Is AI Really ‘For Everyone’? What Malaysian SMEs Must Know

You run a business. You hear about the latest AI announcement, and something inside you sighs. Another promise from a billionaire in California. Another headline telling you that everything is about to change—again. Meanwhile, you’re still trying to get your invoices sent on time and your team to use the CRM system properly.

When Mark Zuckerberg says AI should be “for everyone,” it sounds good. But as a Malaysian SME owner, you’ve learned to be skeptical. What does “for everyone” actually mean for a company with 20 employees in Johor Bahru or Penang? Does it mean you can finally automate those boring tasks without hiring a data scientist? Or does it mean another tool you’ll never touch?

The truth is somewhere in between. Meta’s recent release of Glimmer, an open-weight AI model anyone can download and run on their own hardware, represents a genuine shift in how AI is distributed. But it also comes with asterisks—and those asterisks matter when you’re trying to run a business, not a tech experiment.

TL;DR

• Meta released Glimmer, an open-weight AI model you can download and run yourself, alongside a letter from Zuckerberg arguing AI shouldn’t be controlled by a few labs.
• Open-weight means more choices for SMEs, especially around data privacy and customisation—but it also means you’re responsible for hosting and maintenance.
• For Malaysian SMEs, the practical value depends less on who owns the AI and more on whether it solves your actual problems.

What This Means

Let’s get past the jargon. Most AI tools you’ve used—ChatGPT, Claude, even some of the automation features in your software—are closed. You access them through an API or a website, and the company behind them keeps the model locked away. You never see the inner workings. You just send data in and get results out.

An open-weight model like Glimmer is different. The “weights” are the mathematical settings that make the AI behave the way it does. Releasing them means anyone can download the model and run it on their own computer or server. According to TechCrunch, Meta released Glimmer alongside a more powerful model, Muse Spark, which stays locked behind the company’s own APIs. So even Meta itself isn’t putting everything in the open.

Think of it like a food delivery service. A closed model is like ordering nasi lemak from a restaurant—convenient, consistent, but you don’t know the recipe and you can’t change the spice level. An open-weight model is like getting the recipe handed to you. You can cook it yourself, adjust the ingredients, and you don’t have to ask the restaurant’s permission every time you want to serve it. But now you’re responsible for buying the groceries, cleaning the kitchen, and making sure you don’t burn the place down.

“Giving everyone access to the model is not the same as giving everyone the ability to use it well.”

How This Applies to Malaysian SMEs

So what does this mean for you? Let’s get concrete. Consider your typical business day. There are dozens of repetitive tasks that eat your team’s time—reading and sorting customer emails, extracting details from purchase orders, drafting responses to common inquiries, summarizing long reports. Right now, you might be paying for a SaaS tool that does some of this using AI. That tool works well, but it has limitations. You might be sending customer data to a foreign server. You might be paying a monthly fee for every seat. And you can’t customise the AI’s behaviour to match your specific industry jargon or customer tone.

An open-weight model changes the calculus. If your company handles sensitive personal data—say, you run a clinic, a legal firm, or a financial advisory practice—you can download Glimmer and run it on your own server. Data never leaves your premises. This aligns better with the Personal Data Protection Act (PDPA) and gives your clients more confidence. For a Malaysian SME that handles HR records or customer information, this could be a real advantage.

There’s also the language angle. Many open-weight models are trained on more languages, but they still struggle with Bahasa Malaysia, especially the casual, mix-and-match style many Malaysians use in customer communications. When you run your own model, you can fine-tune it on your own chat transcripts and documents. You can teach it to understand “ok boss, can arrange delivery next week” or “harga boleh nego tak?” This is something a locked, generic model won’t do well for you. The ability to adapt the AI to your local context is where the real value sits.

But here’s the reality check. Running your own AI model requires technical infrastructure. You need a reasonably powerful server, someone to maintain it, and the skills to get the model up and running in the first place. This is not a weekend project for most SME owners. You’re busy running your business. That’s where the gap between “open” and “accessible” appears. Zuckerberg’s vision of AI “for everyone” is technically true—anyone can download Glimmer. But practically, the person who can actually deploy and maintain it is still a specialist. For a business owner in Klang Valley or Kuching, the choice is often between using a simpler to use closed tool or hiring someone with the expertise to set up an open one.

Practical Takeaways

• Don’t rush to download an open-weight model just because it’s free. Start by identifying one or two repetitive tasks where AI could genuinely help you. Map out the effort required to set it up versus the time saved.
• Look at your data privacy needs. If you handle sensitive customer data, running a model on your own hardware might be worth the complexity. If your data is not sensitive, a closed tool is likely fine and faster to deploy.
• Check what hardware you already have. Some open-weight models can run on a good laptop or a modest desktop server. Others need serious computing power. The TechCrunch coverage notes that Glimmer can be downloaded and run on your own hardware, but “your own hardware” varies widely.
• Start small. Try a hosted version of an open model first (many cloud providers now offer them) before committing to on-premises setup. You’ll learn what the model can and cannot do without the infrastructure headache.
• Talk to someone who’s done it. If you have an IT consultant or a business automation partner, ask them to evaluate whether open-weight AI is right for your operations. A good partner can help you avoid the trap of adopting technology for its own sake.

The Bigger Picture

The argument Zuckerberg is making—that AI should not be controlled by “a handful of labs”—is not new. But the release of a capable open-weight model alongside a more powerful locked one shows the industry is moving in two directions at once. Some models will be open, some will be closed, and some will offer a middle path. This is not a bad thing for SMEs. Choice is good.

Over the long term, this trend points to a world where AI becomes more like electricity or internet access—a utility you plug into, rather than a single vendor’s product. Malaysian SMEs will benefit when they can choose between a simple, hosted solution and a more custom, self-managed one based on their actual needs, not based on a tech giant’s marketing message.

But remember: the technology is only part of the equation. The companies that win with AI are not the ones that use the fanciest models. They are the ones that use any model to solve a specific, painful problem. That could be a closed API that reads your supplier emails in 10 seconds. It could also be an open-weight model that runs offline in your factory in Shah Alam because your internet connection is unstable. The model being open or closed tells you quite little about whether it is useful to you.

So, does Mark Zuckerberg really believe AI is “for everyone”? Maybe. But the more relevant question is: do you believe your business can put AI to work in a way that matters? The answer to that question won’t come from a letter or a press release. It comes from looking at your own operations, your own customers, and your own team—and asking where AI can genuinely make a difference.

Aspect	Open-weight model (e.g., Glimmer)	Closed model (e.g., Muse Spark)
Access	Download and run yourself	API access only
Hosting	Your own server or cloud	Provider’s infrastructure
Data control	Full control, data stays with you	Limited, subject to provider’s policies
Customisation	Possible (fine-tuning, editing)	Restricted
Technical skill needed	Higher (installation, maintenance)	Lower (just call an API)

According to the TechCrunch coverage, the open-weight release came alongside a manifesto from Zuckerberg arguing that AI should be accessible to everyone. Yet the same company chooses to keep its most capable model behind APIs. That contradiction tells you everything you need to know: “for everyone” is a statement of direction, not a description of reality. As a business owner, your job is to focus on what works for your business today—while keeping an eye on what might become useful tomorrow.

The open-weight movement is worth watching. Even if you never run a model yourself, the competition it creates tends to push better tools, lower barriers, and more thoughtful data policies across the industry. That benefits you. But the benefit is real only when you apply it to a concrete business problem. Otherwise, it’s just another headline.

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 →