@raunaqn Daily Scan

AI & Robotics Topics Briefing

Wednesday, June 3, 2026 -- What to engage with today

01 Today's Hot Topics

AI Tools Enterprise

Microsoft Bans Claude Code, Forces Engineers Back to Copilot CLI

Microsoft is canceling internal Claude Code licenses for thousands of engineers in its Experiences + Devices division (Windows, 365, Outlook, Teams, Surface) by June 30. The backstory is priceless: Microsoft rolled out Claude Code in Dec 2025 for evaluation, and engineers loved it so much they abandoned Copilot CLI entirely. Now Microsoft is pulling the plug on the competitor it invited in. Meanwhile, GitHub Copilot's new metered billing is driving a separate developer revolt -- users report burning 16% of their monthly Pro+ credits on a single mediocre session.

Source: The Bridge Chronicle, The Register, developer-tech.com -- breaking today
IPO Anthropic

Anthropic Files Confidential S-1 -- The Trillion-Dollar IPO Race Is On

Anthropic submitted a confidential draft S-1 to the SEC on Monday, officially starting its IPO process. This follows the $65B Series H at a $965B valuation and revenue run rate ballooning from $10B to $47B since January. Wedbush calls it "the opening of the floodgates" -- SpaceX already filed (targeting ~$1.75T at NASDAQ debut June 12), and OpenAI is expected to file next. Three potential trillion-dollar-plus IPOs in one year. Fast Company's deep dive ties Anthropic's trajectory directly to Claude Code becoming the "AI killer app" after Opus 4.5 shipped in November.

Source: Fast Company, Inc., The Register, Interesting Engineering -- June 1-2
Robotics OpenAI

OpenAI Launches Robotics Division -- Personal Robots for Everyone

Sam Altman announced OpenAI Robotics on May 31, hiring full-stack hardware, ops, systems, and ML engineers. Led by Aditya Ramesh, the unit grew out of OpenAI's world simulation research. Near-term: robots for skilled infrastructure workers. Long-term: "everyone having a personal robot doing anything they need." This is OpenAI's return to robotics after shutting down Dactyl in 2020. TSLA dropped 3.57% on the news as it threatens Tesla's Optimus positioning. The pitch lands directly on Musk's turf -- right as Tesla is converting its entire Fremont S/X line to Optimus production.

Source: Sam Altman on X, CoinCentral, LatestLY -- May 31
NVIDIA Physical AI

NVIDIA Picks Unitree Over Tesla for Isaac GR00T Humanoid Reference Design

At GTC Taipei/COMPUTEX, Jensen unveiled the Isaac GR00T Reference Humanoid Robot built on Unitree's H2 Plus chassis + Jetson Thor (Blackwell GPU, 2,070 FP4 TFLOPS). The explicit snub of Tesla is the story. Stanford, ETH Zurich, UC San Diego, and Ai2 are initial research partners. Unitree is simultaneously IPO-ing on Shanghai's STAR Market (~$610-620M raise). Other GTC drops: Vera CPU for agents, Cosmos 3 (unified multimodal: text/image/video/audio/actions), RTX Spark for AI PCs, NeMo Neutron 3 Ultra (550B params). Jensen's thesis: "tokens are the new currency."

Source: NVIDIA, RobotsBeat, MoneyCheck, The Street -- June 1-2
Policy Regulation

Trump Signs AI Executive Order -- 30-Day Pre-Release Model Access

Signed privately on June 2 after shelving an earlier 90-day version in May. Creates a voluntary framework for AI companies to submit frontier models for government cybersecurity review up to 30 days before public release. Not mandatory -- no licensing or permitting requirements. David Sacks negotiated the compromise down from 90 days. Altman is in Washington this week meeting White House officials, Mike Johnson, and Hakeem Jeffries. He supports the EO publicly. The real question: does "voluntary" stay voluntary?

Source: WSJ, Reuters, USA Today, The Register -- June 2-3

02 Reply Opportunities

Thread: Microsoft banning Claude Code discussions on X

The "Build vs. Buy Your Own Tools" Irony

Angle: As someone building AI agent infrastructure, this is the most telling data point in the entire AI tools war. Microsoft invited Claude Code in for evaluation and their own engineers defected to it. The lesson isn't about Claude vs. Copilot -- it's about what happens when developer preference and corporate strategy diverge. The engineers who chose Claude Code are voting with their keystrokes on what the actual "AI killer app" looks like. Connect this to the html-docs thesis: the infrastructure layer that developers actually adopt wins, regardless of who owns it.

Thread: Anthropic S-1 and the IPO race

Revenue Growth Tells the Real Story

Angle: $10B to $47B run rate in six months isn't a valuation bubble -- it's Claude Code eating enterprise software budgets alive. Microsoft and Uber blew through their annual AI budgets in months because of Claude usage. The IPO filing is interesting, but the revenue trajectory is the real signal. What does it mean when a single AI coding tool generates more enterprise spend than entire SaaS categories? This is the infrastructure layer repricing itself in real time.

Thread: OpenAI Robotics hiring / NVIDIA Isaac GR00T

The CUDA Playbook for Physical AI

Angle: NVIDIA choosing Unitree over Tesla for Isaac GR00T is doing exactly what it did with CUDA -- creating the software dependency before the market reaches scale. Meanwhile, OpenAI entering robotics with Aditya Ramesh (world simulation background) suggests they'll approach physical AI as an extension of their model stack, not as a hardware company. The next 18 months will determine whether robotics follows the GPU compute model (NVIDIA as platform) or the smartphone model (vertically integrated winners). Both paths need an infrastructure layer.

Thread: Google Antigravity vs. Claude Code vs. Codex comparisons

The Three-Horse Race Has a Clear Hierarchy

Angle: MakeUseOf tested all three on building a Notion clone -- Antigravity's output was a static screenshot where nothing worked, while Claude Code and Codex both shipped functional apps on the first try. The coding agent market isn't a three-horse race yet. It's a two-horse race with a fast follower. The interesting question is whether Gemini 3.5 Flash's cost advantage (4x faster, fraction of the price) matters when the output quality gap is this wide. For now, the answer is clearly no for primary workloads.

03 Post Ideas

Post 1: "Microsoft banned Claude Code internally. Here's what that actually means."

Microsoft gave Claude Code to thousands of engineers for evaluation. They loved it so much they stopped using Copilot CLI entirely. Now Microsoft is revoking licenses by June 30. This is the clearest market signal in the AI coding tools space: when given a choice, engineers overwhelmingly chose the competitor. It's also a case study in the "build vs. buy" trap -- sometimes your own tool isn't the best tool, and your engineers will tell you with their behavior before they tell you with their words.

Hook: "The most expensive product evaluation Microsoft has ever run just backfired spectacularly."

Post 2: "Three trillion-dollar IPOs in one year. The AI infrastructure layer just repriced itself."

SpaceX (filing done, targeting June 12 NASDAQ debut at ~$1.75T). Anthropic (S-1 filed, $965B valuation, $47B revenue run rate). OpenAI (expected to file soon, $852B valuation). We're watching the infrastructure layer of the next computing era go public simultaneously. What makes this different from 2021's IPO wave: these companies have real, fast-growing revenue driven by a single thesis -- AI agents eating everything from code to physical labor. The question for builders: what does the AI agent infra stack look like when these three are public companies with quarterly earnings pressure?

Hook: "We're about to have three public companies valued at $1T+ that didn't exist 5 years ago. All built on the same thesis."

Post 3: "NVIDIA is running the CUDA playbook on humanoid robots, and nobody is paying enough attention."

Jensen picked Unitree over Tesla for the Isaac GR00T reference humanoid design. Stanford, ETH Zurich, and UC San Diego are building on it. Cosmos 3 unifies text/image/video/audio/actions in one model for physical AI. Isaac GR00T is attempting to create the same software lock-in for humanoid robotics that CUDA built for deep learning. If every robotics research lab standardizes on Isaac GR00T, NVIDIA becomes the platform layer for physical AI the same way it became the platform layer for digital AI. Meanwhile, Unitree is IPO-ing at $610M -- the first pure-play humanoid company to go public.

Hook: "CUDA ate AI. Isaac GR00T wants to eat robotics. Here's the playbook."

04 Threads Spotlight

AI Coding Agent Discussion Heating Up on Threads

Threads is seeing active discussion around AI coding tools, with posts from @matt_dancho and @kurtwag8 gaining traction on the Claude Code vs. Codex debate. @obrajesse's post on AI agents (47 likes, 15 comments) was the highest-engagement AI post in the last 24h. The Meta newsroom account (@metanewsroom) also posted about the Anthropic IPO filing (38 likes). The platform is increasingly becoming a venue for dev-focused AI discussion.

Opportunity: Post a Threads-native take on the Microsoft Claude Code ban. The format works well for a short, punchy observation -- "Microsoft gave Claude Code to their engineers as a test. The engineers liked it too much. Now Microsoft is taking it away." Follow up with a comment thread adding the builder-in-public angle. Threads rewards direct, opinionated takes over link-heavy threads.

Physical AI Content Gap on Threads

Robotics content on Threads is significantly underrepresented relative to the news cycle. The NVIDIA Isaac GR00T announcement, OpenAI Robotics launch, and Unitree IPO are getting heavy coverage on X but minimal Threads engagement. This is a clear content gap. Being early to the "physical AI" conversation on Threads positions @raunaqn as a go-to voice on a topic that's about to get much louder.

Opportunity: Post a Threads-first take on the NVIDIA/Unitree partnership. Something like: "NVIDIA just chose a Chinese robotics startup over Tesla for its humanoid robot platform. If Isaac GR00T becomes the CUDA of physical AI, this is the moment it started." Short, visual-friendly, shareable.