Carry phase

The straight-line method for capitalised AI

Carry

Straight-line amortisation charges the amortisable amount evenly across the useful life. IAS 38 makes it the method to use where the pattern of benefit cannot be determined reliably, which is often the case for a token-built asset IAS 38 §98. It is the defensible default: simple, hard to argue is aggressive, and easy to reconcile in the disclosure note.

Why the default fits most token assets

Many token-built assets deliver a steady benefit once live: the copilot answers questions at a broadly constant rate, the classifier runs on each transaction. Where the benefit is consumed evenly, straight-line reflects it, and where the pattern genuinely cannot be pinned down, IAS 38 directs you to straight-line anyway IAS 38 §98. That makes it both the common choice and the safe one.

Computing the periodic charge

The charge is the amortisable amount, cost less residual value, divided by the number of periods in the useful life. Because residual value for an AI intangible is usually zero IAS 38 §100, the charge is typically cost divided by life. Amortisation begins when the asset is available for use, not when it is first put into use IAS 38 §97.

Carrying-value ladderIllustrative example

Cost 360,000 over 36 months, residual zero: an equal charge each year. Illustrative example.

When to reject straight-line

Straight-line is the default, not a mandate. It should be rejected where there is reliable evidence that the benefits are consumed unevenly, for instance an asset built for a fixed campaign whose usage will fall away sharply, or one whose token consumption is forecast to be heavily front-loaded. In those cases a units-of-production profile is the better reflection of the pattern of benefit, and choosing straight-line for simplicity alone would misstate the charge.