Anthropic

Anthropic

Overview

[[Dario Amodei]] and his sister Daniela founded Anthropic in 2021 with a singular mission: to develop safe, reliable, and aligned AI systems. The company emerged from a group of former [[OpenAI]] researchers who believed the AI safety problem deserved dedicated focus and resources at a frontier lab. Anthropic has distinguished itself through a safety-first approach, combining rigorous interpretability research with constitutional AI methodologies and responsible scaling practices. Their philosophy prioritizes understanding why AI systems behave the way they do before scaling them further.

Today, Anthropic represents one of the fastest-growing AI labs globally, having closed a $30 billion Series G funding round in February 2026 that valued the company at $380 billion—the second-largest venture capital deal on record. By April 2026 the lab's annualized revenue had reached $30 billion, surpassing OpenAI's reported ~$24–25 billion ARR for the first time and marking a roughly 1,400% year-over-year run-rate increase from $1B in January 2025. Anthropic has subsequently received multiple investor offers at an $800 billion valuation and is reportedly targeting an October 2026 U.S. IPO that would aim to raise more than $60 billion. Anthropic's trajectory in 2026 has been defined by an acceleration of model releases (Sonnet 4.6 in February, Opus 4.7 in April, Claude Mythos Preview launching in April under restricted access), the introduction of Project Glasswing for partner-led pre-release vulnerability hunting, the May 5 launch of ten preconfigured Claude agent templates for financial services, and a public dispute with the Pentagon that left Anthropic excluded from classified-network AI deployments after the lab declined to permit Pentagon use of Claude for "all lawful" purposes.

Key Details

  • Founded: 2021
  • Founders: [[Dario Amodei]], Daniela Amodei (President), and other former OpenAI researchers
  • CEO: [[Dario Amodei]]
  • Headquarters: San Francisco, CA
  • Funding / Valuation: $30B Series G (February 2026) at $380B post-money valuation; subsequent VC offers at $800B+ valuation as of late April / early May 2026
  • Annualized Revenue: $30B (April 2026) — passed OpenAI ARR for the first time (caveat: OpenAI disputes the figure on accounting grounds)
  • Reported IPO Target: October 2026 U.S. listing aiming to raise >$60B
  • Website: https://www.anthropic.com

Current Models

  • [[Claude Mythos Preview]] — Frontier-tier model in restricted partner preview via Project Glasswing; reportedly ~10T parameters trained on NVIDIA Blackwell hardware; leads Opus 4.7 across all benchmarks; specialized cybersecurity capabilities (April 2026)
  • [[Claude Opus 4.7]] — Generally available flagship model with substantially improved coding and vision; same $5/$25 per-million-token pricing as Opus 4.6 (April 16, 2026)
  • [[Claude Opus 4.6]] — Previous flagship; introduced 1M context and Agent Teams (February 5, 2026)
  • [[Claude Sonnet 4.6]] — High-performance mid-tier model balancing speed and capability
  • [[Claude Opus 4.5]] — Previous-generation flagship model
  • [[Claude Sonnet 4.5]] — Previous mid-tier model
  • [[Claude Haiku 4.5]] — Efficient small model for fast inference
  • [[Claude 3.5 Sonnet]] — Earlier model variant in the Claude 3 family

Key People

  • [[Dario Amodei]] — CEO and co-founder, leads strategic vision and safety research
  • Daniela Amodei — President and co-founder, oversees operations and scaling
  • Mike Krieger — Former Chief Product Officer at Instagram, leads Labs division

Recent Developments

  • Google $40B Investment Commitment (April 24, 2026): Google committed to invest up to $40B in Anthropic — $10B immediately in cash at the same $350B valuation set in February 2026, with another $30B contingent on Anthropic hitting performance targets. As part of the deal Google Cloud will deliver 5 gigawatts of computing power to Anthropic over five years, with optionality on additional gigawatts. The deal stacks alongside Anthropic's existing AWS partnership and the SpaceX/xAI Colossus 1 capacity, giving Anthropic three of the largest compute suppliers in the industry. Pre-deal, Google reportedly held a ~14% stake in Anthropic.
  • SpaceX / xAI Colossus 1 Compute Partnership (May 6, 2026): Anthropic signed a deal with SpaceX (the merged SpaceX–xAI entity) to rent the entire compute capacity of Colossus 1 in Memphis, Tennessee — 300+ MW and 220,000+ NVIDIA GPUs (H100, H200, GB200). The deal directly funds doubled Claude Code rate limits for Pro/Max/Team plans, removal of peak-hour usage caps for Pro and Max accounts, and substantially higher Opus API limits. Anthropic also publicly expressed interest in partnering with SpaceX to develop multiple gigawatts of orbital AI compute capacity. The deal is striking on two dimensions: (1) Anthropic's primary frontier-LLM rival (xAI / Musk) is now its largest single new compute supplier, and (2) it lands roughly two months ahead of Anthropic's reported October 2026 IPO target.
  • 80x YoY Growth Disclosure (Q1 2026): CEO Dario Amodei publicly stated Anthropic delivered 80x year-over-year revenue and usage growth in Q1 2026.
  • ServiceNow Partnership (May 2026): Anthropic and ServiceNow announced a collaboration to deliver autonomous AI agents acting across enterprise workflows with built-in governance, auditability, and secure execution.
  • Financial Services Agent Templates Launch (May 5, 2026): Anthropic released ten preconfigured Claude agent templates for financial services — covering pitchbook construction, KYC screening, ledger reconciliation, month-end close, and other time-consuming workflows in banking, asset management, and insurance. Each template ships as a plugin for Claude Cowork and Claude Code, plus as a cookbook for Claude Managed Agents. Anthropic also expanded its Microsoft 365 integration and introduced new connectors including a Moody's MCP app for financial professionals — an aggressive push into Wall Street that frames Claude Opus 4.7 as the most capable model for financial work yet.
  • $800B Valuation Offers (Mid-April 2026): Multiple VC investor offers valuing Anthropic at $800B or higher; the company has reportedly resisted accepting these terms while preparing for an October 2026 IPO. The $800B+ figure is more than double the $380B Feb 2026 Series G valuation.
  • $30B ARR Passes OpenAI (April 2026): Anthropic crossed $30B in annualized revenue, passing OpenAI for the first time. Trajectory: $1B (Jan 2025) → $9B (end-2025) → $14B (Feb 2026) → $19–20B (March) → $30B (early April). Approximately 80% of revenue is enterprise API / developer; OpenAI disputes Anthropic's $30B headline figure on accounting grounds (gross-vs-net ~$8B differential).
  • Pentagon Exclusion (May 2026): Pentagon awarded classified-network AI deployment contracts to OpenAI, Google, Microsoft, AWS, NVIDIA, SpaceX, Reflection AI, and (subsequently) Oracle — explicitly excluding Anthropic. Anthropic declined to permit Pentagon use of Claude for "all lawful" purposes, citing concerns over domestic mass surveillance and fully autonomous weapons. The exclusion is a significant strategic break from peer labs.
  • Project Glasswing Launch (April 2026): Anthropic launched Glasswing, a controlled program giving select partners — reportedly including AWS, Apple, Cisco, Google, JPMorgan Chase, and Microsoft — early access to Claude Mythos Preview to find and patch critical vulnerabilities ahead of public deployment. Anthropic committed up to $100M in usage credits and $4M in donations to open-source security organizations.
  • Claude Mythos Preview (April 7, 2026): Announced a new frontier-tier model — distinct from the Opus line — with state-of-the-art cybersecurity capabilities. Reportedly identified thousands of zero-day vulnerabilities in major operating systems, browsers, and software (including a 27-year-old OpenBSD bug, a 17-year-old FreeBSD NFS RCE, and a guest-to-host VMM memory corruption primitive). Pricing for partners is $25/$125 per million input/output tokens — 5x Opus 4.7. Available through restricted Project Glasswing access and Vertex AI preview.
  • Claude Opus 4.7 GA (April 16, 2026): Generally available across Claude products, the Anthropic API, Amazon Bedrock, Google Cloud Vertex AI, and Microsoft Foundry. Substantial gains on hardest software-engineering tasks; higher-resolution vision; pricing held at $5/$25 per million tokens (same as Opus 4.6). Released alongside cybersecurity safeguards and the public-beta launch of Memory for Managed Agents.
  • White House Pre-Release Review Pressure (May 2026): Reporting indicates the Trump administration is weighing an executive order to establish a federal-industry working group reviewing frontier models pre-release — a shift partially motivated by concerns about Mythos's offensive cybersecurity capabilities.
  • Claude Sonnet 4.6 Launch (February 17, 2026): Mid-tier model released two weeks after Opus 4.6.
  • Claude Opus 4.6 Launch (February 5, 2026): Introduced 1M context window and native Agent Teams multi-agent collaboration.
  • Claude Code Growth: AI coding assistant on a $2.5B annualized revenue trajectory; becoming a major product pillar.
  • $30B Series G Funding (February 12, 2026): Closed at $380B valuation.
  • 4x Business Subscription Growth: Explosive enterprise growth since the start of 2026.
  • Labs Division Establishment: Mike Krieger leads a dedicated Labs division.
  • Constitutional AI Advancement: Continued development and refinement of constitutional AI methodology.

Why They Matter

Anthropic occupies a critical position in the AI landscape as the primary counterbalance to [[OpenAI]]'s market dominance — and as of April 2026, the new revenue leader among frontier AI labs. While OpenAI pursued rapid capability scaling with GPT-5.5, Anthropic has built institutional culture and technical depth around interpretability and safety. Their ability to raise $30 billion at a $380 billion valuation, then attract subsequent VC offers at $800B+, demonstrates investor confidence in their approach and validates the market demand for safer AI. With $30 billion in annualized revenue (passing OpenAI for the first time), strong enterprise adoption, and the May 5 financial services agent templates anchoring a Wall Street push, Anthropic has proven that the "safety-first" positioning is compatible with — and now leading on — commercial success. Their labs division, Claude Code platform, and the new Mythos research preview position them across consumer, enterprise, and developer segments while continuing to set the bar on frontier safety evaluations. The Pentagon exclusion in May 2026 marks a notable strategic divergence: Anthropic has chosen to forgo classified-deployment revenue rather than relax restrictions on autonomous-weapons and mass-surveillance use cases — a decision likely to define how the safety-first brand is perceived by both governments and enterprise buyers over the next 12–24 months.

Last Updated

May 9, 2026

→ Back to Companies