AI is not automatically sustainable — and for finance teams, that means it is not automatically justifiable.
The real question is not whether capital is allocated to AI.
The real question is whether the right model is deployed against the right operational return.
Running a premium LLM for every routine query is like capitalising a jumbo jet for grocery runs. It may work, but it destroys unit economics.
Historically, finance teams challenged factories for idle capacity and energy waste. Today, organisations are capitalising massive GPU clusters and calling it innovation — even when the workload adds little revenue or margin.
That is not accretive investment.
That is impaired capital efficiency.
In AINNA's work with Malaysian SMEs, every RM of technology spend must earn its place. If a company is serious about ESG and audited sustainability disclosures, computational efficiency must appear in the cost-to-benefit analysis.
- Smart routing reduces cost per inference.
- Small models reduce OpEx and hardware depreciation.
- Detached systems cut redundant licensing and cloud spend.
- Agentic decision-making improves return on AI assets.
Not every business process warrants the highest-cost model.
Sustainable value will not accrue to the companies that deploy the most AI.
It will accrue to companies that allocate AI capital intelligently.
AI can strengthen sustainability reporting and reduce operational waste.
But only if finance and operations teams govern its cost and carbon exposure with discipline.
