AI
Integration is another. A system that sits outside the business process may impress in a demonstration and frustrate in daily use. Employees work around it, duplicate tasks, or correct it. The tool becomes another layer of friction.
What it eats
The physical side of AI is becoming harder to ignore. Data centres need electricity, cooling, backup systems, security, connectivity, and land. Efficiency is improving, but demand is expanding faster.
AI does not remove work from a business. It changes where the work sits. Ownership matters too. Businesses need to know who pays for the system, who governs it, what data it depends on, who checks the outputs, and who answers when the system is wrong. Without that clarity, Patel says, AI becomes experimentation without discipline. The organisation accepts risk it has not yet named.
Energy has already moved from the basement into the boardroom."
For South African companies, this should feel familiar. Energy has already moved from the basement to the boardroom. It is a risk issue, a budgeting issue, and a resilience issue. AI adds another layer.
Build more workflows, customer interactions, analytics processes, and decision systems on AI, and the infrastructure behind those systems becomes part of business risk. A failure in that chain can become a customer-service problem, an operations problem, a compliance problem or a reputational problem.
The language of AI encourages abstraction. It speaks of intelligence, automation, augmentation, transformation, and scale. But models run on machines. Machines need power. Power depends on grids, generation, cooling, and planning.
The machine eats electricity. It eats water. It eats capital. It eats reliable data, skilled people, and governance time.
Companies that ignore the appetite will eventually receive the bill. It will cover more than the software licence. �
78 sabusinessintegrator. co. za