AI Is Becoming an Operations Question, Not Just a Technology Question
If you run a Malaysian SME, you may see headlines about Anthropic signing a major computing agreement and wonder what it has to do with your shop, agency, factory, clinic, or service business. You are not building a frontier AI model. You are trying to answer customers quickly, keep sales records accurate, manage staff, and avoid repetitive administrative work.
That is precisely why this development matters. Large AI companies are investing heavily in the computing capacity needed to run increasingly capable systems. Anthropic’s reported agreement with Nscale is expected to provide computing capacity through Nvidia’s Vera Rubin chip system from late 2027, under a six-year arrangement, according to TechCrunch. You may not need that scale, but the tools built on top of it could affect how you work.
TL;DR: The AI industry is preparing for systems that handle more complex tasks, not only simple chat. For your SME, the practical response is to identify repeatable workflows, organise your business data, and introduce AI with clear human checks.
The best preparation is not buying every new AI tool. It is deciding where your business loses time today and building a reliable process around those pain points.
What This Means
AI systems need computing power to train and operate. Training teaches a model how to recognise patterns across large amounts of information. Operating the model, sometimes called inference, is what happens when you ask it to draft a reply, summarise a document, classify a lead, or extract information from an invoice.
The reported Anthropic-Nscale arrangement involves approximately $45 billion in rented AI computing capacity over six years, according to TechCrunch. The source also reported that Nscale will use Nvidia’s Vera Rubin system, which combines six different chips. The expected start date for this capacity is late 2027, so this is a long-term infrastructure commitment rather than a short-term software update.
Anthropic has also reportedly signed other computing agreements, including a six-year, $10 billion arrangement with Volta and a $5 billion deal connected to AMD, while expanding relationships with Amazon, Google, and Broadcom. These figures and arrangements were reported by TechCrunch.
For you, the important point is not the size of these corporate agreements. It is the direction of travel: AI providers are preparing services that can process more information, perform longer sequences of tasks, and support more specialised business applications.
Key insight: You do not need to compete with AI companies on computing power. You need to make your business processes clear enough for useful AI tools to assist safely.
How This Applies to Malaysian SMEs
1. Customer service can become more consistent. Imagine a Malaysian retailer receiving questions through WhatsApp, Instagram, and its website. Customers may ask about delivery areas, product availability, returns, business hours, or payment methods. A properly configured AI assistant could identify the question, retrieve information from an approved knowledge base, and draft a response in English, Bahasa Malaysia, or a mixture of both. A staff member can review sensitive replies before sending them.
This is useful when your team is small and customer enquiries arrive outside office hours. The AI should not invent delivery promises or make policy decisions. Your role is to provide current product information, response rules, escalation instructions, and a clear handover process.
2. Sales follow-up can become a repeatable workflow. Many SMEs lose opportunities because enquiries remain in personal phones, spreadsheets, or chat threads. An automation system can capture a lead, record the product or service requested, assign a follow-up date, and prepare a tailored message. For example, a renovation contractor could separate enquiries for kitchens, offices, and shoplots, then remind the salesperson when a quotation needs checking.
More capable AI systems may eventually review conversations, identify buying signals, and suggest the next action. You should still require approval before an external message is sent. The objective is not to remove judgement; it is to prevent good leads from being forgotten.
3. Administration can take less attention from your core work. A wholesaler may receive purchase orders in PDF files, email attachments, and photographs. An AI-assisted workflow could extract item codes, quantities, delivery dates, and customer details, then place the information into your business system for checking. A clinic could summarise appointment notes for internal use. An agency could turn meeting notes into tasks and deadlines.
These workflows are suitable when the information follows a recognisable pattern. They need stronger controls when documents involve confidential personal information, medical information, employment records, or financial decisions. Limit access, keep an approval step, and record who confirmed the result.
4. Local language and local operating habits matter. Malaysian customers may communicate in Bahasa Malaysia, English, Mandarin, Tamil, or informal combinations of these languages. They may also refer to local areas, public holidays, delivery constraints, and industry terms that a generic system does not understand. Test AI with real examples from your business instead of assuming that a polished demonstration reflects your customers.
A Simple SME Readiness Table
| Business area | Good first use | Human check needed |
|---|---|---|
| Customer enquiries | Draft answers from approved information | Complaints, refunds, unusual requests |
| Sales | Lead classification and follow-up reminders | Quotations, discounts, commitments |
| Documents | Extract fields from forms and orders | Quantities, names, tax details, exceptions |
| Operations | Summarise meetings and create task lists | Deadlines, staffing, safety decisions |
The table is a practical starting framework rather than a fixed rule. The right level of automation depends on the consequences of an error in your business.
Practical Takeaways
- List five repetitive tasks. Include copying information between systems, answering the same questions, preparing summaries, and reminding people to follow up.
- Choose one low-risk workflow first. Drafting internal summaries or sorting enquiries is usually easier to control than automating approvals.
- Prepare a trusted information source. Keep product details, operating hours, service areas, policies, and standard answers in one maintained location.
- Set approval rules. Decide which actions AI may prepare and which actions require a manager or staff member to confirm.
- Test with Malaysian examples. Use actual language styles, local place names, common customer questions, and industry terminology.
- Measure operational results. Track response time, unresolved enquiries, missed follow-ups, correction rates, and staff hours saved. The number of AI features used is not a useful success measure by itself.
- Review access and retention. Do not place confidential customer or employee information into a tool until you understand how the provider handles that information.
The Bigger Picture
The race between Anthropic, OpenAI, Google, Meta, and other major AI companies is driving investment in computing capacity. TechCrunch reported that these companies are pursuing more computing resources as they compete to build and operate advanced AI systems, with Anthropic’s recent partnerships forming part of that wider trend: TechCrunch.
For SMEs, this may lead to more AI tools that can work across documents, messages, business systems, and voice interactions. However, better models will not automatically fix disorganised operations. If your customer data is incomplete, your policies conflict, or nobody owns the workflow, AI may simply produce faster confusion.
The businesses that benefit most will usually have a few habits in place: clear processes, accurate records, defined responsibilities, and sensible review points. These habits help whether you use AI or not.
Start with one task your team performs repeatedly every week. Document how it should be done, identify the information required, and decide where a person must approve the result. Then test an appropriate automation tool on a small scale. The global compute race may be happening far away, but your practical advantage begins with a well-organised workflow in your own business.
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 →