Frontier AI Valuation Lab
View code (opens in a new tab)
F / 02 A closer look at the assumptions

Explain the gap.

Two views of the same business.
Trace what moves the value, then explore
what a target would require.

Research snapshot · 18 Oct 2026
Operating enterprise valueUSD billions
01 / Reference casePreset
Anthropic
—
Modeled enterprise valueOpen in model
The valuation gap
—
Alternative − reference
02 / Alternative casePreset
Anthropic
—
Modeled enterprise valueOpen in model
Inspect & edit the two cases26 inputs · 13 per case

Make each belief explicit.

All future inputs are analyst assumptions. Rates below are percentages: enter 12.5 for 12.5%. Changing an input recalculates both the attribution and the alternative case’s landscape.

All thirteen assumptions for both cases, with input units, definitions and limits.
Model assumptionReferenceAlternative

Model constraints apply to the complete case. Perpetual growth must remain below WACC and terminal ROIC, and no higher than 2034–2036 growth. Invalid drafts pause the analysis.

01 The attribution

What moved the value?

Every changed assumption is included.
Shared effects are allocated across six groups.

Reference → Alternative
Adds valueReduces value

Calculating the two cases…

Preparing a complete allocation of the valuation change.

Exact grouped Shapley
Inspect the allocation & changed inputs

The allocation averages each group’s contribution across all 720 possible orders of change. It depends on these six groups and this pair of cases; it does not establish causality or scenario probabilities. Long-run growth, WACC and terminal ROIC travel together to preserve their linked model constraints.

02 The value landscape

What would it take?

Explore the trade-off between the cost of capital and operating margin.

Illustrative target · USD billions
Other eleven assumptions: alternative caseTarget margin frontier

Building the value landscape…

Lower EVHigher EV

Each cell is a complete recalculated valuation. The line shows the required margin for the target EV within this window.

Adjust the visible window
Cost of capital · %
2036 EBIT margin · %

Changing an axis samples 13 evenly spaced points between its bounds. Shared windows retain their exact original points until edited. Invalid model combinations are marked unavailable.

Read the landscape data & target frontier

What this varies. The 2036 EBIT margin also changes the interpolated margins in 2031–2036 and the terminal period. WACC discounts every cash flow, including the remaining-2026 stub.

What to read from it. This is a conditional value surface, with the alternative case’s other inputs fixed. The reference EV is not plotted as a point. A target has no probability or financing-price interpretation.

03 Make the analysis portable

An argument you can inspect.

Verify these exact calculations on the server, share both cases, save your comparison to the Research Casebook, or take it into a printable brief.

Calculated in your browser · Server check available

Open Research Casebook
Context, not certainty

Keep the evidence
next to the assumptions.

Company disclosures and reporting establish context. They do not certify the margins, growth or discount rates in either case.

Open the full source register

Loading the dated source context…

Read the model’s limits
  • Forecasts are illustrative analyst assumptions, not probabilities or verified company guidance.
  • Enterprise value measures operating assets. Current net cash, security preferences and a fully diluted ownership bridge remain unverified.
  • Taxes apply only to positive EBIT; no tax-loss carryforwards are modeled. Net reinvestment already includes net capital expenditure and working capital.
  • The terminal period begins after 2036. Changes in its reinvestment policy can create a discontinuity from the final explicit cash flow.
  • Negative EV is preserved. No stock-price floor, investor return, or upside to an equity financing reference is inferred.