Retter Reboot #24

The Retter Reboot : Your weekly dose of ai & tech news

1. SpaceX Goes Public in the Largest IPO in History — SPCX Trades Today

SpaceX priced its IPO at $135 a share on June 11 and opened on the Nasdaq under the ticker SPCX today, June 12 — raising about $75 billion at a $1.75 trillion valuation. That obliterates the previous record (Saudi Aramco's $29.4B in 2019) and makes SpaceX larger than Tesla on day one. The book was roughly twice oversubscribed, and in a rare move ~30% of the float went to retail investors via Robinhood, Fidelity, and Schwab. Starlink is the profit engine (10.3M subscribers, $11.4B in 2025 revenue, 63% EBITDA margin); the absorbed xAI division (now SpaceXAI) posted a $6.36B operating loss. MSCI fast-tracks early index inclusion on June 13, forcing passive funds to buy. Senator Elizabeth Warren asked the SEC to delay over governance concerns; it went ahead anyway.

Source: Kiplinger

2. Anthropic Launches Claude Fable 5 — Its First Public "Mythos-Class" Model

On June 9, Anthropic released Claude Fable 5, the first publicly available model in a new Mythos-class tier that sits above Opus 4.8. Anthropic calls it the most capable model it has ever made generally available — state-of-the-art on nearly all tested benchmarks, with independent testing around 80% on SWE-Bench Pro. It's priced at $10/$50 per million input/output tokens (2x Opus 4.8) under the API id claude-fable-5, and is free on Pro, Max, Team, and Enterprise plans through June 22 before moving to credits. The catch: in high-risk areas (cybersecurity, biology, chemistry, distillation) it blocks and falls back to Opus 4.8, triggering on under 5% of sessions. The unrestricted sibling, Mythos 5, stays locked to Project Glasswing partners.

Source: TechCrunch

3. Apple Rebuilds Siri on Google Gemini at WWDC — But EU Users Are Locked Out

At WWDC on June 8, Apple unveiled Siri AI, the biggest overhaul in the assistant's history — a dedicated app, on-screen awareness, expanded Visual Intelligence, and a Google Gemini backbone under Apple's privacy infrastructure. The asterisk: EU iPhone and iPad users won't get it at launch on iOS 27 / iPadOS 27. Apple blames the Digital Markets Act's interoperability rules; the European Commission fired back on June 10 that the block was Apple's choice, noting Apple simply requested an 18-month blanket exemption that Brussels rejected. EU Mac, Watch, and Vision Pro users still get the features. The episode is a preview of how fragmented AI rollouts are about to become along regulatory lines.

Source: SiliconANGLE

4. Oracle Posts a $638B Backlog as Cloud Revenue Jumps 93%

Oracle reported fiscal Q4 2026 results on June 10, revealing a staggering $638 billion remaining performance obligation (backlog), with OCI cloud-infrastructure revenue up 93% year-over-year to $5.8B and total cloud up 47% to $9.9B. Growth has accelerated every quarter of the year (55% → 68% → 84% → 93%). Yet the stock fell: most of that backlog rides on a handful of mega-deals — most notably a reported $300B agreement with OpenAI — against a record $153B debt load and ~$50B in annual capex. Oracle has effectively become the landlord of the AI infrastructure boom, with all the concentration risk that implies.

Source: Motley Fool

5. OpenAI Comes to Oracle Universal Credits — Frictionless Enterprise Distribution

On June 11, OpenAI announced that enterprise customers can now access its frontier models and Codex through their existing Oracle Universal Credits (UCM) on Oracle Cloud Infrastructure. The significance is procurement, not technology: companies with multi-year, pre-negotiated Oracle commitments — often tens or hundreds of millions of dollars — can route AI workloads to OpenAI models without standing up a new vendor relationship. It deepens the OpenAI–Oracle alliance that began with the $500B Stargate project, and it's a sharp reminder that in enterprise AI, distribution and ease of buying increasingly beat raw model benchmarks.

Source: Build Fast with AI

6. Anthropic's Safety Paradox: A Pause Warning Meets a $35B Compute Deal

The same week it shipped its most powerful public model, Anthropic sharpened its safety message — and exposed the tension at the heart of it. On June 10, Dario Amodei argued governments should be able to block frontier systems judged too dangerous, building on the June 4 "When AI Builds Itself" essay warning that AI is nearing recursive self-improvement (over 80% of code merged into Anthropic's own codebase is now written by Claude). Hours later, a financing platform involving Apollo, Blackstone, and Broadcom unveiled a ~$35B deal giving Anthropic over 1 gigawatt of dedicated compute. The critique writes itself: you can't credibly call for a brake pedal while signing the largest compute deals in your history — unless the slowdown is collective and verifiable, which is exactly Anthropic's argument.

Source: Scientific American

7. Stripe Rebuilds a 50-Million-Line Codebase in a Single Day with Fable 5

The headline proof point from Fable 5's launch wasn't a benchmark — it was Stripe using the model to migrate a 50-million-line codebase in one day, work the company estimated would have taken a team of engineers roughly two months. It's the clearest signal yet that agentic coding has crossed from impressive demo into production-scale infrastructure work. For anyone building on or budgeting around AI engineering, the takeaway is blunt: the unit of work is shifting from "lines a developer writes" to "systems an agent rebuilds," and the timelines are collapsing accordingly.

Source: Medium (Fable 5 coverage)

Jun 12, 2026