Anthropic and the Intelligence Utility Thesis
Anthropic just raised $30 billion at a $380 billion valuation and that got most of the headlines, though the more telling figure is the $14 billion current annualized revenue run-rate, achieved less than three years after the company booked its first dollar. Salesforce required 16 years to reach $14 billion in annual revenue. Anthropic has done it in under three, making it the fastest commercial scaling in enterprise software history.
What makes this trajectory structurally different from a typical venture blitz-scaling story is where the revenue is coming from. Eight of the Fortune 10 are Claude customers, over 500 enterprises are spending more than $1 million annually, and the critical growth engine, Claude Code, Anthropic’s agentic coding tool, hit a $2.5 billion run-rate having only launched publicly in May 2025 with enterprise customers making up more than half of that number. When you look at the shape of this revenue base, it looks far more like an enterprise infrastructure platform, possibly even an enterprise operating system, than an AI chatbot company, which brings us to the core of the thesis.
The Structural Advantage
Anthropic has done something we think is underappreciated: it has inverted the traditional cloud value chain. Rather than paying a premium to sit on top of AWS or Google Cloud, the company has positioned itself so that the hyperscalers compete for the privilege of hosting Claude. AWS, Google Cloud, and Azure all distribute Claude through their marketplaces, giving Anthropic leverage on both compute costs and distribution that no other AI lab enjoys, and that leverage is compounding as enterprise adoption crosses into self-reinforcing territory.
Anthropic's competitive position in the enterprise market has shifted from a product advantage to a structural one, though the survey data tells two different stories depending on what you measure. Workload-weighted enterprise surveys, the ones that track production foundation-model usage rather than brand recall, show Anthropic rising sharply from low-teens share in 2023 to a clear leadership position by late 2025, with estimates around 40% enterprise share versus roughly 27% for OpenAI. In CIO and top-of-mind surveys that ask which model respondents work with most today, OpenAI and ChatGPT remain the default answer at 57%, with Gemini at 28% and Claude at just 7%. We think the gap is a distribution and perception effect. Claude can be a minority named tool while still capturing a disproportionate share of high-value production workloads, the ones that actually drive spend. Where the share shift is clear is coding, the vertical that drives the most measurable ROI and creates the deepest workflow lock-in. Current estimates from Menlo Ventures put Claude at 54% share versus 21% for OpenAI, up from 42% just six months earlier. We view this as the clearest proof point that Claude Code and agentic long-context development workflows are functioning as the wedge that converts bottom-up developer adoption into durable enterprise platform share.
The Business Model Question
Understanding the thesis requires understanding what kind of company Anthropic is actually becoming, and this is where we spent the most time in our analysis. The business combines SaaS revenue from Claude Pro and Enterprise subscriptions, consumption-based API revenue that scales automatically with customer growth, and an emerging infrastructure layer that will soon include proprietary data center capacity. We describe this combination as an intelligence utility, and how you model it determines whether the $380 billion valuation looks demanding or cheap. At 27x current revenue the headline multiple is high, though it compresses to roughly 15x on a year-end 2026 estimate of $26 billion and to about 5.4x on a 2028 revenue estimate of $70 billion, which is consistent with premium enterprise platform valuations when you account for the growth trajectory.
The margin structure is where the real debate sits. We have heard and seen approximate consensus targets ~70s% gross margins by 2028 and our bottoms-up inference cost model, which triangulates public GPU/TPU pricing, silicon roadmaps, and enterprise contract structures, suggests something closer to 58%. The gap between those two numbers represents the difference between the bull case and the base case, and it is the single most important variable to get right in any model you build on this company.
What to Watch
We see three dynamics that will determine whether Anthropic’s trajectory holds. The first is the pace of silicon cost reduction, where Trainium3 and Ironwood chips (all future silicon clusters) could cut per-token compute costs by 50-65%. The second is the enterprise adoption curve for Claude Code, which is creating switching costs that look more like developer platform lock-in than SaaS churn. The third is the cadence of the $50 billion data center buildout that is shifting Anthropic from a model-layer company toward a vertically integrated hyperscaler. And there is a fourth dynamic we explore in the full report that connects directly to our enterprise software bifurcation analysis: the growing ecosystem of independent software vendors building their products on Claude’s API. As the traditional per-seat SaaS model comes under structural pressure from agentic AI, a thesis we laid out in detail in our recent “AI Will Not Kill SaaS” report, the economics increasingly favor the model layer that powers these interactions. Anthropic benefits from both sides of the bifurcation: system-of-record incumbents embedding Claude into their platforms drive token consumption, while commodity SaaS displacement by AI agents drives consumption through the same API. Anthropic sits at exactly the right layer of the stack.
2026 is likely the year the market gets to price Anthropic directly, whether through an IPO or continued private rounds at escalating valuations. When you put all these dynamics together you can see the shape of the opportunity and why we think it warrants the deep analysis we have done in the full report.
What Full Subscribers Get in the Complete Deep Dive.
Three-tiered sum-of-parts valuation framework decomposing Anthropic into SaaS, API/utility, and infrastructure tiers with tier-specific multiples and 2028 implied enterprise value of ~$1.34 trillion
Bottoms-up inference cost model with adjustable assumptions across token volume, silicon pricing, model mix, and utilization rates, built for your own scenario analysis
Complete product and pricing landscape with competitive API pricing tables across Anthropic, OpenAI, Google, Meta, and Mistral
Detailed capex and free cash flow analysis anchored to our “$2-for-$1” infrastructure framework, showing the capital intensity math through 2028
Gross margin bridge analysis breaking down the three discrete levers from today’s ~50% to the base case 58%, and what it would take to reach rumored bull-case 70% target
Competitive benchmarking exhibits across capability, cost efficiency, coding performance, and safety metrics with indexed scoring
Ecosystem analysis examining how the disruption of traditional SaaS business models creates a structural tailwind for Anthropic’s consumption-based API economics
Key risk matrix covering model commoditization, margin compression, capital intensity, competitive dynamics, and regulatory exposure