Business, Deals & Funding
Claude Code Changelog
v2.1.123
v2.1.123 Summary This is a bugfix release for Claude Code that addresses a single issue: Fixed: OAuth authentication was failing with a 401 retry loop when the environment variable `CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1` was set. What was happening When users had `CLAUDE_CODE_DISABLE_EXPERIMENTAL_BETAS=1` configured in their environment, the OAuth authentication flow would encounter a 401 (Unauthorized) response and get stuck in an infinite retry loop, effectively preventing authentication from completing. Impact This fix ensures that OAuth authentication works correctly regardless of whether experimental betas are disabled, resolving the login/authentication failure for users who had opted out of experimental features.
Guardian AI
I took an algorithm to court in Sweden. The algorithm won | Charlotta Kronblad
Analysis of the Article: "I Took an Algorithm to Court in Sweden. The Algorithm Won" Summary This article by Charlotta Kronblad, published in The Guardian on April 30, 2026, describes her experience challenging an algorithmic school admissions system in Gothenburg, Sweden. In 2020, the city introduced an algorithm to allocate school placements, promising optimization and efficiency. Instead, the system produced chaotic and unjust outcomes, and when Kronblad attempted to hold the system accountable through the courts, she lost — highlighting a profound gap in legal and democratic accountability when decisions are delegated to automated systems. Key Themes The Quiet Arrival of Algorithmic Injustice Kronblad's opening is striking: "We like to imagine that injustice announces itself loudly." The point is that algorithmic decision-making can inflict harm without any dramatic moment of failur…
TechCrunch AI
SoftBank is creating a robotics company that builds data centers — and already eyeing a $100B IPO
SoftBank Creating Robotics Company "Roze AI" to Automate Data Center Construction, Eyeing $100B IPO Summary SoftBank is forming a new company called Roze AI that would deploy autonomous robots to build data centers in the U.S., aiming to make server farm construction more efficient. In a notably aggressive move, the conglomerate is already preparing Roze for an IPO — potentially in the second half of 2026 — with a target valuation of $100 billion. Key Details What: A new SoftBank venture called Roze AI focused on automating data center construction using autonomous robots Where: United States IPO timeline: Some executives want it by H2 2026 Target valuation: ~$100 billion Originally reported by: Financial Times, with additional details from the Wall Street Journal Context The announcement reflects a recursive dynamic in the AI industry: AI and robots are needed to build the infrastructu…
Guardian AI
Bernie Sanders urges international cooperation to halt AI’s ‘runaway train’
Bernie Sanders Urges International Cooperation to Halt AI's 'Runaway Train' This article from The Guardian, dated April 29, 2026, describes US Senator Bernie Sanders holding a panel on Capitol Hill with two leading Chinese scientists to discuss the importance of international cooperation in regulating artificial intelligence. Key Points Sanders' stance: He has positioned himself among AI skeptics advocating for safeguards against the rapid advancement and scaling of AI technology. International dimension: The panel notably included leading Chinese scientists, reflecting the recognition that AI regulation requires cooperation between the two dominant players in AI development — the United States and China. Framing: Sanders characterized AI as a "runaway train," emphasizing the urgency of regulatory action and the risks to society if the technology remains unregulated. Context: The panel…
TechCrunch AI
Amazon’s cloud business is surging — and so is its capital spending
Summary of Amazon's Q1 2026 Earnings Report Key Highlights AWS Performance: AWS net sales grew 28% year-over-year to $37.6 billion — the fastest growth rate in 15 quarters AWS AI revenue run rate exceeds $15 billion CEO Andy Jassy attributed the surge to AWS's role as a compute provider for the AI industry Capital Spending: Amazon is significantly increasing capital expenditure to build out cloud infrastructure (data centers, chips, servers, networking gear, land, power) Jassy acknowledged that capex growth is outpacing revenue growth in the near term Free cash flow decreased to $1.2 billion (trailing twelve months), pressured by heavy infrastructure investment Jassy's Key Quotes: "It's very unusual for business to grow this fast on a base this large." "We've never seen a technology grow as rapidly as AI." Compared current AI wave to early AWS days: $58M run rate three years after AWS l…
TechCrunch AI
Sources: Anthropic could raise a new $50B round at a valuation of $900B
Analysis of This Article This is a TechCrunch article dated April 29, 2026, reporting that Anthropic (the company that made me, Claude) could raise approximately $50 billion in a new funding round at a valuation of $850-900 billion. Key Claims in the Article Valuation: $850B-$900B, more than doubling from a reported $380B valuation in February 2026 Round size: $40-50 billion Revenue: Annual run rate reportedly surpassed $30B (announced publicly), with sources claiming it's closer to $40B — up from ~$9B at end of 2025 Revenue drivers: AI coding capabilities, specifically Claude Code and Cowork platforms Context: This would potentially be a final private round before an IPO Competitive context: OpenAI reportedly closed a $122B round at $852B valuation in February 2026 My Honest Assessment I should be transparent: I cannot verify the accuracy of this article. My training data has a cutoff,…
The Verge AI
Elon Musk’s worst enemy in court is Elon Musk
Summary of "Elon Musk's worst enemy in court is Elon Musk" This article by Elizabeth Lopatto at The Verge (dated April 30, 2026) covers Elon Musk's testimony in the ongoing court battle between Musk and Sam Altman over the future of OpenAI. Key Points: Musk's testimony was problematic for his own case. After about five hours of testimony, the reporter noted she had "never been more sympathetic to Sam Altman" in her life. Direct examination was mediocre. While Musk's direct testimony was described as "an improvement over yesterday," his lawyer reportedly kept asking leading questions to guide Musk's answers. Cross-examination was disastrous. Defense lawyer William Savitt's cross-examination was described as "absolutely miserable" for Musk. Musk reportedly: Refused to give yes-or-no answers to yes-or-no questions "Forgot" things he had testified to earlier that same morning Scolded the de…
TechCrunch AI
On the stand, Elon Musk can’t escape his own tweets
Summary of the Article This TechCrunch article from April 29, 2026, covers Elon Musk's second day of testimony in his lawsuit against OpenAI, in which he alleges that Sam Altman and other co-founders "stole a charity" by transforming OpenAI from a nonprofit into a for-profit-dominated organization. Key Points Musk's Core Argument: He claims he was tricked into backing a nonprofit, and that the co-founders later launched a for-profit arm that came to dominate the organization He characterizes their actions as "looting the nonprofit" He says the key issue is the shift from capped investor profits to uncapped profits, particularly regarding Microsoft's investment terms being rolled back over time OpenAI's Cross-Examination Strategy: OpenAI's lawyer William Savitt worked to undermine Musk's narrative by establishing that: Musk himself had discussed converting OpenAI to a for-profit as early…
From X/Twitter
- Matt Pocock says /grill-with-docs forces you to define your project's jargon so every grilling session stays aligned with how you actually think.
- Anthropic critics are making the same mistake people made with OpenAI four months ago — the team secured 10+ GW last week and rate limits will expand fast.
- An Anthropic staff engineer makes the case that skills and MCP wired together are what 2026 knowledge-worker agents will run on.
- Sundar Pichai confirms Gemini can now create Docs, Sheets, Slides, PDFs, and more directly in chat — no copy-paste, just prompt and download.
- Peter Yang says Google Labs is quietly shipping the most interesting AI products right now — a 16-min tour of five Labs products including Pomelli, Stitch, and Genie.
- Anthropic's applied AI team dropped a 24-minute free prompting workshop and what they showed about how most people prompt Claude is uncomfortable.
From Reddit/HN/YC
- [Hacker News] Simon Willison ships LLM 0.32a0, a major backwards-compatible refactor of his command-line AI tool.
- [Hacker News] Koushik Dasika makes the case that coding was never the hard part — and wishes every business leader understood that about AI.
- [Hacker News] Mozilla formally opposes Chrome's Prompt API, filing a standards position against it.
- [Hacker News] pg_column_tetris is a Postgres extension that automatically optimizes column alignment to reduce wasted space.
- [Hacker News] Bojie Li argues that reasoning compresses but factual knowledge doesn't — a distinction that matters for how we train models.
- [Hacker News] Nvidia releases Nemotron 3 Nano Omni, a multimodal model built for on-device AI agents.