Is the DCF Dead? Rethinking Valuation for the Frontier AI Era

Three days ago, Anthropic announced a $30 billion Series G funding round led by GIC and Coatue, valuing the San Francisco-based AI safety company at $380 billion post-money. That figure alone demands attention. But when placed alongside the company's financials — a run-rate revenue of $14 billion, growing over 10x annually for each of the past three years, and yet no disclosed profitability — it raises a question that finance professionals can no longer defer: does the Discounted Cash Flow (DCF) framework, the bedrock of modern corporate valuation, still hold at the frontier of artificial intelligence?

The honest answer is: not in its standard form — and Anthropic's raise makes the case compellingly.

The Case Against the DCF

The DCF model is an elegant machine. Feed it projected free cash flows, a discount rate, and a terminal growth assumption, and it returns intrinsic value with mathematical authority. For a utility, an insurance company, or a consumer staples group, this works well. Cash flows are relatively stable, margins are predictable, and competitive dynamics shift slowly enough to be modelled.

Frontier AI firms violate every one of these premises.

Anthropic earned its first dollar in revenue less than three years ago. Today, its annualised revenue stands at $14 billion. That trajectory is not a trend line — it is a step function. The number of customers spending over $100,000 annually on Claude has grown sevenfold in a single year. Claude Code alone has crossed $2.5 billion in run-rate revenue and has more than doubled in revenue since the turn of 2026. The firm now counts eight of the Fortune 10 among its clients. No conventional DCF model, calibrated to historical base rates, would have forecast any of this.

The terminal value problem is equally severe. In a standard DCF, the terminal value — typically calculated as a perpetuity on normalised free cash flow — often accounts for 60% to 80% of the total implied value. For frontier AI companies, this figure is epistemically meaningless. The technology could render today's model architectures obsolete within 18 months, or it could establish a durable platform monopoly. These are not refinements of the same scenario; they are structurally different worlds. A DCF cannot price optionality of that magnitude. It will either catastrophically undervalue the upside or fail to account for existential competitive risk — and usually both at once.

The discount rate compounds the problem. The weighted average cost of capital for a firm burning capital at this scale, with no public market pricing and a capital structure that is simultaneously equity-like and mission-driven, is genuinely unknowable. Anthropic's Public Benefit Corporation status, its unusual investor base — spanning sovereign wealth funds, strategic technology firms, and long-duration institutional capital — and its explicit safety mandate create obligations that do not map onto a conventional WACC calculation.

The Case For Retaining DCF Logic

That said, abandoning the DCF entirely would be intellectually premature.

The framework's underlying logic — that the value of any asset is the present value of the economic benefits it generates over its life — remains correct. What changes is not the principle but the inputs and the planning horizon. For Anthropic, the near-term DCF is largely noise; the long-term DCF is largely speculation. But the discipline of projecting revenue scenarios, mapping margin trajectories, and stress-testing capital consumption remains valuable as a constraint on magical thinking.

The $380 billion valuation implies a revenue multiple of roughly 27x on the current run-rate. Even granting continued hyper-growth, the terminal assumptions embedded in that figure require Anthropic to sustain margins and market position at a scale that no software business has achieved at comparable speed. A DCF analysis, even a partial one, helps a prudent analyst understand the gap between what the market is pricing and what the fundamentals can plausibly support — a gap that is useful information regardless of whether one intends to invest.

A Defensible Valuation Framework for Frontier AI

Given the above, a credible valuation approach for a firm like Anthropic should be multi-modal and explicitly scenario-weighted. The following framework is proposed:

1. Scenario-Weighted Revenue Multiple (Near-Term Anchor) Use current run-rate revenue as the base. Apply a forward revenue multiple calibrated to hyper-growth SaaS peers with comparable retention and expansion dynamics (e.g., 15x–30x NTM revenue depending on growth deceleration assumptions). Weight three scenarios: accelerated adoption (probability: 40%), steady compounding (45%), and competitive disruption (15%). The output is a probability-weighted Enterprise Value that reflects market-observed pricing without requiring fictitious terminal cash flow assumptions.

2. Real Options Valuation (Strategic Optionality Layer) Identify and value discrete option-like payoffs: the potential to monetise frontier models across new verticals (healthcare, legal, financial services — all of which Anthropic is actively entering); the option value of compute infrastructure buildout; and the call option embedded in a prospective public listing. Use Black-Scholes or a binomial lattice with volatility parameters derived from comparable public AI infrastructure equities. This layer should be additive, not a replacement for the revenue multiple anchor.

3. Qualitative Risk Discount Apply a structured deduction for non-financial risks that markets price imprecisely: regulatory exposure (EU AI Act compliance, potential US oversight escalation), model architecture obsolescence, talent concentration risk (Anthropic's value is, to an unusual degree, human capital), and the governance complexity of the PBC structure. A 10%–20% haircut on the blended intrinsic value is defensible and analytically transparent.

4. Cross-Check Against Transaction Comparables Triangulate against contemporaneous funding rounds in frontier AI — noting that Anthropic's own Series F was completed at a $183 billion post-money valuation, meaning the market has assigned approximately a 2.1x step-up in under twelve months. This is not a valuation input per se, but it disciplines the output against what informed, large-scale institutional capital is actually paying.

5. Periodic Re-anchoring Unlike a traditional DCF, this framework should be explicitly time-stamped and re-run at minimum every six months, given the speed at which the revenue base, competitive landscape, and regulatory environment are moving. A valuation of Anthropic dated 15 February 2026 should be treated as significantly perishable.

The DCF is not dead. But for frontier AI firms, it is insufficient as a standalone tool. The appropriate response is not to abandon financial rigour, but to build a more honest framework — one that names its assumptions, quantifies its uncertainty, and acknowledges that at the edge of technological history, the most important variable is one that no model can calculate: what the next model can do.

Sources: Anthropic Series G Announcement, 12 February 2026; Yahoo Finance / GuruFocus, 13 February 2026.