How to fund AI: central budget, business units or both

When it comes up

During annual planning, when AI requests arrive from many teams at once, when shared platforms or data work need money, or when AI spend shows up unplanned in several budgets.

Your options

Central AI budget

Fits when: Early stage; shared platforms, data and governance need building; you want one portfolio view and economies of scale.

Watch out: Business units may not own outcomes; central teams can become a bottleneck; demand may outstrip the pot.

Business-unit budgets

Fits when: AI is mature in the business; use cases are specific to each unit; you want owners to trade AI against other spend.

Watch out: Duplicate tools and contracts, inconsistent controls, no shared platform, hidden spend.

Hybrid: central platforms, BU use cases (with showback or chargeback)

Fits when: You want shared infrastructure and standards plus business ownership of value.

Watch out: Allocation rules need care; chargeback can discourage experimentation if priced too high.

How to decide

Separate the layers: platforms, data and governance (shared) versus use cases (business-owned).
Fund shared layers centrally, with a multi-year view.
Fund use cases where the value lands, through stage gates (discovery, pilot, scale).
Show costs back to teams (showback) before moving to chargeback.
Keep one portfolio view across all budgets, with the spend tracker and value tracking.
Revisit the model yearly as AI maturity grows.

Trade-offs

Central funding speeds shared foundations and control but weakens business ownership; business-unit funding strengthens ownership but fragments platforms and contracts; hybrids balance both at the cost of allocation complexity.

Real examples

Gartner's 2026 budget research (published February 10, 2026) found 47% of organizations allocate just 1–5% of finance technology spending to AI, even as nearly 60% of CFOs plan to raise finance AI investment by 10% or more. Source: Gartner — CFOs' budget plans prioritize growth functions, technology and AI in 2026 (Feb 10, 2026) ↗
JPMorgan Chase provided its in-house generative AI assistant, LLM Suite, as a shared platform to 140,000 employees in 2024, while individual AI use cases across the bank were valued at close to $2 billion: a central platform with value landing in the businesses. Source: Banking Dive — JPMorgan Chase rolls out LLM Suite generative AI tool to 140,000 employees (Sep 12, 2024) ↗

What to check first

How much AI spend already sits in business-unit budgets, cards and expenses.
Which platforms and data foundations several use cases need.
Whether business owners are ready to own AI value targets.

Learn and use

Sources

Last reviewed 2026-10-01.