AI Is Becoming a Physical Business Decision
You may think of artificial intelligence as software: a chatbot on your website, an automated report, or a tool that summarises documents. But behind every AI feature are physical systems—chips, servers, memory, networks, electricity, cooling equipment, and sometimes robots.
That matters because your business may soon depend on more than choosing the right application. The availability, speed, reliability, and location of the underlying infrastructure can affect how quickly you serve customers, process orders, inspect products, and make decisions.
A recent example is Andreessen Horowitz launching a US$1.1 billion “Machine Age” fund focused on AI hardware and infrastructure. Its stated scope includes computer chips, memory, data centres, robots, cooling, electrical systems, and other supporting facilities. Source: TechCrunch
TL;DR
AI growth is moving beyond apps into the physical infrastructure that makes automation work. For your SME, this means you should evaluate connectivity, data handling, device readiness, and operational processes—not simply subscribe to another AI tool.
Start with one practical workflow, measure the result, and choose systems that can continue working reliably as your business grows.
What This Means
The “Machine Age” idea describes a shift from AI being mainly a software capability to AI being supported by a much larger physical ecosystem. AI systems need computing power to process information, memory to store and retrieve it, fast connections to move data, and facilities that can supply electricity and cooling.
The fund announced by a16z is reported to focus on infrastructure ranging from chips and memory to data centres and robots. Source: TechCrunch This is a useful signal: investors expect AI demand to create opportunities not only for software companies, but also for businesses building the equipment and services underneath them.
For an SME, you do not need to build a data centre or buy specialised chips. The practical lesson is simpler: AI performance depends on the whole operating environment. If your internet connection is unstable, your information is scattered, or your staff record data inconsistently, even an impressive AI application may produce disappointing results.
“The best AI system cannot compensate for a process that is unclear, data that is incomplete, and responsibilities that are undefined.”
There are three broad ways AI infrastructure may affect you:
- Cloud dependence: Many AI services run on remote infrastructure. Your team needs dependable access and clear rules for sending business information to those services.
- On-device processing: Some tools may process information locally on phones, cameras, computers, or machines. This can support faster responses and reduce reliance on continuous connectivity.
- Physical automation: AI may increasingly guide cameras, scanners, warehouse equipment, inspection systems, and robots. This connects digital decisions directly to physical work.
How This Applies to Malaysian SMEs
Suppose you operate a distributor in Selangor, Penang, Johor, or Sabah. Your sales team may receive orders through WhatsApp, email, phone calls, and online marketplaces. The immediate AI opportunity is not necessarily a robot. It may be a system that reads incoming orders, checks product information, identifies missing details, and sends a structured request to your operations team. For that to work, your product catalogue, customer records, and stock information must use consistent names and formats.
If you run a restaurant, café, catering business, or food manufacturer, AI can support demand planning, purchasing, customer responses, and quality checks. A camera-based system might help identify packaging or production issues, while a forecasting tool could compare order history with seasonal patterns. However, the result depends on reliable records for ingredients, production times, delivery schedules, and rejected items. Your first step should be to standardise how staff record these details.
For a service business such as an accounting practice, clinic, repair company, training provider, or property agency, the physical infrastructure may be less visible but still important. Staff need stable access to customer files, document systems, video calls, and automation platforms. If files are stored across personal devices and different chat groups, an AI assistant will struggle to find the correct version. A central document structure and clear access permissions are more valuable than adding several disconnected tools.
Manufacturers and workshops may see the most direct connection between AI and physical operations. Sensors, cameras, barcode scanners, and industrial devices can collect information about production, maintenance, and inventory. AI can then identify unusual patterns or help prioritise inspections. You should not begin by automating every machine. Select one repeated problem—such as rework, delayed maintenance, or incorrect picking—and test whether better data can improve the decision.
Retailers also need to consider the balance between cloud systems and local devices. A point-of-sale terminal, stock scanner, or security camera may need to continue operating during a connectivity problem. When reviewing an AI-enabled system, ask what happens if the internet is interrupted, where information is stored, and how staff can continue working manually.
Infrastructure Questions to Ask Before Adopting AI
| Area | Question for your business | Practical action |
|---|---|---|
| Connectivity | Can staff access essential systems during busy periods? | Record interruptions and identify a backup process. |
| Data quality | Are customer, product, and transaction records consistent? | Create standard fields, naming rules, and ownership. |
| Security | Who can view or export sensitive information? | Use role-based access and remove inactive accounts. |
| Devices | Can existing computers, phones, scanners, or cameras support the workflow? | Test one department before wider deployment. |
| Continuity | What happens when the AI service or internet connection is unavailable? | Document a manual fallback procedure. |
Practical Takeaways
- Choose a workflow, not a buzzword. Focus on one task that is repeated, time-consuming, or prone to mistakes.
- Map the information flow. Write down where data begins, who changes it, and where the final decision is recorded.
- Clean your core records. Standardise customer names, item codes, addresses, service categories, and status labels.
- Check the physical requirements. Review internet reliability, computers, mobile devices, scanners, cameras, and power arrangements.
- Keep people responsible. AI may suggest an answer, but a named staff member should approve important customer, financial, safety, or operational decisions.
- Ask suppliers specific questions. Confirm data location, retention, access controls, downtime procedures, integration options, and export capability.
- Measure operational outcomes. Track response time, order accuracy, unresolved requests, processing delays, or inspection results before and after the pilot.
- Plan for change. Select systems that can connect with your existing accounting, customer, inventory, and communication tools.
The Bigger Picture
The investment in AI hardware suggests that the next stage of adoption will involve more than office software. The infrastructure behind AI is expanding into networks, devices, industrial equipment, logistics, energy systems, and physical workplaces. TechCrunch reports that a16z specifically described needs involving faster systems, memory, interconnects, edge devices, cooling, materials, electrical infrastructure, and real estate. Source: TechCrunch
For Malaysian SMEs, this may create new opportunities. Local businesses could provide equipment maintenance, data preparation, systems integration, warehouse services, industrial inspection, specialised training, or support for AI-enabled devices. You do not need to be an AI laboratory to participate. You may become the trusted operator that applies these tools to a specific industry problem.
It also means technology decisions should be made with a longer view. A tool that works only as a standalone application may become difficult to manage when you add more staff, branches, devices, or customers. Before adopting AI, consider whether the system can share information with the rest of your operations and whether your team can understand its recommendations.
The practical message is not to rush into expensive equipment or follow every announcement. Instead, build a stronger foundation: organised information, dependable processes, secure access, and clear accountability. When AI infrastructure becomes more available, your business will be in a better position to use it responsibly.
A Sensible Next Step
Pick one workflow this week and document it from beginning to end. For example, record how a customer enquiry becomes a quotation, how a purchase order becomes a delivery, or how a service request becomes a completed job. Mark every handoff, delay, duplicate entry, and missing piece of information.
Then ask whether automation can improve that single workflow without removing human oversight. If the answer is yes, run a controlled pilot, train the people involved, and review the results. The physical buildout of AI may happen at a global scale, but your best starting point is local: one process, one team, and one measurable business improvement.
Ready to Streamline Your Operations?
Your business should run itself. AutoRunBiz deploys AI agents to automate your daily operations — WhatsApp orders, invoicing, customer follow-ups, and accounting. Book a free 15-min ops audit to see where automation fits your business →