Anthropic Reportedly in Talks to Raise at Least $30B at a $900B+ Valuation
Reporting compiled this week has Anthropic in early talks to raise at least $30 billion at a pre-money valuation exceeding $900 billion. If it closes anywhere near that mark, it would roughly double the company’s valuation from its prior round and put it on the doorstep of the trillion-dollar tier alongside the largest names in the AI race.
The figure isn’t just a vanity number — it’s the fuel behind everything else in this brief. Compute commitments, enterprise alliances, a $100M partner fund, and a relentless shipping cadence all run on capital, and a raise of this size signals investors are betting that Claude’s enterprise distribution lead is durable. Treat the specifics as reported-not-confirmed until terms are official, but the direction of travel is unmistakable.
Anthropic Commits $100M to the Claude Partner Network
Anthropic is investing $100 million into the Claude Partner Network — funding for the consultancies, system integrators, and software partners that build and deploy Claude for enterprise customers. The money goes toward training, certification, co-selling, and go-to-market support, the connective tissue that turns a model into deployed workflows inside large organizations.
It’s the logical companion to the KPMG and PwC alliances: Anthropic can’t hand-deliver Claude into every Fortune 500 itself, so it’s paying to build an army of partners who can. For buyers, a funded partner network means more certified implementers and fewer half-finished pilots. For Anthropic, it’s distribution leverage that compounds.
Anthropic and the Gates Foundation Form a $200M Partnership
Anthropic committed $200 million in grant funding, Claude usage credits, and technical support to the Gates Foundation over the next four years, aimed at programs in global health, life sciences, education, and economic mobility. It’s a rare mission-driven counterweight to the week’s enterprise land-grab.
Beyond the optics, the deal is a credibility play: deploying Claude against disease modeling, health-systems research, and learning tools puts the model in high-stakes, high-scrutiny domains where getting it right matters. Expect Anthropic to lean on this work in its safety and societal-benefit messaging as the valuation conversation heats up.
“Dreaming” for Managed Agents — Harvey Reports a 6x Jump in Task Completion
Dreaming, Anthropic’s scheduled memory process for Managed Agents, runs between agent sessions: it reviews everything an agent did on its last job, pulls patterns out of those sessions, and writes new memory entries the next session can use — loosely modeled on how the brain consolidates memory during sleep. It’s in research preview and requires the dreaming-2026-04-21 beta header.
The headline result is the one builders will quote: legal AI company Harvey reported task-completion rates climbing roughly 6x after turning on dreaming. The pitch is self-improving agents that get better at recurring work without a human hand-tuning prompts — the kind of compounding that separates a demo from a system you can leave running. Pair it with the new Outcomes evaluation and multi-agent orchestration, also in public beta.
Claude Code Keeps Shipping — Agent View, /goal, a Renamed /code-review, and No Git Bash on Windows
The post-event reliability train rolled on with Claude Code v2.1.141. claude agents opens a single screen showing every session — what’s running, what’s blocked on you, what’s done — with directory-scoped views via claude agents --cwd <path> and pinned background sessions that survive idle and auto-restart to apply updates. /goal keeps Claude working across turns until a completion condition holds, and fast mode now runs on Opus 4.7 by default.
Two changes stand out for daily users: /simplify was renamed /code-review and now reports correctness bugs at a chosen effort level (e.g. /code-review high), with --comment to post findings as inline GitHub PR comments. And on Windows, Git for Windows is no longer required — Claude Code falls back to PowerShell when Bash is absent, removing a long-standing setup headache.
Claude in Chrome Opens to All Paid Plans — Including Pro
The Claude in Chrome extension is now available in beta to every paid plan, Pro included. It reads the page DOM, inspects network requests, and can read, click, and navigate sites alongside you — which makes the build-in-the-terminal, test-in-the-browser loop with Claude Code genuinely practical. “Ask before acting” lets Claude draft a plan for your approval, then execute within those boundaries.
One catch worth knowing: model access scales with plan. Pro is limited to Haiku 4.5 in Chrome, while Max, Team, and Enterprise can reach for Opus 4.7 on the demanding workflows. Still, opening DOM and network access to Pro users is a real expansion of who can let Claude drive the browser — and a reminder to treat agentic browsing with the same caution you’d give any tool with credentials.
Claude for Legal: 20+ Connectors, 12 Practice-Area Plugins, and Thomson Reuters + Free Law Project on MCP
Anthropic went all-in on the legal vertical, shipping more than 20 MCP connectors — DocuSign, Ironclad, iManage, NetDocuments, LexisNexis, Thomson Reuters, Box, Everlaw, and more — alongside 12 practice-area plugins spanning corporate M&A diligence, litigation, employment, privacy, IP, and AI governance. In a notable signal of the protocol’s reach, both Thomson Reuters and the nonprofit Free Law Project launched MCP integrations connecting their legal data directly to Claude.
The proof is in the adoption: Freshfields, Quinn Emanuel, Holland & Knight, and Crosby Legal were named running Cowork on live matters, and Freshfields has deployed Claude across thousands of users while co-developing AI-native workflows with Anthropic. When a research giant and an open-data nonprofit both expose their corpora over the same protocol, MCP stops looking like a convention and starts looking like infrastructure.
Claude for Small Business Hits the Road — 15 Workflows and a 10-City US Tour
Claude for Small Business — the toggle-on package of 15 ready-to-run agentic workflows (payroll planning, month-end close, invoice chasing, lead triage, contract review, cash-flow monitoring, and more) plus 10+ connectors for QuickBooks, PayPal, HubSpot, Canva, Docusign, and Google Workspace — is being taken on tour. Anthropic is running free half-day workshops for 100 local entrepreneurs each across ten US cities, including Chicago, Dallas, and San Jose.
It’s the same embed-where-the-work-happens strategy as the enterprise deals, aimed downmarket: Claude lives inside the SaaS tools owners already pay for, does the work, and waits for approval before anything sends, posts, or pays. Paired with a free PayPal-backed AI Fluency course, it’s a deliberate move to make “agents for the rest of us” a real category rather than an enterprise-only luxury.
Sunday Read — A $900B Number Only Makes Sense If the Distribution Holds
Strip away the individual announcements and the last two weeks tell one story: Anthropic is converting a capability lead into distribution as fast as capital will allow. A reported $30B+ raise at a $900B+ valuation is the financing. The KPMG and PwC alliances, a $100M partner fund, Claude for Legal, and Claude for Small Business are where that money lands — not in flashier benchmarks, but in getting Claude wired into the systems where billable and operational work actually happens.
That’s the bet a $900B valuation has to justify. Models are increasingly substitutable; the moat is whether Claude is the thing already embedded in your firm’s tax engine, your matter-management system, or your QuickBooks. The risk is just as visible: valuations that size assume the distribution lead is durable, and yesterday’s quietly patched sandbox bypass is a reminder that trust is the variable that can unwind it fastest. The capital is betting Anthropic closes the trust gap as quickly as it’s closing the distribution one. The next year decides whether that bet pays.