Subject: OpenAI's $852B Moment + Anthropic Leaks Itself Twice
Hey Operators,
The AI capital race hit a new peak yesterday when OpenAI closed the largest private funding round in Silicon Valley history — $122 billion at an $852 billion valuation, now generating $2B in monthly revenue. But the week's defining irony belongs to Anthropic: just days after accidentally leaking its unreleased Claude Mythos model, the company suffered a second breach, with 500,000 lines of Claude Code source going public via a rogue npm file that was forked over 41,000 times before they could respond. Meanwhile, Microsoft quietly demonstrated that the future of AI products may be multi-model, launching a Copilot feature that puts GPT and Claude to work in tandem. The AI race is accelerating — and the cracks from moving too fast are starting to show.
Operation Check
AI-adjacent mega-caps (NVIDIA, Amazon, SoftBank) got a sentiment lift after OpenAI's record $122B close — expect positive halo effects across AI infrastructure plays today.
Bitcoin holds steady above $90K with broader risk appetite intact; OpenAI's blockbuster round is being read as confirmation that tech capital markets remain wide open.
OpenAI's $2B monthly revenue — growing 4× faster than Alphabet or Meta in their peak growth phases — signals the AI commercialization era is not slowing; the next benchmark to watch is $3B/month by year-end.
Operation Dive
OpenAI Closes $122B Round at $852B Valuation — Largest in Silicon Valley History
OpenAI officially closed i36ts record-breaking funding round yesterday, bringing in $122 billion at a post-money valuation of $852 billion. Amazon led with $50 billion, with NVIDIA and SoftBank each contributing $30 billion — and for the first time, retail investors were invited in, contributing $3 billion through bank channels. ChatGPT now serves 900 million weekly active users, and enterprise customers account for over 40% of the company's $2B in monthly revenue. The Insight: With OpenAI approaching a $1 trillion valuation pre-IPO, the structural bet is clear: whoever controls the leading AI platform at scale controls the next era of computing.
Anthropic Leaks Itself Twice: Claude Code Source Follows the Claude Mythos Exposure
Anthropic's week of security missteps continued on March 31 when Claude Code version 2.1.88 was published to npm with a 59.8MB source map file accidentally bundled in — exposing roughly 500,000 lines of internal TypeScript source code. Within hours, the repository was forked more than 41,500 times. This came just days after a CMS misconfiguration had exposed Claude Mythos, Anthropic's most advanced unreleased model. Anthropic confirmed no customer data or credentials were involved, calling it "a release packaging issue caused by human error." The Insight: Two accidental exposures in one week at one of the world's most security-conscious AI labs reveal how fragile operational hygiene becomes when you're shipping at hypergrowth speed.
Operators in Focus
Grab + WeRide Launch Southeast Asia's First Public Robotaxi Service in Singapore
Today, WeRide and Grab officially launched Singapore's first autonomous public ride service in the Punggol district — an 11-vehicle fleet of GXR robotaxis and Robobuses operating on two approved routes. The Ai.R service has already logged 30,000km of autonomous mileage during a trial period with 1,000+ passengers. Rides will be free until commercial operations begin in mid-2026, and the fleet is powered by NVIDIA DRIVE Hyperion technology. The Insight: Southeast Asia's AV race just got a real public deployment — this is the region's clearest signal, yet that autonomous mobility is graduating from pilot to product.
Oracle Cuts Jobs to Fuel AI Push
Oracle has begun cutting jobs across its business lines, affecting employees in both the U.S. and India, even as it ramps up spending on artificial intelligence infrastructure. The company, which employed about 162,000 people globally as of May, is investing heavily in building out costly data centers to support AI development. Posts on LinkedIn and social media indicate that workers in multiple divisions have had their roles eliminated. The insights: This move highlights a broader trend in tech: companies are reallocating resources from traditional operations toward AI, often at the expense of existing staff.
Operator’s Spotlight read
100+ Baidu Robotaxis Paralyzed in Wuhan: What a Mass AV Failure Looks Like in the Wild
A suspected 'system failure' froze more than 100 Apollo Go autonomous vehicles simultaneously on Wuhan's elevated highways Tuesday night, stranding passengers in stalled cars with heavy traffic on all sides. Police confirmed no injuries, but one school student was trapped for 90 minutes before calling police for help exiting the vehicle. Baidu confirmed it is investigating the cause, describing it as a temporary outage. The incident went viral in China within hours. The Insight: When autonomous systems fail at scale, they don't fail quietly — a single software event can strand 100 vehicles simultaneously, and that's the edge-case safety story that regulators and insurers will be studying closely.
Operator Industry Radar
EU Tightens Rules: No AI in Official Texts→The European Union has banned staff from using AI‑generated images and videos in official communications, citing risks of deepfakes and credibility erosion. While the rule allows technical enhancements like sharpening or resizing visuals, it prohibits synthetic content creation.
Court Docs: Musk Texted Zuckerberg to Co-Fund $97.4B OpenAI Takeover → Newly unsealed documents from the Musk vs. OpenAI lawsuit reveal that in early 2025, Musk asked Zuckerberg to join his $97.4B bid to acquire OpenAI Zuckerberg declined to commit, and OpenAI CEO Sam Altman rejected the offer outright.
LiteLLM Supply Chain Attack Hits AI Recruiting Platform Mercor → Hacking group Lapsus$ exploited a vulnerability in open-source library LiteLLM to breach AI hiring platform Mercor, stealing user data a stark reminder that the AI tooling supply chain is an expanding attack surface.
Was this email forwarded to you? Don’t miss any updates Subscribe to TechWithAdit for sharp, no - noise tech intelligence stay sharp, – Adit