Retter Reboot #23

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

1. Anthropic Files for IPO at $965B — Beats OpenAI to Wall Street

On June 1, Anthropic confidentially filed its S-1 with the SEC, setting up one of the largest tech IPOs ever and beating rival OpenAI to the public markets. The filing follows a fresh $65B Series H round (closed May 28) that valued the company at $965 billion — surpassing OpenAI's $852B for the first time and making Anthropic the most valuable AI startup in history. Annualized revenue hit ~$47B in May, up roughly 5x from $10B in December, driven by Claude Code and enterprise demand. A listing is targeted for October, with a trillion-dollar debut now the base case if markets cooperate. Filing first lets Anthropic set the narrative for the entire category.

Source: Fortune

2. Microsoft Launches Its Own MAI Models at Build — Cutting the OpenAI Cord

At Build 2026 on June 2, Microsoft unveiled seven in-house AI models under the MAI brand, anchored by MAI-Thinking-1 — its first reasoning model (35B active params, 256K context, sparse mixture-of-experts). Critically, it was trained from scratch with zero distillation from OpenAI, Anthropic, or anyone else — a data-lineage pitch aimed at enterprise buyers. It ships alongside MAI-Code-1-Flash, a 5B coding model already rolling out to every paying GitHub Copilot user. The subtext is loud: after $13B+ invested in OpenAI, Microsoft now has its own frontier reasoning model, its own coding model, and a credible path off OpenAI's bill.

Source: Yahoo Finance

3. OpenAI Turns Codex Into an Enterprise Work Platform — Not Just for Coders

On June 2 at its "Intelligence at Work" event, OpenAI repositioned Codex from a coding agent into a general enterprise platform. Alongside the new Sites feature, it shipped Annotations (in-place editing across docs, sheets, and slides) and six role-specific plugins bundling 62 business apps (Snowflake, Figma, Salesforce) with 110 automated skills. Codex now has 5M+ weekly users — and non-developers (analysts, marketers, operators) are 20% of the base and growing 3x faster than engineers. Codex is also coming directly into the ChatGPT app in the coming weeks. The play: make Codex the operating system for knowledge work, not just code.

Source: VentureBeat

4. Three Trillion-Dollar IPOs Are Now Racing to Market in 2026

Anthropic's filing kicked off what Wedbush analysts called "an opening of the floodgates" for a dormant IPO market. SpaceX (merged with xAI earlier this year) is targeting a $2 trillion valuation and seeking $75B+ in its own offering later this month. OpenAI is expected to file its confidential S-1 within weeks, working with Goldman Sachs and Morgan Stanley toward a fall listing. AI chipmaker Cerebras already listed two weeks ago. For the first time, public markets will get audited financials and real margin scrutiny on near-trillion-dollar AI valuations.

Source: Radical Data Science

5. Top AI Labs Are Now Formally Researching Machine Consciousness

The Financial Times reported on June 3 that Google DeepMind, Anthropic, and Meta have all expanded research into machine consciousness and AI welfare — hiring philosophers, psychologists, and ethicists to assess whether models could have morally relevant experiences. Anthropic has reportedly been testing its models for behaviors resembling "panic" and "anxiety" and is pursuing formal model-welfare research. No lab claims current systems are conscious, but the question is being treated as a serious open research problem for future systems rather than science fiction.

Source: Financial Times (via TipRanks)

6. The Flat-Fee Era Ends: GitHub Copilot Switches to Token-Based Billing

On June 1, GitHub retired Copilot's flat monthly subscription and moved every plan to usage-based billing — replacing Premium Request Units with GitHub AI Credits consumed by actual token usage (input, output, and cached context). Base prices hold ($10/mo Pro includes $10 in credits), and code completions stay unlimited, but token-heavy agentic coding sessions can burn $30–$40 in credits, and some heavy users warn of 10x–50x cost jumps. The backlash is loud, but the signal is bigger than one product: AI coding is leaving its all-you-can-eat promotional phase and entering metered, managed-infrastructure economics. The honest question it forces: was AI coding ever cheap, or was someone else eating the cost?

Source: GitHub Blog

7. OpenAI Ships "Sites" — Build and Host Internal Apps From a Single Prompt

The standout from OpenAI's June 2 launch is Sites: describe an internal tool in plain English and Codex builds, deploys, and hosts it on OpenAI infrastructure with a shareable workspace URL — no Vercel project, no hosting account, no DevOps ticket. The framing is deliberate: not a Squarespace or Webflow competitor for public sites, but a killer of the homegrown internal apps every company builds and abandons (dashboards, project trackers, review rooms, scenario planners, KPI boards). It handles storage, environment variables, and access control natively, with Sign in with ChatGPT auth baked in. Currently in preview for ChatGPT Business and Enterprise, with broader rollout in the coming weeks. The trade-off: apps are hosted by OpenAI with limited code portability — for full code export, Lovable or Replit remain stronger.

Source: OpenAI

Jun 5, 2026