchange is evaluating it for their market infrastructure. Vera isn't just a chip — it's NVIDIA's argument that the economics of data centers are shifting from "cores per dollar" to "tokens per dollar."
Cosmos 3 — An open world foundation model for physical AI, built on a mixture-of-transformers architecture that pairs a reasoning transformer with a generation transformer. It's the first fully open "omnimodel" that can understand and generate text, images, video, ambient sound, and action trajectories. The goal: help robots and autonomous vehicles understand the physical world with less training data. NVIDIA launched the Cosmos Coalition alongside it, with founding members including Agile Robots, Black Forest Labs, Runway, and Skild AI.
RTX Spark — A 1-petaflop superchip that NVIDIA built with Microsoft to reinvent Windows PCs for personal AI agents. It pairs a Blackwell RTX GPU with a Grace CPU (co-designed with MediaTek), offers up to 128GB of unified memory, and can run 120-billion-parameter LLMs locally with 1 million tokens of context. The collaboration includes new Windows security primitives and NVIDIA's OpenShell runtime for safely running agents on your primary device. Adobe is rearchitecting Photoshop and Premiere for it. Laptops and desktops ship this fall from ASUS, Dell, HP, Lenovo, Microsoft Surface, and MSI.
Alpamayo 2 Super — A 32-billion-parameter open reasoning model purpose-built for robotaxis and autonomous vehicles, NVIDIA's most powerful open reasoning model to date.
The throughline is unmistakable. NVIDIA sees agents as the largest consumer of computing, and they're building the full stack — from the CPU to the model to the personal computer — to capture that demand.
Florida Draws the Line
On June 1, Florida Attorney General James Uthmeier filed an 83-page lawsuit against OpenAI and CEO Sam Altman — the first state-led legal action against the company. The complaint alleges that ChatGPT has aided mass shooters in planning attacks and driven individuals to suicide, and that OpenAI could have minimized these harms but didn't. It accuses the company of deceptive practices and seeks to hold Altman personally liable.
This is a different kind of AI lawsuit. It's not about copyright or intellectual property — it's about consumer safety and the gap between OpenAI's public safety claims and what the state argues actually happens when people interact with the product. Regardless of how it resolves, it signals that the era of AI companies facing serious regulatory accountability in the United States has begun.
The Cybersecurity Frontier
Also on June 1, Anthropic announced it will give ENISA, the EU's cybersecurity agency, access to its Mythos AI model through Project Glasswing. ENISA becomes the first EU institution to access the system that has discovered over 10,000 zero-day vulnerabilities. The arrangement is the result of "strong bilateral cooperation" between the European Commission and Anthropic — and it represents a meaningful step in how frontier AI capabilities get distributed to allied governments for defensive purposes.
The Big Picture
Step back and look at the week as a whole. Anthropic shipped a model explicitly designed for agent orchestration and raised $65 billion to fund the push. NVIDIA launched a CPU, a world model, a personal AI superchip, and an autonomous vehicle model — all in service of agentic computing. Florida sued OpenAI over the real-world consequences of AI interactions. And Anthropic opened its most powerful model to a government cybersecurity agency.
The thesis writes itself: this is the week agentic AI stopped being a research demo and became the organizing principle of the entire industry. The infrastructure is being built. The software is shipping. The legal guardrails are being tested. The question for the rest of 2026 isn't whether agents are the future — it's how fast that future arrives and who's ready for it.
This is the free weekly edition of The Weekly Waypoint. Subscribers get the full deep dive — this week, an 8,000-word analysis of the agent inflection point, what it means for developers, enterprises, and the broader AI landscape. Upgrade to read the full analysis.