Standards-anchored to primary IFRS and FASB text. Worked figures are illustrative. Not accounting advice.
TCTokenCapitalisation
Scope and standing

About this reference

Scope
TokenCapitalisation.com is a standards-anchored reference for the balance sheet life of capitalised AI and token spend. It is written for a controller mid-close, not a learner, and it takes the capitalise decision as already made. It is maintained by Oliver Wakefield-Smith at Digital Signet. It is a reference, not accounting advice, and it exists to be quoted with the governing paragraph attached.

Who this is for

The reader we write for is a group financial controller nine months past the decision to capitalise internal AI development spend, now mid-way through year-end close and board-pack preparation. She is not asking whether to capitalise. She is asking what useful life she can defend, which amortisation method fits, what triggers an impairment review, and how to build a carrying-value roll-forward and a ROI view the audit committee will not tear apart. The FP&A lead and the technical-accounting owner read here too. Every page is written to be lifted into a memo or a board pack with its citation intact.

Why standards-anchored, not vendor-driven

Most writing on AI cost sits on one side of a chasm. The accounting standards explain how an intangible is recognised and amortised but never treat the token as the unit of measurement. The FinOps and CFO sources meter token spend but stop at the profit and loss account and never reach the balance sheet. This reference deliberately spans both, and it anchors every position to primary IFRS and FASB text rather than to a vendor narrative. No product is being sold here, so no treatment is being talked up to suit one.

What is in and out of scope

Not accounting advice

This is a reference. It sets out defensible positions and the paragraphs they rest on, and it flags where judgement is required. It does not replace your auditor or your own technical-accounting judgement on your specific facts. Every worked figure on the site is an illustrative example, never real company data.