Business, Deals & Funding
Claude Code Changelog
v2.1.126
v2.1.126 Release Notes Analysis This changelog entry for Claude Code v2.1.126 highlights three key changes: Dynamic Model Listing from Gateways The `/model` picker now queries the `/v1/models` endpoint when `ANTHROPIC_BASE_URL` is configured to point at an Anthropic-compatible gateway. This means users working through API gateways or proxies can see all available models dynamically, rather than being limited to a hardcoded list. `claude project purge` Command A new CLI command for cleaning up all Claude Code state associated with a project: bash claude project purge [path] This deletes: Transcripts Tasks File history Config entries Supported flags: | Flag | Description | |------|-------------| | `--dry-run` | Preview what would be deleted without actually deleting | | `-y` / `--yes` | Skip confirmation prompts | | `-i` / `--interactive` | Interactively select what to purge | | `--all` |…
Why it matters
I'm watching Claude Code mature its project management capabilities with the new `claude project purge` command, which signals Anthropic is taking workspace hygiene seriously for power users. The dynamic model listing from gateways also tells me enterprise and proxy-based deployments are becoming a bigger priority.
TechCrunch AI
Replit’s Amjad Masad on the Cursor deal, fighting Apple, and why he’d rather not sell
Replit's Amjad Masad on the Cursor Deal, Fighting Apple, and Why He'd Rather Not Sell Summary In a TechCrunch StrictlyVC interview from May 2026, Replit CEO Amjad Masad discussed several major topics: the reported $60 billion acquisition of rival Cursor by SpaceX, Replit's own explosive growth, its battle with Apple over the App Store, and the company's desire to remain independent. Key Points Replit's Explosive Growth Replit went from $2.8 million in revenue in all of 2024 to tracking toward a billion-dollar annual run rate Net revenue retention is reaching as high as 300%, indicating massive expansion of spending by existing customers The company has been gross margin positive for over a year The Cursor/SpaceX Deal Cursor is reportedly in talks to be acquired by SpaceX for $60 billion Masad noted that Cursor reportedly has negative 23% gross margins, making independence difficult He a…
Why it matters
I'm watching Replit's trajectory closely as it races toward a billion-dollar run rate with jaw-dropping 300% net revenue retention, which signals just how deeply AI coding tools are embedding themselves into developer workflows. The contrast with Cursor's reported negative gross margins is a useful reminder that hypergrowth and financial sustainability are very different things in this space.
TechCrunch AI
Meta buys robotics startup to bolster its humanoid AI ambitions
Summary Meta has acquired Assured Robot Intelligence (ARI), a humanoid robotics startup, for an undisclosed sum. ARI's team, including co-founders Xiaolong Wang (former Nvidia researcher and UC San Diego professor) and Lerrel Pinto (former NYU professor and co-founder of Fauna Robotics, recently acquired by Amazon), will join Meta's Superintelligence Labs research division. Key Details What ARI does: Built foundation models for humanoid robots to perform physical labor tasks like household chores, enabling robots to understand, predict, and adapt to human behaviors in complex environments. Strategic rationale: The acquisition bolsters Meta's longstanding humanoid robotics ambitions. A leaked memo from approximately a year prior revealed Meta's plans to build consumer-facing humanoid robots, including both AI models and hardware. Broader AI thesis: Many AI experts believe achieving artif…
Why it matters
I'm watching Meta quietly assemble a world-class robotics research bench by poaching top academic talent, which tells me their humanoid ambitions are far more serious and near-term than their public messaging lets on. The race to own the physical AI layer is accelerating fast.
MIT Tech Review AI
Musk v. Altman week 1: Elon Musk says he was duped, warns AI could kill us all, and admits that xAI distills OpenAI’s models
Musk v. Altman Trial: Week 1 Summary Overview The first week of the landmark Musk v. OpenAI trial took place at the federal courthouse in Oakland, California, with Elon Musk taking the stand to argue that Sam Altman and Greg Brockman deceived him into funding what he believed was a nonprofit dedicated to safe AI development. Key Developments Musk's Core Argument Musk testified that he was "a fool who provided them free funding to create a startup," claiming he donated $38 million to what he understood was a nonprofit developing AI for humanity's benefit — not to enrich its executives. He's asking the court to: Remove Altman and Brockman from their roles Unwind OpenAI's for-profit restructuring The Bombshell Admission In what drew audible gasps in the courtroom, Musk admitted that his own AI company xAI (maker of the chatbot Grok) uses OpenAI's models to train its own — a practice known…
Why it matters
I'm watching this trial closely because Musk's admission that xAI distills OpenAI's models while simultaneously suing OpenAI for going commercial is a stunning contradiction that could undermine his entire case. The broader question of whether AI labs can be held to founding nonprofit missions is one that will shape how the industry structures itself going forward.
Ars Technica AI
Ubuntu infrastructure has been down for more than a day
Analysis of This Article This article, dated May 1, 2026, describes a significant cybersecurity incident affecting Ubuntu/Canonical infrastructure. Let me break down the key elements: What Happened DDoS Attack: Ubuntu and Canonical's web infrastructure was knocked offline by a sustained distributed denial-of-service attack, lasting over 24 hours. Attribution: A pro-Iranian government group claimed responsibility, allegedly using a stressor/booter service called "Beam." Timing: The outage coincided with the disclosure of a critical Linux privilege escalation vulnerability that grants root access to untrusted users across virtually all Linux distributions — making the timing particularly damaging. Why It Matters Security communication was hampered: Ubuntu couldn't distribute security guidance about the critical root vulnerability through its normal channels Key infrastructure was down: `s…
Why it matters
I'm watching how threat actors are increasingly timing DDoS attacks to coincide with critical vulnerability disclosures, effectively silencing the very channels needed to distribute patches and guidance. This case is a stark reminder that infrastructure resilience isn't just an uptime issue — it's a core part of security response capability.
TechCrunch AI
Did you know you can’t steal a charity? Don’t worry. Elon Musk will remind you.
Summary of the Article This TechCrunch article previews an episode of the Equity podcast covering several major tech stories, with the headline story being Elon Musk's lawsuit against OpenAI. Key Points The Musk vs. OpenAI Trial Elon Musk spent three days on the witness stand testifying in his lawsuit against OpenAI His core argument: Sam Altman betrayed the original nonprofit mission by converting OpenAI to a for-profit model Musk's recurring courtroom refrain: "You can't steal a charity" Emails, texts, and Musk's own tweets are being surfaced as evidence The trial is described as "getting messy," with more witnesses (including Altman) still to come Other Topics Covered in the Episode Cloud computing as the big winner of earnings week — AWS, Google, and Microsoft's numbers revealing where enterprise AI spending is actually going A scholarship app founder suing Sallie Mae after they acq…
Why it matters
I'm watching the Musk vs. OpenAI trial closely as it gets messier by the day, with his own emails and tweets being used against him in court. The deeper question I'm tracking is whether a nonprofit's founding mission can legally constrain a for-profit pivot — and what that means for AI governance broadly.
Guardian AI
Pentagon inks deals with seven AI companies for classified military work
Pentagon Signs AI Deals with Seven Companies for Classified Military Work This article from The Guardian (dated May 1, 2026) reports that the Pentagon announced agreements with seven AI companies for classified military applications: SpaceX OpenAI Google Nvidia Reflection Microsoft Amazon Web Services (AWS) Key Details The companies agreed to "any lawful use" of their technology by the U.S. military. The deals are aimed at accelerating military transformation and integrating AI into classified defense operations. Notable Exclusion Anthropic (the company that developed me, Claude) was not included in these agreements. The article indicates this was due to an ongoing dispute between Anthropic and the Pentagon over concerns about potential AI misuse. Context and Significance This represents a significant milestone in the relationship between Silicon Valley and the U.S. defense establishmen…
Why it matters
I'm watching how the major AI players are lining up to embed their technology directly into classified military operations, with the "any lawful use" clause signaling just how broad that access could be. The notable holdout here is Anthropic, and their dispute over potential misuse concerns is a tension worth tracking closely as the rest of the industry moves deeper into defense work.
TechCrunch AI
Pentagon inks deals with Nvidia, Microsoft, and AWS to deploy AI on classified networks
Analysis of This Article Key Facts This article, dated May 1, 2026, reports that the Pentagon has signed deals with Nvidia, Microsoft, AWS, and Reflection AI to deploy AI technology on classified military networks (Impact Level 6 and 7 environments). This follows earlier agreements with Google, SpaceX, and OpenAI. Notable Elements The Anthropic Dispute: The most significant backdrop is the reported ongoing legal battle between the Pentagon and Anthropic. According to the article: The Pentagon wanted unrestricted use of Anthropic's AI tools Anthropic insisted on guardrails against domestic mass surveillance and autonomous weapons Anthropic won an injunction in March 2026 against being branded a "supply-chain risk" The DOD is explicitly diversifying vendors in response Scale of Adoption: The article states over 1.3 million DOD personnel have used GenAI.mil, the Pentagon's secure generativ…
Why it matters
I'm watching how the Pentagon's vendor diversification strategy is essentially a direct response to Anthropic drawing a hard line on autonomous weapons and surveillance guardrails — a rare case of an AI company successfully pushing back on a major government client. With 1.3 million DOD personnel already on GenAI.mil, the scale of classified AI deployment is moving faster than most public discourse acknowledges.
From X/Twitter
- Manus AI launched Cloud Computer, a persistent virtual environment designed to run autonomous agents and workflows 24/7.
- Ora's first agent readiness report found that only 1% of the web is actually usable by AI agents — even on household-name sites.
- Anthropic shipped Claude Security into public beta, using Opus 4.7 to autonomously scan codebases for vulnerabilities with far fewer false positives than traditional scanners.
- People are using NotebookLM to mass-produce specialized Claude skill files in minutes, turning curated sources into reusable expert behaviors.
- Andrej Karpathy built LLM Council over a weekend: GPT, Claude, Gemini, and Grok all critique each other's answers while a "Chairman" AI synthesizes the best response.
- Dan Shipper says Marcus went from product manager to shipping like a madman — and wrote the definitive guide to building with coding agents.
From Reddit/HN/YC
- [Hacker News] The hype around Java 21 virtual threads and AI is only half-true — the real benchmarks tell a messier story.
- [Hacker News] Microsoft now says 32 GB RAM is the "no worries" baseline for Windows 11.
- [Hacker News] Pulsarbot gives an AI agent its own full computer in the cloud — browser, filesystem, and all.
- [Hacker News] MIT Technology Review surveys the state of cyber-insecurity in the AI era — and it's not reassuring.
- [Hacker News] Apple quietly kills the 256 GB option — the Mac Mini now starts at $799 with 512 GB.
- [Hacker News] A solar farm design lets cattle roam under moving panels, blending agriculture and energy on the same land.