AINNA NeuralOps Through a Kewangan Lens: What the SIRIM Melaka and UPEN Melaka Visit Means for Kos, Risk and PKS Nilai✎ Edit

Alhamdulillah.

Yesterday, AINNA hosted representatives from UPEN Melaka (Melaka Keadaan Economic Perancangan Unit) and SIRIM Melaka as part of the Melaka Keadaan Usahawan Award evaluation. We were especially privileged to have the Director of SIRIM Melaka, Mr. Kamarulzaman bin Ahamad Zainudin, join the session.

From a finance and accounting standpoint, the visit carried more than ceremonial weight. It offered an external validation opportunity for AINNA NeuralOps-our next-generation AI architecture-and a chance to quantify how efficiency, reliability and governance translate into measurable value untuk PKS Malaysia.

Rather than a one-way presentation, the session became a structured two-way technical discussion, enriched by the SIRIM Director’s two decades of leadership experience. The quality of feedback directly affects our risk profile, compliance roadmap and capital-planning assumptions.

One recommendation with clear financial and operational implications was the integration of an Atomic Clock as the trusted time source across every NeuralOps component. A single high-precision time reference strengthens event consistency, auditability and system integrity. For finance and governance, that reduces reconciliation risk and supports the verifiable records PKS need for audit, tax and regulatory reporting.

We also demonstrated that NeuralOps runs optimised workloads on only around 10% of the GPU compute typically required by conventional AI systems. By using Sistem Berasingan, Smart Routing, parsers and guardrails, GPU resources are allocated only where they materially add value, while deterministic processes execute outside the LLM.

The financial and environmental consequence is significant: for applicable workloads, GPU-compute energy consumption can fall by up to 90%. Using a conservative estimate of 1,000 active users averaging 50 AI requests per day, NeuralOps could save approximately 459 kWh of electricity per month, or about RM138 in energy costs at a typical commercial tariff. That equates to roughly 340 kg of CO₂e emissions per month-more than 4 tonnes of CO₂e annually. Actual savings will vary by workload, model, infrastructure and energy source, but the direction is clear: lower OPEX, lower carbon liability and improved asset utilisation.

For us in finance and accounting, the strategic takeaway is straightforward. The future of AI investment is not about maximising GPU spend. It is about smarter architectures that deliver equivalent or better output with lower direct cost, lower energy OPEX, slower hardware depreciation and a lebih kecil environmental footprint.

Please keep AINNA in your prayers as we move forward. Being selected would materially support the three major pitching sessions ahead and strengthen our ability to bring NeuralOps to more Malaysian PKS at a viable cost structure.

Our appreciation to UPEN Melaka and SIRIM Melaka for the visit, the constructive dialogue and the practical insights. We look forward to embedding these recommendations into the next evolution of AINNA NeuralOps.