Anthropic

Safety-First AI

Founded: 2021 | HQ: San Francisco, CA | Key People: Dario Amodei (CEO), Daniela Amodei (President), Andrej Karpathy (Pretraining Lead) | Valuation: $965 billion (Series H, May 28, 2026) | Flagship Models: Claude Opus 4.7, Claude Sonnet 4, Claude Haiku 3.5, Claude Code

The Origin Story

Anthropic was founded in early 2021 by Dario and Daniela Amodei alongside seven other former OpenAI employees who left over growing concerns about AI safety practices at their former employer. Dario Amodei, who holds a PhD in computational neuroscience from Princeton and had served as Vice President of Research at OpenAI, had previously lobbied to keep cofounder Greg Brockman off the GPT-3 development team—foreshadowing the governance tensions that would later erupt into public view at OpenAI. The Amodei siblings built Anthropic around a singular thesis: that AI safety required dedicated research at the frontier of capabilities, not as an afterthought. The company pioneered Constitutional AI, a technique that trains models to follow a set of principles rather than relying solely on human feedback, giving Claude its distinctive cautious and reflective character. This safety-first positioning was not merely branding—it was embedded in the company's technical architecture from day one.

Key Milestones

Anthropic launched Claude in March 2023, initially available only through a waitlist. The model quickly gained a reputation for nuanced, careful responses that avoided the confident-but-wrong tendencies of competitors. In September 2023, Amazon invested $1.25 billion, expanding to $4 billion by year-end and eventually $8 billion total—making Amazon Anthropic's largest cloud and infrastructure partner. Google also invested approximately $2 billion. The Claude 2 launch in July 2023 expanded context windows to 100,000 tokens, then industry-leading. The Claude 3 family, launched in March 2024, introduced a three-tier model architecture—Haiku (fast and cheap), Sonnet (balanced), and Opus (powerful)—that became an industry standard. Claude 3.5 Sonnet, released incrementally through 2024, emerged as the model of choice for software developers, frequently outperforming GPT-4 on coding benchmarks. The launch of Claude Code in 2025 transformed Anthropic's commercial trajectory: the AI coding agent hit $1 billion in revenue within months and now generates over $2.5 billion annually, accounting for more than half of Anthropic's enterprise revenue. Claude Code rate limits were doubled in May 2026, reflecting continued infrastructure scaling. Anthropic's fundraising has been extraordinary. The company raised a $3.5 billion round in March 2025 at a $61.5 billion valuation, a $13 billion round in September 2025 at $183 billion, and a $30 billion Series G in February 2026 at $380 billion. In May 2026, Google invested up to $40 billion ($10 billion upfront, $30 billion contingent on milestones) and Amazon committed up to $25 billion in additional infrastructure funding—making these the largest single-investor commitments in AI startup history. On May 28, 2026, Anthropic closed a $65 billion Series H round at a $965 billion post-money valuation, surpassing OpenAI's last valuation of $730 billion and making it the most highly valued AI startup in the world. The funds will go toward advancing safety research, expanding compute, and scaling products. Revenue run rate reached $14 billion by early 2026, with over 500 enterprise customers paying $1 million or more annually. The company booked $21 billion in orders with Broadcom for custom TPU chips and expects to spend approximately $12 billion on model training in 2026. Andrej Karpathy, one of the most prominent figures in AI research and a founding member of OpenAI, joined Anthropic's pretraining team on May 19, 2026—a significant talent acquisition that reinforces Anthropic's frontier research capabilities. Claude Opus 4.7, released on April 16, 2026, became Anthropic's most powerful publicly available model, leading benchmarks across the hardest reasoning and coding tasks and beating competitors on reasoning evaluations. The model was specifically optimized for financial research and complex analytical workloads, addressing enterprise demand for AI-assisted analysis in regulated industries. Anthropic is also leveraging SpaceX's Colossus 1 data center (300 MW, 220K+ GPUs) for approximately $1.25 billion per month in compute, signaling a new level of cross-company infrastructure sharing in the AI industry.

Current Position

Anthropic has established itself as the primary enterprise alternative to OpenAI, winning over organizations that prioritize reliability, safety, and predictable model behavior. Its multi-cloud strategy—running on Amazon Web Services, Google Cloud, and now SpaceX's Colossus 1 infrastructure—provides enterprise customers with unmatched infrastructure flexibility. The product line has expanded beyond conversational AI into Claude Cowork for collaborative work, Claude Code Enterprise for large-scale software development, and specialized solutions for healthcare (HIPAA-compliant), government, and cybersecurity. The addition of Andrej Karpathy to lead pretraining research strengthens Anthropic's position in the frontier model race. Anthropic's technical differentiator remains its focus on honest, harmless, and helpful outputs—a positioning that resonates strongly with regulated industries.

What Leaders Should Know

Anthropic is the enterprise-safe choice in the AI race. For organizations in regulated industries—financial services, healthcare, government—Claude's safety-first architecture and Constitutional AI framework offer a lower-risk on-ramp to generative AI than competitors. The Claude Code product line has become the fastest-growing developer tool in enterprise AI; if your engineering teams are evaluating AI coding assistants, Claude Code should be on the shortlist. The $965 billion post-money valuation from the Series H round surpasses OpenAI's latest valuation, signaling extraordinary market confidence. Google's $40 billion commitment and Amazon's $25 billion commitment mean both AWS and GCP customers will see first-class Claude integration for years to come. The Colossus 1 compute partnership at $1.25 billion per month underscores the capital intensity of frontier AI—leaders should model for continuing price evolution as Anthropic invests in custom silicon to reduce long-term inference costs. Karpathy's arrival and Opus 4.7's benchmark leadership suggest Anthropic is not just the safe choice but increasingly the performance-leading choice.