2. Anthropic Eyes Custom AI Chips as Revenue Hits $30B Run Rate — Three sources confirm early-stage talks on Anthropic-designed silicon. Annualized revenue tripled since end of 2025; over 1,000 customers now spend $1M+/year on Claude.

3. CoreWeave Inks Multi-Year Deal to Power Claude — Anthropic adds CoreWeave to its compute roster on top of Google TPUs, AWS Trainium, and Nvidia. Stock popped 11%; nine of the top 10 model labs now run on its platform.

Read full report →

2. Claude.ai and Claude Code Hit by Login Outage — A 51-minute authentication failure locked users out of Claude.ai and took Claude Code offline for 15 minutes. Now fully resolved.

3. Anthropic Consulted 15 Christian Leaders on Claude's Moral Future — A two-day summit at Anthropic HQ covered grief, self-harm responses, and whether Claude could be a "child of God."

Read full report →

2. "Claude Mania" Sweeps HumanX Conference — 6,500 attendees and one consensus: Anthropic has stolen the spotlight from OpenAI. Claude Code was the tool on everyone's lips.

3. Claude Code 2.1.101 Ships Team Onboarding — New /team-onboarding command, auto-trust for OS CA certificates, and smarter remote-session setup.

Read full report →

2. Software Stocks Slide 2.6% as Mythos Reignites Disruption Fears — S&P 500 Software Index extends its 2026 decline to 25.5% YTD. Cybersecurity names CrowdStrike, Cloudflare, and Okta fell 5-6.5%.

3. Claude Haiku 3 Retires April 19 — Eight days left to migrate from the original budget model. API calls to claude-3-haiku-20240307 will fail after the deadline.

Read full report →

2. Anthropic Revenue Hits $30B Annualized Run Rate — Up more than 3x from $9B at year-end 2025, with enterprise contracts making up roughly 80% of the business.

3. Advisor Tool Launches in Public Beta — New API feature pairs a fast executor model with Opus 4.6 as a mid-generation advisor for long-horizon agentic tasks.

Read full report →

2. Accenture Signs Multi-Year Partnership With Anthropic — The consulting giant will stand up a dedicated Claude practice and train 30,000 consultants on Anthropic's agentic platform, targeting Fortune 500 rollouts.

3. Snowflake Expands Partnership With $200M Commitment — The expanded deal embeds Claude directly in Snowflake for 12,600+ customers and launches a joint go-to-market for agentic workloads across AWS, Azure, and Google Cloud.

Read full report →

2. Claude Goes Down for Third Consecutive Day — Users reported login failures, chat errors, and degraded performance on April 8, less than 24 hours after the previous major outage.

3. Anthropic Revenue Run Rate Passes $30 Billion — Bloomberg reports annualized revenue has tripled since end of 2025, with over 1,000 enterprise customers spending $1M+ annually.

Read full report →

2. Claude Mythos Enters Limited Early Access With Cybersecurity Warning — Anthropic confirmed its most powerful model yet is being tested with enterprise partners, flagging "unprecedented cybersecurity risks" after an accidental source code leak.

3. OpenClaw Fallout Continues as Devs Scramble for Alternatives — Anthropic's subscription ban on third-party harnesses has sparked community backlash and a migration wave, with OpenClaw's open-source fork seeing a surge in GitHub stars.

Read full report →

2. UK Moves to Woo Anthropic Amid US Defense Clash — Britain is pitching Anthropic on a London expansion and dual LSE listing. The GOV.UK AI assistant, already powered by Claude, anchors the offer.

3. Claude API Gets 300k Token Batches, Capability Fields, Data Residency — A wave of API improvements landed this week, plus model retirements for Sonnet 3.7 and Haiku 3.5. The 1M token beta ends April 30.

Read full report →

2. Claude Code Security Vulnerabilities Stack Up After Source Leak — CVE-2026-21852 let malicious repos steal API keys before trust confirmation. A separate 50+ subcommand bypass evades security analysis.

3. Agent Skills Ecosystem Explodes Past 2,300 Skills — Together AI open-sourced 12 agent skills for Claude Code and Codex. Community marketplaces now list 2,300+ skills and 770+ MCP servers.

Read full report →

2. Message Batches API Now Supports 300k Output Tokens — Claude Opus 4.6 and Sonnet 4.6 unlock 300k output via new beta header. Web search and web fetch graduate out of beta.

3. 1M Context Window Retiring April 30 for Older Models — Claude Sonnet 4.5 and Claude Sonnet 4 lose 1M context on April 30. Migrate to Sonnet 4.6 or Opus 4.6 now.

Read full report →

2. Trojanized Leak Repos Spreading Vidar Malware — Fake Claude Code source repos on GitHub are delivering infostealers and proxy malware. Hundreds of forks before detection.

3. Usage Limits Apology Fuels More Backlash — Anthropic tried to explain why Pro users hit limits faster. The community response made things worse.

Read full report →

2. Claw-Code Hits 100K Stars in One Day — A clean-room Rust rewrite of Claude Code's architecture is now the fastest-growing repo in GitHub history.

3. Claude Buddy Easter Egg Goes Live — Found in the leaked source: an April Fools terminal pet with 18 species, RPG stats, and a 1% shiny rate. Live through April 7.

Read full report →

2. Claude Code v2.1.89 Ships Same Day — New release adds deferred hook permissions, named subagents for @mention typeahead, and non-blocking MCP connection mode.

3. 1M Token Beta Retiring April 30 — The context-1m beta header stops working for Sonnet 4.5 and Claude 4 at month's end. Migrate to Sonnet 4.6 or Opus 4.6 now.

Read full report →

2. Claude Paid Subs More Than Double in 2026 — Credit card data confirms record subscriber growth. Super Bowl ads, Claude Code, and the Pentagon feud all fueled the surge.

3. Anthropic's Madcap March Wraps Up — 14+ launches, 5 outages, a court victory, an IPO filing in the works, a model leak, and an OSS program. The most consequential month in Anthropic's history ends today.

Read full report →

2. Mythos Cyber Fears Hit Washington — Axios reports Anthropic is privately warning top government officials that Mythos makes large-scale cyberattacks significantly more likely in 2026.

3. Pentagon Appeal Clock: 3 Days Left — DOJ has until ~April 2 to appeal Judge Lin's injunction. Legal experts widely expect a Ninth Circuit filing.

Read full report →

2. Claude Paid Subs More Than Double — 1M Signing Up Daily — Daily signups quadrupled from 250K to over 1M since January. Super Bowl ads and Claude Code are the primary growth drivers.

3. Pentagon Has One Week to Appeal Injunction — Judge Lin delayed implementation by seven days. If no appeal is filed by ~April 2, the supply-chain risk designation stays frozen.

Read full report →

2. Accenture Launches Cyber.AI Powered by Claude — The enterprise security platform cuts scan turnaround from days to under an hour. API test coverage jumped from 10% to 80% at Accenture's own infrastructure.

3. MCP Hits 97M Monthly Downloads — Moves to Linux Foundation — Anthropic donates MCP governance to the Agentic AI Foundation, co-founded with OpenAI, Block, Google, and Microsoft. New peak-hour usage limits also take effect today.

Read full report →

2. Anthropic Weighs October IPO at $60B+ Valuation — Bloomberg reports early talks with Goldman Sachs, JPMorgan, and Morgan Stanley. Racing OpenAI to go public first.

3. Judge Blocks Pentagon Ban — Calls It First Amendment Retaliation — Judge Rita Lin granted the injunction, suspending the supply-chain risk designation and calling it an "Orwellian notion."

Read full report →

2. Pentagon Ruling Imminent — Anthropic Requested Decision by Today — Judge Rita Lin called the ban "troubling" and said she'd rule within days. Meanwhile, Pentagon CIO confirmed Claude is being used in Operation Epic Fury.

3. Economic Index: Experienced Users 10% More Successful — Anthropic's March report shows 6+ month users do harder work and get better results. Each year of usage correlates with tackling more complex tasks.

Read full report →

2. Claude Code Auto Mode Ships — Safety Classifier Guards Every Action — A new permissions mode where Claude decides autonomously, with a per-action classifier checking for destructive actions and prompt injection before each tool call runs. Team plan now, Enterprise soon.

3. Claude Code Channels: Text Your AI Coder from Telegram or Discord — Push tasks into a running Claude Code session from your phone. Full filesystem and MCP access. Replies when done. No open ports required.

Read full report →

2. Pentagon Injunction Hearing Today at 1:30pm PST — Judge Rita Lin hears oral arguments in San Francisco. First judicial test of whether the Pentagon's supply-chain risk designation is lawful or unconstitutional retaliation.

3. Ramp AI Index: Nearly 1 in 4 Businesses Now Pay for Anthropic — Adoption hit 24.4% with 4.9% MoM growth. Anthropic wins 70% of head-to-head matchups against OpenAI among first-time buyers. OpenAI saw its biggest monthly decline.

Read full report →

2. DOJ Calls Pentagon Blacklist "Lawful and Reasonable" — Hearing Tomorrow — Opposition brief filed ahead of Tuesday's 1:30pm PST injunction hearing before Judge Rita Lin. Anthropic fired back in a reply brief calling the designation preemptive retaliation.

3. Claude.ai Login Outage Resolved; 2x Promo Ends March 27 — Sunday's auth issues are cleared. Four days left on the off-peak 2x usage boost for Free, Pro, Max, and Team plans.

Read full report →

2. Microsoft & 22 Retired Generals Back Anthropic; DoD Claims 6-Month Transition — Former service secretaries and flag officers call the Pentagon designation "retribution." Military users on the ground say replacing Claude is far harder than leadership admits.

3. Extended Thinking Display Field Ships; 2x Usage Promo Ends March 27 — Developers can now omit thinking blocks for faster streaming with thinking.display: "omitted". The off-peak 2x usage boost for Free, Pro, Max, and Team plans has five days left.

Read full report →

2. Claude Code Channels Launches as Research Preview — Connect Telegram or Discord to a running Claude Code session and control it from anywhere. Plugin runs locally, no open ports, no public webhook. Requires v2.1.80+ and a paid plan.

3. 1M Context Window Now GA at Standard Pricing — Opus 4.6 and Sonnet 4.6 get the full 1M token context window at standard rates -- no more 2x premium. Media cap raised to 600 images or PDF pages per request.

Read full report →

2. Claude Code Voice Mode & /loop Fully Rolling Out — Push-to-talk voice mode is expanding to all users, alongside the /loop command for recurring task automation and new /effort and /batch controls.

3. 2x Usage Promo Enters Final Week — Anthropic's bonus usage promotion doubles limits for Free, Pro, Max, and Team plans during off-peak hours through March 27.

Read full report →

2. "Claudy Day" Vulnerability Chain Disclosed — Three chained flaws in Claude.ai enable silent data exfiltration and malicious redirects. Primary injection flaw patched; two mitigations still in progress.

3. Claude Captures 40% of Enterprise LLM Spend — Business subscriptions up 4.9% MoM while OpenAI fell 1.5%. Claude now holds ~40% of enterprise AI spend, up from 24% a year ago.

Read full report →

2. Cowork Gets Persistent Agent Threads — Anthropic begins rolling out persistent agent thread support in Cowork for Max plans today, with Pro plan access following within 48 hours.

3. Claude Code Voice Mode Expands to 5% of Users — Push-to-talk voice coding is live for an initial slice of Pro, Max, Team, and Enterprise subscribers. Hold spacebar to speak, release to send — now in 20 languages.

Read full report →

2. Anthropic Acquires Vercept to Supercharge Computer Use — The 9-person AI perception startup joins Anthropic. Claude already hit 72.5% on OSWorld, up from under 15% when computer use launched in late 2024.

3. Third Claude Outage This Month Hits Free Users — Service disruption on March 17 appears limited to free-tier users on Claude.ai. Anthropic has identified the issue and is deploying a fix.

Read full report →

2. Palmer Luckey: Pentagon Should Have Been "More Forceful" — Anduril founder told Axios that private companies shouldn't dictate how the military uses their tech, backing the Pentagon's blacklist decision.

3. US Army Awards Anduril $20B Contract — A 10-year deal covering autonomous systems, counter-drone tech, and AI-powered command and control. Announced the same week Luckey backed the blacklist.

Read full report →